creiht | hah | 04:10 |
---|---|---|
*** jeromatron has joined #openstack | 04:11 | |
antonym | looks like todd fixed the links | 04:11 |
creiht | cool | 04:12 |
creiht | http://www.readwriteweb.com/cloud/2010/07/openstack-rackspace-and-nasa-n.php | 04:13 |
lbieber | http://trends.ellerdale.com/etc/search?q=OpenStack | 04:13 |
creiht | nice | 04:14 |
*** sirp2 has joined #openstack | 04:16 | |
*** sirp2 is now known as sir_p | 04:17 | |
creiht | mtaylor: we should add a link to the docs at swift.openstack.org on the swift launchpad page | 04:19 |
mtaylor | creiht: yes | 04:20 |
creiht | is there a way I can edit that page? | 04:20 |
mtaylor | creiht: you should be able to | 04:20 |
creiht | and should we really point to the wiki? | 04:20 |
mtaylor | creiht: "Change Details" | 04:20 |
lbieber | mtaylor: and on the nova page as well | 04:20 |
creiht | since a. it is completly open and b. it is a little of a mess at the moment | 04:21 |
* creiht doesn't see a "Change Details" | 04:21 | |
mtaylor | creiht: I recommend putting links and stuff in the description section... | 04:21 |
mtaylor | hrm | 04:21 |
mtaylor | ah. | 04:21 |
mtaylor | hang on | 04:21 |
creiht | I have change details for swift-core | 04:21 |
creiht | cool | 04:21 |
mtaylor | creiht: you should have it for swift now | 04:22 |
mtaylor | lbieber: can you do it for nova? | 04:22 |
creiht | cool thanks | 04:22 |
lbieber | mtaylor: sure | 04:22 |
mtaylor | lbieber: thanks | 04:22 |
*** justinsb has joined #openstack | 04:23 | |
*** bretpiat1 has joined #openstack | 04:26 | |
greenisus | we're now #2 on hacker news: news.ycombinator.com | 04:28 |
creiht | and #4 | 04:29 |
_0x44 | #2 and #3 now | 04:29 |
*** bretpiatt has quit IRC | 04:30 | |
creiht | there are a lot of funky ?'s on the faq page | 04:32 |
creiht | http://openstack.org/projects/openstack-faq/ | 04:32 |
creiht | � | 04:32 |
mtaylor | creiht: that sounds like an encoding problem and/or an encoding declaration problem | 04:33 |
creiht | The "Who has contributed?" question has an incomplete sentence for an answer | 04:33 |
* mtaylor cries a little bit that the hackers are finding english errors | 04:34 | |
creiht | haha | 04:34 |
lbieber | mtaylor: are you requesting the swift mailing list on launchpad? | 04:34 |
mtaylor | lbieber: yes. spm is working on it | 04:35 |
lbieber | mtaylor: Thanks! | 04:35 |
*** toddmorey has joined #openstack | 04:37 | |
*** seats has joined #openstack | 04:37 | |
creiht | toddmorey: howdy! | 04:37 |
lbieber | ozone team needs to subscribe to https://launchpad.net/~nova/+mailing-list-subscribers if they haven't already | 04:37 |
creiht | toddmorey: the faq page has weird ? symbols on it | 04:38 |
lbieber | toddmorey: watch out, you are about to get an onslaught of requests :) | 04:38 |
*** pquerna has joined #openstack | 04:38 | |
toddmorey | howdy! | 04:38 |
creiht | haha | 04:38 |
creiht | seats: !!! | 04:38 |
toddmorey | I'm sure I am. :) Bring 'em on! | 04:38 |
creiht | toddmorey: The "Who has contributed?" question has an incomplete sentence for an answer | 04:38 |
antonym | seats: oh hai | 04:38 |
lbieber | toddmorey: Remove "San Antonio, Texas" from the job description for the Community Manager, it can be any location | 04:38 |
seats | creiht, howdy | 04:38 |
seats | hey ant | 04:38 |
_0x44 | Hey seats | 04:39 |
antonym | how's it going? | 04:39 |
creiht | <- chuck | 04:39 |
creiht | :) | 04:39 |
seats | thier, i got the backward name | 04:39 |
mtaylor | lbieber: well, any location except for washington | 04:39 |
mtaylor | :( | 04:39 |
creiht | hehe | 04:39 |
lbieber | mtaylor: :( :( | 04:39 |
seats | just dropping in to lurk | 04:39 |
creiht | http://thenextweb.com/us/2010/07/19/rackspace-issues-a-challenge-to-the-cloud-industry-goes-open-source-with-openstack/ | 04:39 |
toddmorey | lbieber: done! | 04:41 |
*** greenisus has quit IRC | 04:44 | |
lbieber | toddmorey: Thanks! Did you put in the "How to Contribute" section on openstack.org, I can't seem to find it. | 04:44 |
lbieber | toddmorey: its on the wiki at http://wiki.openstack.org/HowToContribute | 04:45 |
toddmorey | That was certainly a positive writeup on thenextweb | 04:45 |
creiht | yeah... overall pretty decent press | 04:45 |
mtaylor | anybody with blogs ... please add yourself to planet.openstack.org | 04:46 |
toddmorey | lbieber: there's a link to that very page in the second box at the top of the community page | 04:46 |
lbieber | toddmorey: Ahhh yes, very nice | 04:46 |
creiht | haha | 04:48 |
creiht | and it begins | 04:48 |
creiht | asenchi Seriously, openstack.org uses bzr and launchpad? #failhard | 04:48 |
creiht | :) | 04:48 |
antonym | heh | 04:49 |
mtaylor | haters are so boring | 04:51 |
creiht | CloudCEO RT @jclouds: #openstack is the biggest cloud news since #ec2<-Why? Why is this bigger news than any of the other open stacks out there? | 04:53 |
creiht | only other negative tweet so far | 04:53 |
*** patcito has joined #openstack | 04:53 | |
creiht | at least that I have seent | 04:53 |
creiht | erm seen | 04:53 |
creiht | jesse is steve wozniak's son? | 04:55 |
antonym | hah, i saw that too | 04:55 |
*** kuadrosx has joined #openstack | 04:55 | |
creiht | that's awesome :) | 04:55 |
creiht | mtaylor: bumping to 1.0.1? | 04:56 |
mtaylor | creiht: yeah - I finally got around to cutting the 1.0 release I was supposed to do on friday :) | 04:57 |
creiht | ahh | 04:57 |
creiht | cool | 04:57 |
*** anotherjesse has joined #openstack | 04:57 | |
patcito | hi | 04:57 |
anotherjesse | congrats guys | 04:57 |
mtaylor | creiht: so - http://launchpad.net/swift/1.0 | 04:58 |
creiht | anotherjesse: are you really the woz's son? | 04:58 |
anotherjesse | first I ever heard of that | 04:58 |
creiht | heh | 04:58 |
patcito | what do you guys think of http://www.eucalyptus.com/ ? it is supposed to be another "open source cloud stack" | 04:58 |
anotherjesse | is there a jesse at rackspace? | 04:58 |
creiht | or is that a different jesse at nasa? | 04:58 |
creiht | I don't think so | 04:58 |
*** silassewell has joined #openstack | 04:59 | |
mtaylor | patcito: I don't want to get all flame-orientied - but eucalyptus is really more open core than open source | 05:00 |
patcito | mtaylor, ok, thanks for the info | 05:00 |
mtaylor | ok. I'm supposed to be on a beach vacation... :) I'm going to go to bed... I'll obviously be around and online though | 05:02 |
creiht | mtaylor: do we have to bump version in the debian stuff? | 05:02 |
*** mtaylor is now known as mtaylor|beach | 05:02 | |
creiht | hah | 05:02 |
mtaylor|beach | creiht: well, we should probably properly cut debs based off of the now-released 1.0.0 tarball | 05:02 |
creiht | k | 05:02 |
* mtaylor|beach adds to todo | 05:03 | |
creiht | mtaylor|beach: no worries.. enjoy the vacation | 05:03 |
mtaylor|beach | creiht: thanks! | 05:03 |
creiht | http://bartongeorge.net/2010/07/18/talkin-to-the-project-lead-of-openstack-object-storage/ | 05:09 |
*** howardroark1 has joined #openstack | 05:10 | |
*** dendrobates is now known as dendro-afk | 05:11 | |
*** howardroark1 has quit IRC | 05:12 | |
*** jakedahn has joined #openstack | 05:13 | |
*** jdmaturen has joined #openstack | 05:14 | |
jdmaturen | "written in Python, using the Tornado and Twisted frameworks, and relies on the standard AMQP messaging protocol as well as the Redis distributed KVS." <= Redis isn't distributed, did you guys do additional work on making redis clusterable? | 05:15 |
jdmaturen | re: OpenStack Compute | 05:15 |
*** dendro-afk is now known as dendrobates | 05:15 | |
dendrobates | jdmaturen: no we did not. | 05:16 |
*** pharkmillups has joined #openstack | 05:16 | |
jeromatron | any chance of moving to something like cassandra? :) | 05:16 |
dendrobates | jdmaturen: YES! | 05:16 |
jdmaturen | dendrobates: so is "distributed" a misnomer? | 05:17 |
soren | Very few things are set in stone at this point. We're not married to Redis at all. | 05:17 |
jeromatron | dendrobates - what did you say YES to? | 05:17 |
bretpiat1 | jdmaturen, where did you pull that from? I'll get the marketing folks to fix it ;) | 05:17 |
dendrobates | there are some things in the current version that will be fixed | 05:17 |
*** bretpiat1 is now known as bretpiatt | 05:18 | |
jdmaturen | bretpiatt: http://openstack.org/projects/compute/ | 05:18 |
jdmaturen | got my hopes up ;) | 05:18 |
jeromatron | soren - if you are interested in using cassandra - the #cassandra channel has a ton of people who would be willing to help I think :) | 05:18 |
dendrobates | we plan on adding cassandra support as soon as we can. | 05:19 |
bretpiatt | jdmaturen, Does Redis have specific features you want that a Cassandra implementation would not? | 05:19 |
soren | jeromatron: Let's just say that we have pretty good contacts in the Cassandra community :) | 05:19 |
dendrobates | not sure about the first release though | 05:19 |
creiht | soren: haha! | 05:19 |
jeromatron | (disclaimer - I work at rackspace on cassandra) | 05:19 |
creiht | jeromatron: I sit on the other side of the partition from you :) | 05:20 |
soren | jeromatron: Oh :) | 05:20 |
creiht | jeromatron: if you are who I think you are :) | 05:20 |
jeromatron | creiht - do you work in Austin or San Antonio? | 05:20 |
creiht | San Antonio | 05:21 |
bretpiatt | jeromatron, take a look at http://nova.openstack.org and look at where Redis plugs in, you could help with the implementation :) | 05:21 |
creiht | but I just realized that there are no more cassandra guys on the other side of the partition in SA, except for Eric | 05:21 |
patcito | dendrobates, so OpenStack Object Storage is like a CDN and OpenStack Compute a DB to do map/reduce stuff? not sure | 05:21 |
jeromatron | okay - eric and gary work on cassandra in san antonio. we have 4 guys up in austin working on cassandra now. I'm up in Austin though I occasionally go to SA. | 05:22 |
creiht | patcito: Object Storage is more like Amazon's S3 | 05:22 |
jeromatron | bretpiatt: cool - tx! | 05:22 |
creiht | For Rackspace's cloud files, we do integrate with Limelight to provide CDN functionality | 05:22 |
creiht | we are currently working with Limelight to open source the connecter to them as well | 05:23 |
patcito | creiht, ok, and what about OpenStack Compute? | 05:23 |
* creiht passes the buck to the compute guys :) | 05:23 | |
jdmaturen | bretpiatt: in general? failure scenarios for cassandra are too painful. I need redis for in memory data structures w/ atomic operations | 05:23 |
jdmaturen | i don't use redis as kv personally | 05:24 |
jeromatron | jdmaturen - failure scenarios? atomic operations like multi-row transactions? | 05:25 |
jdmaturen | lists, sets, sorted sets, etc | 05:25 |
antonym | patcito: OpenStack Compute is the software to provision large scale deployments of compute instances, it currently runs KVM but more support is being added | 05:26 |
*** hornbeck has joined #openstack | 05:26 | |
jdmaturen | jeromatron: not so much transactions | 05:26 |
patcito | antonym, are there any equivalent (proprietary or not)? so I can get an idea | 05:26 |
jeromatron | jdmaturen: I'm not sure why cassandra couldn't handle lists, sets, and sorted sets - its architecture borrows from bigtables columnfamily for example. you may have already looked into the data model though. | 05:28 |
soren | patcito: The software that runs Rackspace Cloud Servers or Amazon EC2. That sort of thing. | 05:28 |
jeromatron | (well column family/super column familiy) | 05:28 |
jdmaturen | jeromatron: i have looked into it, yea, perhaps its just the perceived difficulty compared to looking at the redis command list :) | 05:30 |
justinsb | If you're going to look at Cassandra, you could consider working from my branch (or at least looking at it): https://code.launchpad.net/~justin-fathomdb/nova/abstract-data-stores | 05:30 |
justinsb | My focus is on getting traditional SQL support (for e.g. moderate sized private clouds), but I'm really abstracting out the data storage layer | 05:31 |
patcito | soren, ok thanks, so to sum it up I could build images of eg. ubuntu10.04 and Compute would manage those images while OpenStorage would provide an easy interface for CDN? | 05:31 |
*** kuadrosx has quit IRC | 05:31 | |
anotherjesse | good luck guys! I have to sleep to be in DC tomorrow :( 5am flight | 05:31 |
creiht | mtaylor|beach: doh!!! /doc directory isn't in the tarball for swift | 05:32 |
creiht | anotherjesse: safe travels | 05:32 |
*** jpp has joined #openstack | 05:32 | |
soren | patcito: That will eventually be true, yes. | 05:32 |
*** jpp is now known as Guest60859 | 05:32 | |
jeromatron | justinsb - cool. I'll check that out. there was a paper on PIQL that talked about a generalized query mechanism for stuff like - not sure if that would be interesting - http://www.eecs.berkeley.edu/Pubs/TechRpts/2010/EECS-2010-8.html | 05:33 |
patcito | soren, cool, maybe you should explain it like that on the web page because at first look it's kind of confusing if I may suggest | 05:33 |
*** ChanServ changes topic to "Welcome to OpenStack! Find out more and get involved here: http://www.openstack.org" | 05:33 | |
justinsb | jeromatron: Interesting stuff. I think there's a solid need for traditional SQL databases for the private cloud case though. Some enterprises will insist on Oracle or DB2 or whatever... | 05:37 |
jeromatron | justinsb: I'm sure... there are just times when you need scalability and that gets expensive with proprietary dbs... | 05:38 |
jeromatron | justinsb - is your launchpad fork something that might make it into mainline? would it be better to look at how redis is currently plugged into nova or look more at your stuff to hack at making cassandra plug in? | 05:39 |
justinsb | jeromatron: Well, another advantage is that we can actually benchmark them head-to-head :-) Digging in to the data-store code, it seems that a K-V DB might not be quite so advantageous, because many operations that could be a single operation in SQL end up being multiple operations in Redis. But I want to avoid religious wars here! | 05:40 |
justinsb | I would certainly hope that my fork will make it into mainline. It's not really a fork so much as a branch - the standard way of working on features in launchpad. | 05:41 |
jeromatron | justinsb - sure - makes sense. just wanted to see the best way to plug in. cool - just wondered since it said experimental in the description. | 05:41 |
Guest60859 | From scoble's post: "Every startup, nearly, uses the LAMP stack and those who don’t go that way tell each other at private industry events like the one I’m at right now in Jackson Hole, Wyoming, that jumping off the open source stack brings negative results for companies." I guess that's not vendor lockin, but now it's LAMP lockin? Kidding, this looks fantastic. | 05:42 |
justinsb | jeromatron: I would look at how nova does it, but then I'd start from my fork. It's super-buggy right now, but most of the work was actually in abstracting out redis, not in the SQL stuff. I haven't really tested the SQL stuff much yet. | 05:42 |
*** brainsik has joined #openstack | 05:43 | |
*** lbieber has quit IRC | 05:44 | |
*** brainsik has quit IRC | 05:45 | |
*** hornbeck has quit IRC | 05:47 | |
*** anotherjesse has quit IRC | 05:48 | |
*** pvo has joined #openstack | 05:49 | |
*** driftx has joined #openstack | 05:50 | |
*** anotherjesse has joined #openstack | 05:50 | |
*** anotherjesse has joined #openstack | 05:51 | |
*** pvo has quit IRC | 05:58 | |
*** silassewell has quit IRC | 05:58 | |
*** pvo has joined #openstack | 06:05 | |
*** anotherjesse has joined #openstack | 06:09 | |
*** greenisus has joined #openstack | 06:09 | |
*** jhalogen has joined #openstack | 06:10 | |
*** anotherjesse has quit IRC | 06:11 | |
*** holoway has joined #openstack | 06:16 | |
holoway | congrats, guys | 06:16 |
pvo | Thanks! Lots of work ahead.... | 06:17 |
*** joshuamckenty has joined #openstack | 06:17 | |
creiht | hehe... | 06:17 |
* creiht is already getting bug reports :) | 06:17 | |
bretpiatt | creiht, can you give me a link to the bug report? | 06:18 |
*** joshuamckenty has quit IRC | 06:18 | |
creiht | bretpiatt: just little things so far | 06:18 |
creiht | https://bugs.launchpad.net/swift/+bug/607074 | 06:18 |
*** cglee has joined #openstack | 06:19 | |
creiht | and | 06:20 |
creiht | https://bugs.launchpad.net/swift/+bug/607068 | 06:20 |
* creiht points his finger at mtaylor|beach | 06:20 | |
creiht | :) | 06:21 |
*** Predominant has joined #openstack | 06:22 | |
Predominant | Hey stackers | 06:22 |
*** pvo has quit IRC | 06:23 | |
Guest60859 | Heh | 06:23 |
Predominant | :) | 06:23 |
Predominant | What is everyone up to? | 06:23 |
*** pvo has joined #openstack | 06:24 | |
pvo | Hey there | 06:24 |
pvo | Don't really like stackers... What | 06:24 |
dendrobates | Predominant: just experiencing the excitement of the launch | 06:24 |
pvo | Other name can we use? | 06:24 |
Predominant | stackites? | 06:25 |
antonym | better than slackers :) | 06:25 |
Predominant | Stackies? | 06:25 |
creiht | racker stacker | 06:25 |
creiht | :) | 06:25 |
Predominant | stax | 06:25 |
redbo | why? | 06:26 |
pvo | Why don't I like it? | 06:26 |
creiht | stackets? | 06:26 |
pvo | Do we have to be something? | 06:27 |
redbo | why is everyone trying to make me cry? | 06:27 |
creiht | awww | 06:27 |
Predominant | Whats the license on the projects | 06:27 |
pvo | Apache2 | 06:27 |
*** joshuamckenty has joined #openstack | 06:28 | |
pvo | Hey hey it's joshuamckenty | 06:28 |
joshuamckenty | evening, all | 06:28 |
joshuamckenty | just watching the twitter traffic | 06:28 |
*** jhalogen has quit IRC | 06:28 | |
Predominant | Does Apache2 allow me to host a copy of this elsewhere? | 06:29 |
joshuamckenty | Oh, I have a patch for mac dev environments again, btw | 06:29 |
joshuamckenty | Predominant: yup | 06:29 |
joshuamckenty | Apache2 allows you to do basically anything you want | 06:29 |
pvo | joshuamckenty: Oh awesome | 06:29 |
*** spackest has joined #openstack | 06:30 | |
*** joshuamckenty has quit IRC | 06:30 | |
spackest | looking to me like openstack is going to be the internal cloud building winner? that sound about right? | 06:30 |
Predominant | Whas the bzr link for the repo | 06:30 |
*** joshuamckenty has joined #openstack | 06:30 | |
soren | Predominant: lp:nova or lp:swift depending on which you want. | 06:31 |
Predominant | Cheers | 06:31 |
spackest | we've been playing with eucalyptus with about 80 cores and ten terabytes, but looking like openstack is way ahead | 06:31 |
*** joshuamckenty has quit IRC | 06:31 | |
spackest | we just might be a month or two early for open stack, does that sound right? | 06:31 |
Predominant | Who is working on the web control panel? | 06:32 |
*** joshuamckenty has joined #openstack | 06:32 | |
pvo | Ther should be a ref panel coming soon | 06:32 |
spackest | so is openstack going to speak the amazon or rackspace api? think I saw something about ec2 api, is that right? | 06:33 |
greenisus | @Predominant, I'm building the web control panel | 06:33 |
larissa | greenisus: Error: "Predominant," is not a valid command. | 06:33 |
Predominant | Cool | 06:33 |
creiht | spackest: storage is already what is what is in production, so if you want to start looking at that, by all means go ahead :) | 06:33 |
Predominant | greenisus: Have you begun work on that? | 06:33 |
creiht | compute still has some work to do | 06:34 |
spackest | creiht: right now we run gpfs to our walrus and storage controller with five two TB luns, that going to be easy to switch over? | 06:34 |
creiht | greenisus: this is irc, not twitter :) | 06:34 |
Predominant | creiht: There is only twitter. | 06:34 |
spackest | we're kind of on our last legs with eucalyptus | 06:34 |
joshuamckenty | Predominant: both apis for now, but ec2 is essentially just legacy | 06:34 |
* creiht googles gpds | 06:34 | |
creiht | erm gpfs | 06:35 |
greenisus | sorry about that | 06:35 |
creiht | hehe | 06:35 |
joshuamckenty | sorry, spackest: ==^ | 06:35 |
pvo | The rackspace API will be the default long term but can support any APi if you want to write support | 06:35 |
antonym | getting the bot all confused heh | 06:35 |
spackest | you ever play with deltacloud? | 06:35 |
greenisus | yeah, i've begun work on the control panel, and i'll get it in launchpad soon | 06:35 |
spackest | anyone? it speaks rackspace and amazon | 06:35 |
greenisus | at the moment it offers compute support. working on storage next | 06:35 |
greenisus | compute speaks amazon at the moment, but storage speaks the rackspace API only | 06:36 |
pvo | It speaks amazon today but will speak rackspace by default soon | 06:36 |
greenisus | right | 06:36 |
*** joshuamckenty has quit IRC | 06:36 | |
creiht | spackest: from a quick look at ibm's site on gpfs, I'm not sure | 06:36 |
spackest | our infrastructure team has been beating up eucalyptus pretty good, would be nice to have a real system that it sounds like powers rackspace, the second largest public cloud? | 06:36 |
spackest | creiht: we'll have to play around | 06:36 |
creiht | I would have to research it a bit further | 06:36 |
creiht | how does it interface with walrus? | 06:36 |
spackest | we end up just mounting everything to a single directory | 06:37 |
spackest | for walrus, and another for the storage controller | 06:37 |
* creiht isn't very familiar with walrus either :/ | 06:37 | |
*** joshuamckenty has joined #openstack | 06:37 | |
creiht | so swift is a object storage backend that provides a RESTful interface | 06:37 |
spackest | basically you get a single directory that ends up be your walrus (s3) | 06:37 |
pvo | joshuamckenty: You on mobile? | 06:37 |
*** joshuamckenty has quit IRC | 06:37 | |
pvo | Heh | 06:38 |
spackest | guess I should go look at the storage api | 06:38 |
creiht | the api has some similarities to s3, but doesn't match one to one | 06:38 |
spackest | anything for ebs? | 06:38 |
pvo | Not yet | 06:38 |
creiht | It isn't a filesystem or block storage | 06:38 |
spackest | we're much more interested in ebs | 06:38 |
pvo | Swift is object storage | 06:38 |
spackest | or at least something mountable | 06:39 |
*** joshuamckenty has joined #openstack | 06:39 | |
creiht | we are interested in ebs as well, just don't have a solution *yet* | 06:39 |
creiht | :) | 06:39 |
spackest | but I guess there are plans for ebs / filesystem storage? | 06:39 |
spackest | lots of folks would be interested :) | 06:40 |
*** joshuamckenty has quit IRC | 06:40 | |
antonym | spackest: you might want to check this out, http://nova.openstack.org/volume.html | 06:40 |
pvo | Yes were working on it | 06:40 |
*** joshuamckenty has joined #openstack | 06:40 | |
spackest | you guys all rackspace folks? | 06:40 |
creiht | some are | 06:41 |
pvo | I am | 06:41 |
greenisus | i am, but we're a diverse group | 06:41 |
antonym | some rs, some nasa, some others as well | 06:41 |
spackest | and rackspace's motivation? | 06:41 |
creiht | http://www.rackspacecloud.com/blog/2010/07/19/open-stack/ | 06:42 |
spackest | we're currently using ata over ethernet for our eucalyptus cloud, so that sounds pretty reasonable | 06:42 |
pvo | meaning? Why open up? No one was real doing this. Fully open design and collab | 06:42 |
creiht | might go a ways to explaining that | 06:42 |
spackest | yes, why help another company build a potential rs competitor | 06:43 |
spackest | reading the post . . . | 06:43 |
creiht | spackest: We differentiate ourselves by service, not by technology | 06:44 |
creiht | just like anyone can build hosting witha a LAMP stack, we want everyone to be able to build a cloud with openstack | 06:44 |
creiht | we want open standards, and portability | 06:44 |
spackest | fanatical support! | 06:45 |
creiht | :) | 06:45 |
spackest | sounds great | 06:45 |
spackest | not sure I will be able to sleep :) | 06:45 |
creiht | spackest: http://bartongeorge.net/2010/07/18/introducing-openstack-an-open-source-cloud-platform/ | 06:45 |
spackest | we've been in talks with vmware and they are less than compelling to me | 06:45 |
creiht | is an interview with Lew Moorman, which explains a bit about it as well | 06:45 |
creiht | hah | 06:45 |
spackest | we have a decent sized internal vmware farm, but it still isn't develop self-serve | 06:46 |
antonym | we'd like to see vmware contribute to the project | 06:46 |
spackest | developer | 06:46 |
pvo | Developer self serve is awesome | 06:47 |
spackest | does amazon consider rs a threat? like rs is number two, but is it close? | 06:47 |
patcito | if I want to build my own rackspace competitor, will openstack allow me to track what each of my customer are using (cpu/ram etc) so I can charge my customer easily by writing some glue code or is that not open source? | 06:47 |
creiht | spackest: that would be a good question to ask amazon :) | 06:48 |
spackest | ha | 06:48 |
spackest | you guys seen this? http://code.google.com/p/lasic/wiki/UsersGuide | 06:48 |
spackest | I manage a cloud services team and we're trying to help developers provision their own stuff using lasic | 06:49 |
*** jdmaturen has quit IRC | 06:49 | |
spackest | could pretty well change our world, just need a beter cloud to offer folks | 06:49 |
creiht | spackest: I think that is what we are hoping for :) | 06:49 |
pvo | patcito: It can | 06:49 |
spackest | wouldn't be too hard to add an openstack provider to lasic | 06:50 |
patcito | cool | 06:50 |
spackest | been pondering adding a deltacloud provider | 06:50 |
patcito | gotta go | 06:50 |
patcito | cheers | 06:50 |
pvo | Laters | 06:50 |
spackest | what do folks use to describe the instances they are launching? like it has to be scriptable, right? | 06:51 |
spackest | just python or the like? | 06:51 |
spackest | or ruby for deltacloud | 06:51 |
spackest | folks have liked lasic | 06:51 |
spackest | we have scripted instance launch for several teams generally in less than an hour | 06:52 |
spackest | and yes, it's nice an open | 06:52 |
*** bretpiatt has quit IRC | 06:52 | |
spackest | seems like openstack has need for a nice, simple way for describing / launching / managing instances | 06:53 |
greenisus | what do you mean? | 06:53 |
pvo | Outside of the API? | 06:54 |
spackest | yes | 06:54 |
spackest | writing something simpler than code | 06:54 |
greenisus | we'll have a few things up soon: | 06:54 |
spackest | like lasic :) | 06:54 |
greenisus | web control panel, iphone app, android app, ipad app | 06:54 |
spackest | so our teams write lasic, which brings up their instance(s) to a viable state | 06:55 |
pvo | And chrome and ff apps | 06:55 |
creiht | greenisus: he's talking about something a bit different than that | 06:55 |
greenisus | oh yes, forgot about the chrome/ff app | 06:55 |
pvo | spackest: Go for it | 06:55 |
spackest | like bringing up a load balancer, tomcat(s) and database(s) all at once | 06:55 |
pvo | spackest: At would be awesome | 06:55 |
pvo | At/that | 06:56 |
*** patcito has quit IRC | 06:56 | |
antonym | pvo: that auto correct is killing you :) | 06:56 |
spackest | we're using it to manage 30-40 instances that make up a viable reference | 06:56 |
driftx | last I checked eucalyptus/walrus was pretty weak, SPOF everywhere | 06:56 |
pvo | antonym: It sure makes it interesting :) | 06:56 |
spackest | so much easier than manual or a control panel | 06:56 |
spackest | the 30-40 instances are at amazon | 06:57 |
antonym | yeah, i disabled it on mine | 06:57 |
spackest | eucalyptus says they are doing a ha version, but it is a quarter or two away | 06:57 |
driftx | mongodb says similar crap. | 06:57 |
spackest | and who knows if it will really be ha | 06:57 |
spackest | you guys must have seen the need for scriptable launch, right? | 06:58 |
pvo | spackest: Sure | 06:58 |
spackest | when one of my teams started playing at amazon we spent way too much time tinkering with instances | 06:58 |
spackest | I wrote some code using typica http://code.google.com/p/typica/ which lasic uses | 06:59 |
spackest | honestly, I think openstack could just crush | 07:01 |
spackest | betting eucalyptus, opennebula, nimbula and the like aren't incredibly excited | 07:01 |
spackest | or maybe they could just offer support for openstack of building their own technologies | 07:01 |
pvo | Heh. I bet they aren't happy | 07:01 |
Guest60859 | This could be seen as a good move against all other cloud providers, since Rackspace isn't just limited to the cloud. It says, "We'll commoditize the hell out of the cloud, and we'll win the game by driving down everyone else's profit margin in that arena while consolidating our hold on custom/managed hosting." | 07:02 |
*** greenisus has quit IRC | 07:03 | |
spackest | thanks for the links | 07:03 |
pvo | Later all. Long day.... | 07:04 |
* creiht needs some sleep as well | 07:04 | |
*** pvo has quit IRC | 07:04 | |
spackest | congrats, I look forward to following / implementing your progress :) | 07:05 |
*** sir_p has quit IRC | 07:08 | |
*** Guest60859 has quit IRC | 07:10 | |
*** greenisus has joined #openstack | 07:10 | |
*** pvo has joined #openstack | 07:14 | |
*** bpiatt has joined #openstack | 07:17 | |
joshuamckenty | gnight, all | 07:18 |
joshuamckenty | openstack just hit cnet, btw: http://news.cnet.com/8301-13846_3-20010870-62.html | 07:18 |
*** jakedahn has quit IRC | 07:22 | |
joshuamckenty | hmmm | 07:24 |
joshuamckenty | http://www.computerworlduk.com/community/blogs/index.cfm?entryid=3078&blogid=41 | 07:24 |
*** jakedahn has joined #openstack | 07:25 | |
*** novas0x2a has joined #openstack | 07:25 | |
*** pvo has quit IRC | 07:25 | |
*** pvo has joined #openstack | 07:26 | |
*** pvo has quit IRC | 07:26 | |
*** pvo has joined #openstack | 07:26 | |
*** chuhnk has joined #openstack | 07:29 | |
chuhnk | I salute your openness | 07:29 |
*** pvo has quit IRC | 07:30 | |
*** spike has joined #openstack | 07:32 | |
*** joey_ has joined #openstack | 07:33 | |
*** spackest has quit IRC | 07:33 | |
*** brainproxy has joined #openstack | 07:36 | |
brainproxy | hey, congrats on the launch-announcement guys | 07:36 |
brainproxy | as a former racker I'm really proud to see this happen :) | 07:37 |
chuhnk | oh btw http://docs.novacc.org is down | 07:37 |
chuhnk | thank god for google cached pages | 07:38 |
*** jakedahn has left #openstack | 07:41 | |
joshuamckenty | chuhnk: thanks for the heads up | 07:46 |
joshuamckenty | I think the hudson bot is going to rebuild that somewhere | 07:46 |
joshuamckenty | to go along with http://hudson.openstack.org/job/nova-pylint/ | 07:47 |
chuhnk | cool, I was just trying to checkout docs on the architecture. Does it have to be used with EC2 or will this allow cloud infrastructure to be built on ones own machines? | 07:47 |
*** greenisus has quit IRC | 07:50 | |
*** calavera has joined #openstack | 08:16 | |
joshuamckenty | any hardware, anywhere | 08:17 |
joshuamckenty | docs are up at http://nova.openstack.org/ | 08:17 |
joshuamckenty | but they haven't been updated since before the openstack work started | 08:17 |
joshuamckenty | so there aren't any docs for the rackspace APIs yet | 08:18 |
*** calavera has quit IRC | 08:23 | |
*** joshuamckenty has quit IRC | 08:23 | |
chuhnk | ah ok | 08:24 |
chuhnk | I look forward to seeing some stuff and experimenting with it | 08:24 |
*** joshuamckenty has joined #openstack | 08:24 | |
spike | hi, great to see this project happening | 08:25 |
*** joshuamckenty has quit IRC | 08:25 | |
spike | one thing that is not clear to me is if it integrates with ec2 or other cloud services like eucalyptus does | 08:27 |
chuhnk | (09:17:41) joshuamckenty: any hardware, anywhere | 08:27 |
*** joshuamckenty has joined #openstack | 08:27 | |
chuhnk | I just asked the same question | 08:27 |
spike | chuhnk: nope, I'm asking a different question | 08:27 |
spike | I'm asking about integration, not requirements | 08:27 |
spike | I'd like to be able to bridge my private cloud with a public one like eucalytus does with rightscale | 08:28 |
dendrobates | rightscale does that, not eucalyptus | 08:28 |
dendrobates | and rightscale is working with openstack | 08:29 |
dendrobates | so hopefully it will be work soon | 08:29 |
spike | dendrobates: well, ecucalyptus does have code to integrate with righscale, no? | 08:29 |
spike | ok, cool, that's great news | 08:29 |
spike | that said, I'd like to remove the middleman, I have no particular interest or necessity to work with rightscale so I'd like my cloud to plug straight into ec2 | 08:30 |
spike | I guess it's "just" a matter of adding plugins/modules to the controller so that whenever you issue some "start" command and target "othercloud" it can tell what other cloud is and translate that "start" command to the appropriate API call | 08:32 |
*** pdelgallego has joined #openstack | 08:32 | |
joshuamckenty | spike: we've been looking at that as a concept called "cloud brokering / brokerage" | 08:34 |
joshuamckenty | but it's still mostly theoretical, since we want to support seamless migration / splits, etc. | 08:34 |
joshuamckenty | and there's still not a good conceptual model for a globall-portable virtual network | 08:35 |
*** calavera has joined #openstack | 08:37 | |
spike | joshuamckenty: is there any document on that design? i'd be very happy to try to help with that | 08:37 |
joshuamckenty | it's still internal to NASA right now, I've got a meeting this week to try and break that loose | 08:37 |
spike | awesome, thanks | 08:38 |
joshuamckenty | np | 08:38 |
* spike is very excited about openstack | 08:38 | |
*** tupshin has joined #openstack | 08:40 | |
*** joshuamckenty has quit IRC | 08:40 | |
*** miclorb_ has joined #openstack | 08:40 | |
*** joshuamckenty has joined #openstack | 08:41 | |
*** joshuamckenty has quit IRC | 08:43 | |
*** calavera has quit IRC | 08:47 | |
*** calavera has joined #openstack | 08:48 | |
*** Garo_ has joined #openstack | 09:03 | |
*** cglee has quit IRC | 09:04 | |
*** truff has joined #openstack | 09:07 | |
Garo_ | hi, I'm wandering around Switft code. Are there somewhere examples how to add new objects to the cloud? if there are then I haven't found those examples nor docs yet :< | 09:19 |
*** joey_ has quit IRC | 09:26 | |
*** joey_ has joined #openstack | 09:26 | |
*** jtdowney has joined #openstack | 09:29 | |
*** miclorb_ has quit IRC | 09:31 | |
*** joey_ has quit IRC | 09:32 | |
*** joey_ has joined #openstack | 09:32 | |
*** joey_ has left #openstack | 09:33 | |
*** miclorb_ has joined #openstack | 09:36 | |
*** bpiatt has quit IRC | 09:37 | |
*** bpiatt has joined #openstack | 09:37 | |
*** barismetin has joined #openstack | 09:44 | |
*** matclayton has joined #openstack | 09:49 | |
*** andyss_ has joined #openstack | 09:49 | |
*** calavera has quit IRC | 10:00 | |
*** chilts has joined #openstack | 10:10 | |
chilts | sweet! | 10:10 |
*** miclorb_ has quit IRC | 10:11 | |
*** calavera has joined #openstack | 10:13 | |
*** oz` has joined #openstack | 10:18 | |
*** calavera has quit IRC | 10:21 | |
*** brainsik has joined #openstack | 10:43 | |
*** brainsik has joined #openstack | 10:44 | |
*** bpiatt has quit IRC | 11:11 | |
*** bpiatt has joined #openstack | 11:11 | |
redbo | Garo_: there's sort of a reference client in the swift source code between bin/st (cli) and swift/common/client.py (library) | 11:11 |
redbo | Garo_: there's also the rackspace cloud api developer guide, most of which also applies to swift from http://bit.ly/bc7zK8 | 11:15 |
redbo | We probably need to duplicate that in the swift docs. | 11:16 |
redbo | creiht: ^^^ :) | 11:20 |
*** jtdowney has quit IRC | 11:49 | |
*** arcane has joined #openstack | 11:55 | |
arcane | ooh | 11:55 |
*** larsemil has joined #openstack | 11:58 | |
*** Lartsa has joined #openstack | 12:13 | |
*** cglee has joined #openstack | 12:19 | |
*** calavera has joined #openstack | 12:24 | |
*** ttx has joined #openstack | 12:38 | |
ttx | mtaylor|beach, soren, dendrobates: congrats on the announcement ! | 12:38 |
*** brainsik has quit IRC | 12:42 | |
*** brainsik_ has joined #openstack | 12:42 | |
*** brainsik_ has quit IRC | 12:44 | |
*** gdusbabek has joined #openstack | 12:56 | |
*** carsonm has joined #openstack | 12:59 | |
*** W3b3r_ has joined #openstack | 12:59 | |
W3b3r_ | Hello guys | 12:59 |
W3b3r_ | Openstack is a all-software solution? | 12:59 |
W3b3r_ | Or is a frontend to a another software, like Eucalyptus? | 12:59 |
*** ivan has joined #openstack | 13:01 | |
*** glangst1 has joined #openstack | 13:04 | |
*** rocambol2 has joined #openstack | 13:07 | |
*** HansF has joined #openstack | 13:10 | |
*** lbieber has joined #openstack | 13:12 | |
*** andrewvhs has joined #openstack | 13:14 | |
mtaylor|beach | W3b3r_: it's a complete solution in itself. | 13:15 |
W3b3r_ | mtaylor|beach, hum, thank you | 13:16 |
ttx | mtaylor|beach: I think the question is, does it build on top of other software, like KVM | 13:16 |
mtaylor|beach | ttx: ah - well... | 13:16 |
mtaylor|beach | W3b3r_: if that is your question, then yes, it does build on top of other software like kvm | 13:16 |
ttx | I think frontend is a bit reductive :) | 13:16 |
W3b3r_ | hehe | 13:17 |
mtaylor|beach | W3b3r_: but it does not build on top of other cloud implementations such as eucalyptus | 13:17 |
mtaylor|beach | oh, and thanks ttx | 13:17 |
ttx | mtaylor|beach: you're welcome ;) | 13:17 |
ttx | mtaylor|beach: you're actually at the beach, or at the beach of Portland ? | 13:18 |
mtaylor|beach | ttx: the beach | 13:19 |
mtaylor|beach | ttx: cozumel | 13:19 |
ttx | mtaylor|beach: looks nice, enjoy it :) | 13:20 |
*** malthe has joined #openstack | 13:20 | |
mtaylor|beach | ttx: mmm. already am :) | 13:20 |
*** lbieber has quit IRC | 13:24 | |
*** astrolin has joined #openstack | 13:29 | |
*** issackelly has joined #openstack | 13:31 | |
*** avi has joined #openstack | 13:31 | |
issackelly | What would a minimal hardware setup be to tinker with the openstack storage? | 13:32 |
*** bpiatt has quit IRC | 13:32 | |
creiht | redbo: re: developer docs, check! | 13:33 |
* creiht goes to add a bug | 13:33 | |
mtaylor|beach | creiht: saw the setup.py bug you filed... | 13:33 |
mtaylor|beach | creiht: I'm thinking perhaps we fix these little small bugs and cut a 1.0.1? | 13:33 |
*** stefano_s has joined #openstack | 13:34 | |
creiht | mtaylor|beach: sounds good... did you see that the tarfile was missing files? | 13:34 |
mtaylor|beach | yup | 13:35 |
*** greg-g has joined #openstack | 13:35 | |
kirkland | dendrobates: soren: mtaylor|beach: hey guys, is there a complete list of all of the Etherpad documents from Wednesday? | 13:40 |
creiht | mtaylor|beach: did you get a chance to upload a new one? | 13:40 |
kirkland | dendrobates: soren: mtaylor|beach: I have all of them from Tuesday, but missed a couple of URLs from Wednesday | 13:40 |
mtaylor|beach | creiht: not yet | 13:41 |
*** jonesy has joined #openstack | 13:41 | |
creiht | k | 13:41 |
jonesy | :-D | 13:41 |
kirkland | dendrobates: soren: mtaylor|beach: nevermind, found links on the wiki | 13:43 |
*** SEJeff has joined #openstack | 13:47 | |
notmyname | issackelly: I'm running swift on a VM for dev purposes. so minimal storage requirements are really light. you could run it on one server with several (~4) drives or a small collection of servers (again, ~4) | 13:51 |
*** rocambol2 has quit IRC | 13:51 | |
issackelly | notmyname: cool, I'll probably try it out on a VM and see how it goes. I really like the idea of a local 4-16TB datastore | 13:51 |
notmyname | issackelly: check out the SAIO page on swift.openstack.org for instructions on how to set up a VM | 13:52 |
*** HansF has quit IRC | 13:58 | |
*** glangst1 has quit IRC | 13:58 | |
*** W3b3r_ has quit IRC | 13:59 | |
*** glangst1 has joined #openstack | 13:59 | |
*** blakeyeager has joined #openstack | 14:02 | |
jonesy | I see that nova uses libvirt, but I'm wondering if there's anything else in the code base that assumes or requires kvm as the virtualization mechanism...? | 14:03 |
creiht | jonesy: it will be plugable | 14:03 |
jonesy | good to know, but the current launchpad stuff assumes kvm? | 14:04 |
*** oz` has quit IRC | 14:04 | |
* jonesy doesn't care, just wants to know so he can set things up as the code downloads | 14:04 | |
creiht | I believe so | 14:04 |
jonesy | thanks! | 14:05 |
creiht | http://nova.openstack.org/ | 14:05 |
creiht | has docs that you can browse while the code is downloading :) | 14:05 |
jonesy | I'm reading them now - that's where I found the references to libvirt and kvm. | 14:08 |
jonesy | thanks! | 14:08 |
creiht | cool | 14:08 |
*** dizzl_ has joined #openstack | 14:09 | |
*** astrolin has quit IRC | 14:09 | |
*** tonywolf has joined #openstack | 14:11 | |
*** murkk has joined #openstack | 14:11 | |
*** glangst1 has quit IRC | 14:14 | |
*** tomh has joined #openstack | 14:14 | |
*** dhdenny has joined #openstack | 14:15 | |
*** jaypipes has joined #openstack | 14:34 | |
*** jdarcy has joined #openstack | 14:35 | |
*** pvo has joined #openstack | 14:36 | |
*** lbieber has joined #openstack | 14:40 | |
*** calavera_ has joined #openstack | 14:40 | |
*** calavera has quit IRC | 14:42 | |
*** calavera_ has quit IRC | 14:46 | |
*** calavera_ has joined #openstack | 14:49 | |
*** jeromatron has quit IRC | 14:50 | |
*** jeromatron has joined #openstack | 14:51 | |
dendrobates | jonesy: the work to start adding xen support is under review now | 14:52 |
*** johnpur has joined #openstack | 14:53 | |
*** ChanServ sets mode: +v johnpur | 14:53 | |
*** calavera_ has quit IRC | 14:54 | |
*** EricInBNE has joined #openstack | 14:54 | |
*** sirp2 has joined #openstack | 14:55 | |
*** sirp2 is now known as sir_p | 14:57 | |
EricInBNE | is swift built on cassandra? | 14:58 |
notmyname | NO! | 14:58 |
creiht | :) | 14:58 |
notmyname | EricInBNE: swift is all python developed in-house and uses packages like eventlet and WebOb | 14:59 |
EricInBNE | notmyname, so rackspace is sticking with it? | 15:00 |
*** jhalogen has joined #openstack | 15:00 | |
matclayton | how does swift actually serve files? | 15:00 |
notmyname | yup. we're running it in production now as Cloud Files | 15:01 |
creiht | matclayton: Through a RESTful api | 15:01 |
creiht | in the same way you would interact with Rackspace's cloud files product | 15:01 |
EricInBNE | notmyname, what does rackspace use cassandra for? | 15:02 |
matclayton | sure, more wondering whether it is sendfile based or now | 15:02 |
matclayton | *not | 15:02 |
creiht | EricInBNE: Not sure, and if we did know, not sure if we could talk about it yet :) | 15:02 |
creiht | matclayton: There are proxy servers in front that serve the files via HTTP | 15:03 |
matclayton | creiht: these python also? | 15:03 |
creiht | indeed | 15:03 |
*** h1tman has joined #openstack | 15:03 | |
matclayton | creiht: any reason why nginx etc aren't used for the proxy part? | 15:04 |
notmyname | matclayton: sendfile isn't a good fit because this is run on multiple servers | 15:04 |
creiht | matclayton: we looked at using nginx for load balancing, but the biggest issue is that nginx spools all content before sending on to backend services | 15:04 |
gholt | nginx spools to disk :( | 15:04 |
creiht | yup | 15:04 |
matclayton | yeah we know! | 15:04 |
gholt | We wanted to patch it up, but as the author of nginx says: It's hard. | 15:05 |
creiht | for web apps that is fine, but for a storage system, not so much :) | 15:05 |
matclayton | creiht: we are an entire python shop, and have ended up using nginx direct to disk, and a proxy infront which essentially redirects the user to the correct nginx machine | 15:05 |
notmyname | swift's use case is pretty mush the opposite of web pages: we have large variable uploads with small downloads | 15:05 |
notmyname | that is, we have to handle large uploads that have a small return payload | 15:06 |
creiht | matclayton: the biggest issue is that we want to store all replicas of an incoming object at once, so in swift as an object comes in throught the proxy, the object is written to all replicas | 15:06 |
matclayton | fair enough, just trying to figure out it is a good fit for what we are trying to do | 15:06 |
creiht | we didn't want to write once, then replicate | 15:06 |
*** pharkmillups has quit IRC | 15:06 | |
creiht | matclayton: no worries, and ask away | 15:07 |
creiht | :) | 15:07 |
matclayton | creiht: makes a lot of sense | 15:07 |
matclayton | so our use case is, we essentially handle ~100Meg files on averge, and have about 250k of these, incoming is about 5Mbit constant and growing | 15:07 |
creiht | sure | 15:08 |
matclayton | outgoing is about 3.6Gig at peak, although we have reduced this a lot recently | 15:08 |
creiht | given the right hardware setup, that should be doable | 15:08 |
matclayton | currently we are ducting it together with a lot of sshfs and nginx | 15:08 |
matclayton | and logic in the app layer | 15:09 |
notmyname | sounds complicated | 15:09 |
creiht | matclayton: http://swift.openstack.org/development_saio.html | 15:09 |
creiht | walks you through setting up a dev vm instance of swift | 15:09 |
matclayton | saw that, biggest concern was python performance reading from disk->network | 15:09 |
creiht | if you would like to try poking at it | 15:09 |
matclayton | need to get it setup and have a play I guess | 15:10 |
notmyname | matclayton: eventlet helps a lot with the IO | 15:10 |
matclayton | cool, in generally what have you found is the limitation? | 15:11 |
matclayton | or what is it bound by? | 15:11 |
creiht | your network and disk io | 15:11 |
*** jbd has joined #openstack | 15:11 | |
creiht | and can be cpu if you are pushing a lot of network (like 10g) | 15:11 |
creiht | the proxies tend to be more cpu bound | 15:12 |
creiht | the storage nodes are more io bound | 15:12 |
matclayton | all our stuff is 1g | 15:12 |
creiht | swift can easily saturate the 1g links | 15:12 |
matclayton | do you guys typically run different proxy/storage nodes? or run on the same boxes? | 15:12 |
jdarcy | Are objects streamed directly to replicas, or buffered at the proxy? | 15:12 |
gholt | Proxies will be very cpu bound if you ask them to do SSL, but you can scale them out. | 15:12 |
creiht | jdarcy: streamed directly | 15:12 |
*** pharkmillups has joined #openstack | 15:13 | |
*** gondoi has joined #openstack | 15:13 | |
creiht | matclayton: we run proxy's separate from the storage nodes, but that is mostly an implementation detail... there is nothing stopping you from running them all on the same server | 15:13 |
creiht | the system is very flexible on that side of things | 15:15 |
matclayton | btw one minor point I found really frustrating when scanning the docs is much like MogileFS there isn't a clear one page saying what each different piece of architecture does and what a complete setup should look like. unless I missed it, meant to be constructive :) | 15:15 |
creiht | matclayton: Oh I hear you... we are working on that :) | 15:15 |
matclayton | sure you are! | 15:15 |
jdarcy | I was fairly impressed by this morning's examination of the code, and am looking forward to taking it for a spin. Nice work, guys. | 15:15 |
gholt | matclayton: Absolutely; we were busy coding. | 15:15 |
creiht | The open source timeline got ramped up, so I haven't had time to work on that yet | 15:16 |
jonesy | thanks, dendrobates -- I'm pretty vm-agnostic (at least for the time being), but good to know that it's not as simple as 'libvirt==anything' | 15:16 |
creiht | jdarcy: awesome, and thanks! | 15:16 |
matclayton | cool, what typical spec of machines do you run each part on? | 15:16 |
* gholt thinks, "Yeah, it's all creiht's fault." | 15:16 | |
creiht | haha | 15:16 |
creiht | matclayton: I can't go into a whole lot of detail, other than the storage boxes are 4U machines with 24 2G spindles | 15:17 |
creiht | proxies are smaller machines that are mostly cpu | 15:17 |
*** JordanRinke has joined #openstack | 15:17 | |
* JordanRinke looks around | 15:17 | |
creiht | JordanRinke: howdy | 15:17 |
* h1tman wave's at JordanRinke | 15:17 | |
*** pengwynn has joined #openstack | 15:18 | |
*** bpiatt has joined #openstack | 15:18 | |
notmyname | matclayton: the point is to optimize proxies for CPU and storage nodes for disk in the most cost-effective way | 15:18 |
matclayton | sure of course, more wondered if they are cheap commodity stuff or high end servers | 15:18 |
creiht | it is designed to run on commodity stuff | 15:18 |
*** dubs has joined #openstack | 15:19 | |
*** EricInBNE has left #openstack | 15:19 | |
creiht | and thusly designed to handle failures :) | 15:19 |
jdarcy | So a single machine with 24 spindles is still I/O-bound rather than CPU-bound? Nice. | 15:19 |
matclayton | thanks, our limits are also the ratio of BW to HD space | 15:19 |
JordanRinke | creiht: Hello | 15:19 |
creiht | jdarcy: yes | 15:20 |
jonesy | Haven't seen any mention of monitoring in docs yet. There are plenty of tools to do it, but just wondering if there's some integrated monitoring component somewhere that has yet to be released? | 15:21 |
creiht | We do pretty standard monitoring with zenoss | 15:21 |
*** dendrobates is now known as dendro-afk | 15:21 | |
jonesy | k | 15:21 |
*** pvo has quit IRC | 15:22 | |
*** pvo has joined #openstack | 15:22 | |
notmyname | jonesy: also, we have a log processor system that can be set up to feed a stats system (but it is yet-to-be-released) | 15:23 |
jonesy | thanks | 15:24 |
matclayton | thanks guys, will try and get a small cluster setup and tested the next week or so. Are the rest enpoints compatitble for the rackspace cloud ones, (wondering if the django rackspace cloud storage classes will work out of the box) | 15:24 |
creiht | matclayton: should be | 15:25 |
notmyname | matclayton: it should be compatible (other than auth differences) | 15:25 |
creiht | It also includes a dev auth server that emulates rackspace auth | 15:25 |
creiht | matclayton: if you run into issues, let us know | 15:25 |
notmyname | what creiht said ^ ;-) | 15:25 |
*** johnblue has joined #openstack | 15:26 | |
matclayton | dont care about auth, will disable that straight away :) will let you know if we have any issues. | 15:26 |
creiht | :) | 15:26 |
gholt | What IP are you going to set it up on? ;) | 15:26 |
creiht | hah | 15:26 |
matclayton | all in a VPN anyways | 15:26 |
*** pvo has quit IRC | 15:27 | |
h1tman | hmm.. does the current release require a specific python version? | 15:28 |
gholt | I believe 2.6 is required atm. | 15:28 |
notmyname | h1tman: we use 2.6 | 15:28 |
h1tman | thanks | 15:28 |
*** iRTermite has joined #openstack | 15:28 | |
JordanRinke | iRTermite: HELLO! | 15:29 |
iRTermite | :P | 15:29 |
h1tman | thought that might be the issue.. I tried running the install on CentOS(2.4.3) as well as Ubuntu, and the CentOS install failed with syntax errors | 15:29 |
h1tman | sorry that was not clear.. the python version on CentOS is 2.4.3 | 15:29 |
jdarcy | I'll be trying on RHEL6, so I'll keep note of issues I find. | 15:30 |
johnblue | Is there a particular OS distro that the install is ez'r on? I prefer FreeBSD ... | 15:30 |
creiht | h1tman: yeah we are aware of that... not sure we can easily make it 2.4 compatibale, but 2.5 compatible should be doable | 15:31 |
*** urandom has joined #openstack | 15:31 | |
* creiht needs to add a getting started page to the docs :) | 15:31 | |
*** cglee has quit IRC | 15:32 | |
creiht | johnblue: We target Ubuntu Server when developing swift, but it *should* work on freebsd, as long as you can get the dependencies | 15:32 |
h1tman | np, just didnt see any documentation on that.. or I am blind | 15:32 |
creiht | h1tman: not it isn't there yet... on my list :) | 15:32 |
h1tman | ah ha! :p | 15:32 |
*** admboom has joined #openstack | 15:33 | |
*** rando has joined #openstack | 15:34 | |
*** dendro-afk is now known as dendrobates | 15:35 | |
johnblue | creiht: roger. I haven't started yet .. still reading about OpenStack trying to get my arms around it. Has a list of dependencies already been created? Could whip up a BSD install port .. | 15:35 |
creiht | johnblue: not yet, unless you look in the debian packaging stuff for it :) | 15:35 |
johnblue | creiht: ;) *smack* lol. | 15:36 |
*** hunter has joined #openstack | 15:36 | |
*** pvo has joined #openstack | 15:36 | |
creiht | main dependencies are python 2.6 and the latest eventlet, python-xattr, webob, simplejson | 15:37 |
jaypipes | mornin folks. :) | 15:40 |
*** epicurus has joined #openstack | 15:40 | |
creiht | morning | 15:40 |
johnblue | morn'n. Had your your cup O joe yet? :D | 15:41 |
jaypipes | johnblue: yep :) | 15:41 |
jaypipes | johnblue: actually, already had a 5-mile walk, breakfast, and two cups of coffee this morning :) ready to go! | 15:41 |
*** JordanRinke has quit IRC | 15:41 | |
jaypipes | johnblue: oh, and fixed a bug too ;) | 15:41 |
*** silassewell has joined #openstack | 15:42 | |
jdarcy | I tried to fix a bug once, but the tweezers kept shaking. | 15:42 |
johnblue | jaypipes: ummm .. I'll just be over here --->. lol. PT in the morn'n gotta love it bro! | 15:42 |
jaypipes | jdarcy: lol | 15:42 |
* jaypipes laughing a bit at all the hubbub about "open core" vs "open source"... | 15:44 | |
*** Predominant has quit IRC | 15:44 | |
jdarcy | I've only looked at Swift so far. Does Nova store its images there, or can it? | 15:49 |
creiht | jdarcy: that is an upcomming milestone | 15:49 |
*** tobym has joined #openstack | 15:51 | |
jaypipes | jdarcy: Nova's image store is pretty adaptable. In the next milestones, Swift will be an image store possibility (currently it uses s3), and I'm sure other options will materialize in the next 6-12 months. | 15:52 |
tobym | Is there a good description somewhere describing differences between Compute & Storage APIs vs EC2 and S3 APIs? | 15:52 |
jaypipes | tobym: the Compute API is currently the ec2 API, but the Rackspace API will become the canonical API shortly (without dropping support for ec2 api of course) | 15:53 |
mtaylor|beach | jaypipes: you should already have indirect memebership in openstack-planet via swift-core or nova-core | 15:53 |
jdarcy | tobym: Just as one data point, I already had some CURL-based code to talk to S3. It took me half a day on Friday to adapt it to use CloudFiles as well. | 15:54 |
jaypipes | mtaylor|beach: doesn't work. readonly error. | 15:54 |
creiht | tobym: http://www.rackspacecloud.com/blog/2010/06/15/a-close-look-at-the-rackspace-cloud-servers-api-and-how-it-compares-to-amazons-ec2-api/ | 15:54 |
mtaylor|beach | jaypipes: weird. looking in to it then | 15:54 |
jaypipes | mtaylor|beach: I wouldn't have edited that wiki page if everything had worked as described :P | 15:54 |
creiht | Is a good post that describes how the rackspace compute apis differ | 15:54 |
mtaylor|beach | jaypipes: oh - you aren't actually in swift-core or nova-core | 15:54 |
jdarcy | tobym: It wasn't even what I'd call a hard day's work. Basically they're similar enough semantically that if you know one you can rip through the API differences easily. | 15:54 |
jaypipes | creiht: nice article... ++ | 15:54 |
tobym | jaypipes, creiht, jdarcy: thanks | 15:55 |
creiht | tobym: and the storage api is similar | 15:55 |
mtaylor|beach | jaypipes: try again | 15:55 |
jaypipes | mtaylor|beach: done. cheers. | 15:57 |
mtaylor|beach | jaypipes: how had I not added you to swift-core or nova-core before now... | 15:57 |
mtaylor|beach | sigh | 15:57 |
mtaylor|beach | jaypipes: how do you like the add-your-own-damn-self process there? | 15:57 |
jaypipes | mtaylor|beach: because the team was called ozone, then ozone-core, then nova-core, then nova? ;) | 15:57 |
mtaylor|beach | jaypipes: well, yeah | 15:57 |
johnblue | super-nova ;) | 15:57 |
mtaylor|beach | jaypipes: nope - nova-core and nova are two separate teams | 15:57 |
jaypipes | mtaylor|beach: it was fine. I updated the wiki page with proper instructions, though ;) | 15:57 |
mtaylor|beach | jaypipes: sweet | 15:58 |
jaypipes | mtaylor|beach: s/config.ini/planet.ini | 15:58 |
jaypipes | mtaylor|beach: s/heads/images | 15:58 |
mtaylor|beach | jaypipes: heh. thanks :) | 15:58 |
jaypipes | mtaylor|beach: np | 15:58 |
* jaypipes just glad the damn thing is public now and we don't need to hide for some PR release :) | 15:59 | |
mtaylor|beach | jaypipes: zomg. yes | 15:59 |
creiht | haha | 15:59 |
pvo | mtaylor|beach: no kidding | 15:59 |
jaypipes | pvo: mornin. how's the weather in SA today? hotter than Austin last week?! | 15:59 |
pvo | in Portland.... who knows about SA. Probably hot. : ) | 16:00 |
* jaypipes enjoying Columbus weather... | 16:00 | |
jaypipes | pvo: ah, yes! OSCON :) | 16:00 |
pvo | weather here is beyootiful | 16:00 |
jaypipes | pvo: nice :) | 16:00 |
* jaypipes loves Portland. | 16:00 | |
jaypipes | anyone know if Josh McKenty is at OSCON? | 16:00 |
pvo | jaypipes: dunno | 16:01 |
jdarcy | What's a practical minimum for running some tests? Four nodes? | 16:01 |
pvo | he was bouncing off irc late last night | 16:01 |
creiht | jdarcy: what types of tests would you like to run? | 16:02 |
creiht | you can run everything on one machine if you want to just play around with it | 16:02 |
creiht | if you want to start playing with a cluster, then 4-5 nodes is a pretty good start | 16:02 |
creiht | though we don't have good docs for running a cluster yet, so if you do that, let me know if you have any questions, or run into issues | 16:03 |
*** Predominant has joined #openstack | 16:03 | |
*** mtaylor|beach has quit IRC | 16:04 | |
jdarcy | I think I'll want to run some very basic perf tests, just so I can do a fair comparison between Swift and Hail (S3-compatible) as a back end for another project, so I'll probably scrounge up four actual machines. | 16:06 |
*** taf2 has joined #openstack | 16:06 | |
*** mtaylor|beach has joined #openstack | 16:06 | |
jaypipes | jdarcy: I'm working on end-to-end performance regression testing for nova. I'll be adding in the same for swift as well. I'll update the ML when progress is sufficient to show off. | 16:07 |
*** mtaylor|beach has quit IRC | 16:07 | |
*** mtaylor has joined #openstack | 16:07 | |
*** ChanServ sets mode: +v mtaylor | 16:07 | |
jdarcy | jaypipes: Cool, that'll be helpful. | 16:08 |
jaypipes | jdarcy: we want to get to the point where Tarmac/Hudson simply fires off a regression run for each commit into staging and automatically logs a history of performance/scalability data over time. | 16:08 |
*** andrewvhs has quit IRC | 16:08 | |
jaypipes | jdarcy: hudson.openstack.org | 16:08 |
jaypipes | jdarcy: automation++ | 16:08 |
mtaylor | jaypipes: when you get it - I also want a pass/fail response from that system | 16:08 |
mtaylor | jaypipes: so that we can properly hook it in to the pre-merge testing hook | 16:09 |
jaypipes | jdarcy: we do the same for Drizzle over at hudson.drizzle.org and I wrote automation software for checking regressions for Drizzle that is very helpful to show regressions (for instance: https://lists.launchpad.net/drizzle-benchmark/msg03027.html). Hoping to do the same for openstack. | 16:09 |
creiht | jdarcy: when you get around to doing that, let us know, as there are some system tweeks that make a huge difference | 16:09 |
*** croteb has joined #openstack | 16:10 | |
mtaylor | jaypipes: also - there are some hudson plugins that will display the graphing trends and stuff for us ... I'd love to chat with you on making sure your system can output things in a format that hudson can read and process | 16:11 |
mtaylor | or at least can do so optionally | 16:11 |
jaypipes | mtaylor: hells yeah. | 16:13 |
*** ttx has quit IRC | 16:15 | |
jdarcy | creiht: Will do. Thanks! | 16:15 |
mtaylor | jaypipes: also - just while I'm making feature requests on software you haven't released yet... | 16:15 |
mtaylor | jaypipes: if you're going to model it after d-a at all, it would be great if there were at least a mode where, if I'm running it from hudson, I can let hudson handle the code branching and then run your software on a given branch, cause right now d-a and hudson are duplicating effort (although I know why d-a does what it does) | 16:16 |
jaypipes | mtaylor: understood. | 16:17 |
*** kuadrosx has joined #openstack | 16:17 | |
*** jdmaturen has joined #openstack | 16:18 | |
*** jdmaturen has joined #openstack | 16:18 | |
*** lbieber has quit IRC | 16:20 | |
*** bd_ has joined #openstack | 16:26 | |
*** pvo has quit IRC | 16:30 | |
*** pvo has joined #openstack | 16:30 | |
*** docgnome has joined #openstack | 16:32 | |
*** barismetin has quit IRC | 16:32 | |
*** docgnome has left #openstack | 16:33 | |
*** pharkmillups has quit IRC | 16:33 | |
*** pvo has quit IRC | 16:35 | |
justinsb | "qemu: linux kernel too old to load a ram disk" Anyone seen that from KVM while trying to launch an instance? | 16:35 |
*** pvo has joined #openstack | 16:39 | |
*** joshuamckenty has joined #openstack | 16:39 | |
*** tobym has quit IRC | 16:40 | |
*** joshuamckenty has quit IRC | 16:40 | |
*** joshuamckenty has joined #openstack | 16:41 | |
eday | justinsb: nope, but I can guess what the fix would be :) | 16:42 |
*** sirp3 has joined #openstack | 16:42 | |
*** enmand has joined #openstack | 16:43 | |
*** joshuamckenty has quit IRC | 16:43 | |
*** joshuamckenty has joined #openstack | 16:44 | |
*** Goldfisch has joined #openstack | 16:44 | |
*** Goldfisch has left #openstack | 16:44 | |
justinsb | eday: Upgrading the kernel? I'm on Lucid Lynx, so I'd be surprised if that was the fix... | 16:44 |
*** Goldfisch has joined #openstack | 16:45 | |
*** sir_p has quit IRC | 16:45 | |
*** murteas has joined #openstack | 16:49 | |
*** joshuamckenty has quit IRC | 16:51 | |
eday | justinsb: ahh, ok. I'm not sure then | 16:51 |
jaypipes | mtaylor: https://launchpad.net/~openstack-discuss needs non-privating... | 16:53 |
mtaylor | jaypipes: actually, working on that right now | 16:53 |
*** PiotrSikora has joined #openstack | 16:53 | |
jaypipes | mtaylor: cheers | 16:53 |
mtaylor | jaypipes: we're gonna rename ~openstack to ~openstack-admins and make a new open ~openstack team (so that it'll match the ~swift/~nova open team approach) | 16:54 |
mtaylor | jaypipes: then we'll hang the mailing list off of ~openstack | 16:54 |
mtaylor | jaypipes: but yes. | 16:54 |
jaypipes | mtaylor: l | 16:54 |
jaypipes | mtaylor: k | 16:55 |
PiotrSikora | guys, i'm not sure if you've noticed, but ~swift overview page isn't accessible by anonymous users while ~nova and ~openstack are... | 17:00 |
*** Predom[osx] has joined #openstack | 17:00 | |
*** SEJeff has quit IRC | 17:00 | |
mtaylor | PiotrSikora: hrm. that should change... | 17:01 |
*** poli has joined #openstack | 17:01 | |
poli | And this was just announced today?! | 17:02 |
mtaylor | PiotrSikora: aroo? it works for me? | 17:03 |
mtaylor | PiotrSikora: (as an anonymous user...) | 17:03 |
*** Predominant has quit IRC | 17:04 | |
*** Predom[osx] is now known as Predominant | 17:04 | |
PiotrSikora | mtaylor: erm... sorry... ~ works, but /swift doesn't (but /nova and /openstack does) | 17:06 |
mtaylor | PiotrSikora: ah. ok. thanks | 17:06 |
mtaylor | PiotrSikora: thanks. I've filed a ticket to look in to that. very odd | 17:08 |
*** dendrobates is now known as dendro-afk | 17:09 | |
PiotrSikora | np :) | 17:10 |
*** nickmbailey has joined #openstack | 17:11 | |
*** greenisus has joined #openstack | 17:14 | |
*** bretpiatt has joined #openstack | 17:17 | |
mtaylor | creiht: ok. cut the 1.0.1... I'll write up the walkthrough of that any old time now... but writing docs are less interesting than the beach :) | 17:17 |
*** tupshin has quit IRC | 17:24 | |
*** Goldfisch has left #openstack | 17:29 | |
*** issackelly has left #openstack | 17:30 | |
*** tobym has joined #openstack | 17:30 | |
*** g0rdy has joined #openstack | 17:35 | |
johnblue | mtaylor: sand, water, sun .. all over-rated. ;) | 17:38 |
*** clayg has quit IRC | 17:38 | |
*** clayg_ has joined #openstack | 17:39 | |
*** tupshin has joined #openstack | 17:42 | |
PiotrSikora | hmm... i've got a few questions about design of swift... | 17:43 |
PiotrSikora | 1) why do you guys use X-Storage-{User,Pass} instead of Authorization header? | 17:43 |
PiotrSikora | 2) why do you use rsync'ed sqlite instead of mysql/drizzle/postgres for object database? | 17:43 |
notmyname | PiotrSikora: the auth system we included is compatible with the one that Rackspace Cloud uses | 17:44 |
PiotrSikora | notmyname: heh, i thought so... | 17:44 |
notmyname | PiotrSikora: as to sqlite->it's simple, fast, and good enough. the complexity of the others was too high a cost | 17:45 |
poli | two releases in 12 hours. I am falling in love with this project ;) | 17:45 |
*** tupshin has quit IRC | 17:45 | |
*** joshuamckenty has joined #openstack | 17:45 | |
PiotrSikora | but it doesn't make it really good open standard when its just open-sourced software that was make to meet specific needs of a single company | 17:46 |
PiotrSikora | not that i don't appreciate the you work, but it seems that mogilefs or cornea are still better "general" fit than swift | 17:47 |
notmyname | true. but up until now, it has been for one company ;-) patches are welcome | 17:47 |
notmyname | I would need to see better performance for a more complicated db system before I would be interested in including it in the main trunk, though | 17:48 |
creiht | PiotrSikora: It depends on your use case... for example mogilefs can not scale to the level of storage that we required | 17:48 |
PiotrSikora | creiht: do you mean performance problems with metadata storage? | 17:48 |
notmyname | we didn't choose sqlite because of company needs, per se. I mean, we have drizzle and cassandra devs working here. | 17:48 |
creiht | mogilefs is likely a good candidate for a lot of use cases | 17:48 |
PiotrSikora | this is easily solved by sharding metadata database | 17:49 |
redbo | "easily" | 17:49 |
PiotrSikora | it was good enough for yahoo china ;) | 17:49 |
creiht | PiotrSikora: it has been a while since I have looked at it, but for example the centralized tracker makes it hard to scale | 17:49 |
PiotrSikora | creiht: AFAIK tracker isn't centrailized | 17:49 |
creiht | and replace hard with other things like expensive | 17:50 |
*** matclayton has left #openstack | 17:50 | |
PiotrSikora | the bottleneck was always metadata database, but this is sharedable(?) ;) | 17:50 |
*** joshuamckenty has quit IRC | 17:51 | |
redbo | So... what happens after my patch has been approved for merge? | 17:51 |
creiht | redbo: it should get automerged | 17:51 |
notmyname | I think *magic* | 17:51 |
redbo | when does that happen? | 17:51 |
redbo | and what happens if there's a conflict? | 17:52 |
*** oz` has joined #openstack | 17:52 | |
creiht | mtaylor: -^ | 17:53 |
creiht | PiotrSikora: Like I said... MogileFS is probably a reasonable solution for many people and many problems. It wasn't for us | 17:53 |
creiht | nor did it work for NASA | 17:54 |
PiotrSikora | ;) | 17:54 |
creiht | and I'm a big fan of the danga guys (we use memcached quite a bit for example) | 17:54 |
bretpiatt | PiotrSikora swift does not have a central metadata database, the database is shared across all of the nodes in the cluster so it is already "sharded" | 17:55 |
mtaylor | creiht: arroo? | 17:55 |
creiht | mtaylor: if a merge request is accepted, but auto-merge fails, what happens? | 17:56 |
PiotrSikora | bretpiatt: yeah... but that's pretty much what yahoo did with their mogilefs fork (ylem) | 17:56 |
PiotrSikora | sharding metadata database | 17:56 |
mtaylor | creiht: then tarmac makes a comment on the merge request and sets the status back away from approved | 17:56 |
creiht | oh so they had to fork mogilefs to make it work for them? :) | 17:56 |
mtaylor | creiht: the comment will contain the failure | 17:56 |
creiht | mtaylor: ahh cool thanks | 17:56 |
PiotrSikora | creiht: since sharding isn't present in default mogilefs release, then i guess so... ;) | 17:57 |
redbo | What we have now is essentially a metadata table sharded by container, with automatic replication and balancing. It's not a real distinction. | 17:57 |
creiht | redbo: nicely put :) | 17:58 |
*** lbieber has joined #openstack | 17:58 | |
mtaylor | would have been nice if yahoo.cn had contributed back their sharding to mogile... | 17:59 |
PiotrSikora | mtaylor: whole project is here: http://github.com/chaoslawful/ylem | 17:59 |
PiotrSikora | AFAIK it was released after yahoo.cn social projects were canceled, so its not maintained whatsoever | 18:00 |
mtaylor | PiotrSikora: great. but that's just a random renamed fork ... if they're going to publish - why not work with upstream? | 18:00 |
mtaylor | I know all of the mogile guys, I'm sure they would have been happy to receive and then _maintain_ patches | 18:00 |
* mtaylor grumbles, rants and tosses arms around like a crazy me | 18:00 | |
mtaylor | man | 18:01 |
mtaylor | it would be great if I could type | 18:01 |
mtaylor | :) | 18:01 |
PiotrSikora | mtaylor: don't ask me, i have no answers :P | 18:01 |
mtaylor | PiotrSikora: don't mind me - I'm just crotchey - which I think means it's time for lunch :) | 18:01 |
*** mtaylor is now known as mtaylor|lunch | 18:01 | |
PiotrSikora | mtaylor|lunch: bon appetit | 18:02 |
PiotrSikora | oh... and one more question... | 18:02 |
PiotrSikora | since i wasn't able to find this in docs | 18:03 |
PiotrSikora | or actually, nvm | 18:03 |
*** rdw has joined #openstack | 18:03 | |
notmyname | PiotrSikora: ask. we may need to add it to the docs or someone else may want to know | 18:05 |
*** blakeyeager_ has joined #openstack | 18:06 | |
PiotrSikora | notmyname: erm... i wanted to ask how do you serve the files to end-user | 18:07 |
*** pvo has quit IRC | 18:07 | |
PiotrSikora | kind of high-level flow overview | 18:07 |
*** pvo has joined #openstack | 18:07 | |
notmyname | request -> proxy -> (account|container|object) -> proxy -> response | 18:08 |
*** blakeyeager has quit IRC | 18:09 | |
*** blakeyeager_ is now known as blakeyeager | 18:09 | |
PiotrSikora | proxy being python wsgi app? | 18:09 |
notmyname | yes | 18:09 |
notmyname | the proxy server code | 18:09 |
redbo | Mogilefs+sharding is very similar to our original architecture, dozens of db servers, they were a constant headache for availability, scale, and maintenance. | 18:09 |
*** eleanor has joined #openstack | 18:10 | |
*** adrian_otto has joined #openstack | 18:10 | |
notmyname | the proxy handles the requests and serves out to the appropriate account, container, or object servers and then figures out the best response to send to the client (based on what the storage servers responded with) | 18:10 |
redbo | You can make it work (youtube notably does) | 18:10 |
*** tomh has left #openstack | 18:10 | |
PiotrSikora | ok, so let me ask the ultimate question, how can 3rd party application locate requested file without touching your proxy app? | 18:11 |
*** rtyler has joined #openstack | 18:11 | |
PiotrSikora | is this even possible? | 18:11 |
rtyler | creiht: why didn't you do it in my favorite way? | 18:11 |
* creiht smacks rtyler | 18:11 | |
*** silassewell has quit IRC | 18:11 | |
redbo | that's what... aww... | 18:11 |
*** silassewell has joined #openstack | 18:11 | |
* rtyler cries and runs back to #eventlet | 18:11 | |
_0x44 | creiht: Yeah, you could have written it in PHP and chewing gum! | 18:11 |
*** rtyler has left #openstack | 18:12 | |
*** bytee has joined #openstack | 18:12 | |
creiht | PiotrSikora: if the 3rd party application is outside the system, then it has to use the proxy | 18:12 |
notmyname | PiotrSikora: the 3rd party app would essentially have to act like a proxy server: look up the right server(s) in the ring | 18:12 |
bytee | hi! the Object Store (swift) - what is this based on, drizzle? i can't seem to find more details online | 18:12 |
*** pvo has quit IRC | 18:12 | |
adrian_otto | PiotrSikora: It's possible to bypass the proxy if you know where the content is stored. | 18:12 |
creiht | bytee: It is custom built using python | 18:12 |
notmyname | bytee: not based on anything. completely in house | 18:12 |
creiht | and some other libs and tools | 18:12 |
bytee | creiht, notmyname ah ok, thanks | 18:13 |
adrian_otto | but you essentially need the function of the proxy built into the app to do that | 18:13 |
creiht | plus the proxy holds all the smarts to handle failure, etc. | 18:13 |
PiotrSikora | and access to all sqlite databases? | 18:13 |
notmyname | no, those are handled by the account and container servers | 18:13 |
PiotrSikora | or do you also rolled-up your own "storage server"? | 18:14 |
adrian_otto | creiht: exactly, which keeps the clients simple, which is a key maintenance benefit. | 18:14 |
notmyname | PiotrSikora: I used "storage server" as a stand-in for account, container, and object servers | 18:14 |
*** gondoi has quit IRC | 18:15 | |
creiht | each "server" is a self contained service, responsible for its own innards | 18:15 |
creiht | so the proxy doesn't have to know anything about sqlite | 18:15 |
creiht | you could for example replace the account and container to use some other technology if you like | 18:16 |
creiht | as long as it exposed the same RESTful interface | 18:16 |
PiotrSikora | wouldn't it be more efficient to use webdav or existing webservers infrastructure to do that? | 18:16 |
*** patcito has joined #openstack | 18:16 | |
PiotrSikora | i would imagine that calling python for all this stuff isn't "free" | 18:17 |
glange | import free | 18:17 |
adrian_otto | PiotrSikora: trouble with using WebDav is then you would need to come up with a way to handle the per object metadata | 18:17 |
adrian_otto | the existing solution solves the metadata issue with XFS | 18:17 |
*** KnuckleSangwich has joined #openstack | 18:17 | |
creiht | PiotrSikora: Python isn't a bottle neck. You run into problems with network and disk io long before running into issues with python | 18:18 |
adrian_otto | so that you get an atomic window to updating objects and related metadata together | 18:18 |
creiht | the metadata is stored in xattrs which is available in most filesystems now adays | 18:18 |
KnuckleSangwich | exciting stuff guys! | 18:19 |
creiht | KnuckleSangwich: thanks! we think so too ;) | 18:19 |
PiotrSikora | adrian_otto: ah, you're probably right :) | 18:19 |
creiht | PiotrSikora: Most of these components have been through many iterations of different designs | 18:20 |
creiht | to get to where we are today | 18:20 |
PiotrSikora | creiht: yeah... it isn't bottle-neck, but it also isn't free ;) | 18:21 |
creiht | If someone finds a better way to do any component, our ears are open :) | 18:21 |
creiht | PiotrSikora: it is good enough TM :) | 18:22 |
PiotrSikora | but the reason why i'm asking all those question is because with cornea (and mogilefs, i think) i was able to write nginx module that would issue single query to memcached/mysql/postgresql to get file location (node address + path) | 18:22 |
PiotrSikora | and it would serve it right away from the node | 18:22 |
johnblue | Does swift talk directly to curl or to libcurl via a python interface? | 18:22 |
PiotrSikora | without involving number of different services to get to the actual file | 18:23 |
creiht | johnblue: swift provides a restful interface which you can hit using curl if you want... internally we do not use curl | 18:23 |
creiht | PiotrSikora: You are still talking about a design that doesn't inheritly scale easily | 18:23 |
creiht | and when I say scale, I'm talking billions of objects in 100's of petabytes | 18:24 |
PiotrSikora | well, i still assume sharded version of mogilefs/cornea ;) | 18:24 |
PiotrSikora | and i assume that rdbms scale as good as sqlite | 18:25 |
PiotrSikora | which might be wrong assumption :P | 18:25 |
creiht | PiotrSikora: If it works for you, then that is great :) | 18:25 |
PiotrSikora | ;) | 18:26 |
PiotrSikora | hmm... is sqlite filename always the same? | 18:27 |
PiotrSikora | or does it change over time? | 18:27 |
*** chmouel has joined #openstack | 18:28 | |
creiht | PiotrSikora: it should be the same | 18:28 |
creiht | and also remember each sqlite db is responisble for just one container | 18:28 |
spike | kind of OT question, but I saw it mentioned several times and I'm failing at googling it: what's tarmac? | 18:28 |
spike | part of the CI? hudson plugin? | 18:29 |
creiht | spike: tarmac is a bot for launchpad that assists with workflow type things | 18:29 |
spike | ah, ok | 18:29 |
creiht | it will spawn hudson jobs when new code is merged for example | 18:29 |
creiht | those container dbs are thusly spread evenly accross the cluster | 18:29 |
PiotrSikora | hmm... than i could probably use libdrizzle's sqlite->drizzle server | 18:29 |
PiotrSikora | so it would act as rdbms | 18:30 |
PiotrSikora | ^^ | 18:30 |
creiht | PiotrSikora: if you really wanted to, sure :) | 18:30 |
*** justinsheehy has joined #openstack | 18:30 | |
redbo | why do you want it to be a database server process? | 18:30 |
PiotrSikora | i want to be able to query it remotely | 18:31 |
creiht | what types of queries? | 18:31 |
PiotrSikora | "get me location of that file" type of queries ;) | 18:31 |
glange | PiotrSikora: you could put a database in front of swift :) | 18:32 |
*** Manganeez has joined #openstack | 18:32 | |
creiht | if you know an account, container, and object name, then that can be obtained from the ring | 18:32 |
* creiht smacks glange | 18:32 | |
rdw | does your hudson instance automatically file bugs for you? | 18:32 |
creiht | rdw: hah! | 18:32 |
glange | PiotrSikora: you could name that database server, "bottleneck" :) | 18:32 |
*** scotticus has joined #openstack | 18:32 | |
PiotrSikora | glange: lol ;) | 18:32 |
creiht | PiotrSikora: and we actually have command line tools that make it easy to find information like that | 18:33 |
bytee | where does OpenStack see a traditional rdbms like MySQL fitting in? | 18:34 |
PiotrSikora | bytee: it doesn't :P | 18:35 |
PiotrSikora | creiht: cool, i'll look into it | 18:35 |
bytee | PiotrSikora: ok, and drizzle, then? | 18:35 |
redbo | The object layer is basically a distributed hash table. The sqlite stuff is an index of that. | 18:35 |
* bytee hasn't read up on everything... trying to sneak a peek, etc. while at oscon | 18:35 | |
PiotrSikora | bytee: same, and same for postgresql... it's sqlite all the way ;) | 18:36 |
creiht | bytee: Compute and Storage are the first two projects that fit into openstack. I can imagine that other things will eventually fit in as well, but I don't know if any other parts are defined yet | 18:36 |
justinsb | I'm working on an traditional rdbms backend for Nova Compute. Might not prove applicable to Rackspace scale, but hopefully useful for private clouds. But it's an open source project, so I think it's more about sensible contributions than policy | 18:37 |
PiotrSikora | bytee: duh... sorry, guess i mis-understood your question | 18:37 |
*** dec3pti01 has joined #openstack | 18:37 | |
PiotrSikora | creiht: but yeah... swift definiately needs some high-level overview like the one mogilefs has (http://code.google.com/p/mogilefs/wiki/HighLevelOverview) | 18:38 |
creiht | PiotrSikora: it is on my list :) | 18:38 |
PiotrSikora | great | 18:38 |
creiht | and you aren't the first to ask | 18:38 |
bytee | creiht: is the Storage code available ? i got a launchpad login note (or it could just be interent - lp seems slow) | 18:38 |
*** joshuamckenty has joined #openstack | 18:39 | |
creiht | mtaylor|lunch: I've seen what bytee is talking about before, to access swift you have to be logged into launchpad | 18:39 |
creiht | bytee: at the moment I think you just have to be logged in to launchpad | 18:39 |
rdw | and their login interface is terrible | 18:39 |
PiotrSikora | bytee: bzr branch lp:swift | 18:39 |
creiht | bytee: you can also browse current docs at http://swift.openstack.org | 18:40 |
* bytee just has a selfish interest to see where mysql/drizzle might fit in, that's all ;) | 18:40 | |
PiotrSikora | creiht: mtaylor|lunch already filled ticket for that ;) | 18:40 |
bytee | PiotrSikora: ok, lemme try that though i can imagine this branch taking ages | 18:40 |
johnblue | bytee: have ya got any cool bling from OSCON yet? | 18:40 |
bytee | johnblue: i took my bag yesterday, and didn't get a chance to look at it yet, sorry | 18:41 |
johnblue | ;) | 18:41 |
creiht | bytee: if you are at oscon, I highly recommend going to Will Reeses talk (one of my coworkers) | 18:41 |
bytee | creiht: i will. lemme check the schedule | 18:42 |
creiht | he will give an overview of the architecture and how we got there | 18:42 |
bytee | creiht: thanks a bunch. i think i'll also go to the summit tomorrow | 18:42 |
creiht | cool | 18:42 |
*** ChanServ changes topic to "Find out more and get involved here: http://www.openstack.org | Nova Docs: nova.openstack.org | Swift Docs: swift.openstack.org | Wiki: wiki.openstack.org" | 18:43 | |
creiht | time for a meeting :) | 18:44 |
exlt | ls -al | 18:47 |
exlt | ugh. | 18:47 |
* exlt is tired of fighting with bzr-buildpackage for the day | 18:47 | |
*** dendro-afk is now known as dendrobates | 18:49 | |
*** greenisus has quit IRC | 18:49 | |
jaypipes | dendrobates: question for ya.. | 18:51 |
soren | exlt: Need help? | 18:52 |
*** Twitch has joined #openstack | 18:53 | |
exlt | soren: pristine-tar is creating a very small orig.tar.gz and the diff.gz ends up with lots that should be in orig.tar.gz - that's the biggest issue | 18:56 |
exlt | so I started looking for bzr export --exclude=.. that doesn't seem to exist | 18:57 |
exlt | so I'm reading on pristine-tar delta - didn't really think the default of "everything but /debian dir and .bzr" would fail so badly | 18:59 |
soren | I've never used pristine-tar directly. | 18:59 |
*** pvo has joined #openstack | 18:59 | |
*** pvo has quit IRC | 18:59 | |
*** pvo has joined #openstack | 18:59 | |
*** dendrobates is now known as dendro-afk | 18:59 | |
*** dendro-afk is now known as dendrobates | 19:00 | |
jaypipes | dendrobates: around? | 19:00 |
exlt | bzr-buildpackage calls pristine-tar to generate the orig.tar.gz from the bzr repo | 19:00 |
* jaypipes has a feeling OSCON-itis has affected many colleagues... ;) | 19:01 | |
exlt | and I've never used it directly, either | 19:01 |
pvo | jaypipes: whats shakin? | 19:01 |
pvo | I haven't seen him | 19:01 |
comstud | jay, yeah... SOME of us are actually working! | 19:01 |
*** soulresin has joined #openstack | 19:01 | |
comstud | oh... hi pvo | 19:01 |
comstud | :) | 19:01 |
pvo | comstud: hey there | 19:01 |
jaypipes | pvo: oh, just wondering about blueprints and nova project on Launchpad...should I throw a ping to the openstack-discuss ML about ones I've created and want feedback on, or...something else? | 19:02 |
eday | jaypipes: he's here next to me, but talking with folks | 19:02 |
jaypipes | comstud: morning Chris. | 19:02 |
comstud | mornin.. or well.. it's after noon now | 19:02 |
pvo | jaypipes: I would. Drop them here too to get more eyes on it. | 19:03 |
jaypipes | ah, well, yes it is.. | 19:03 |
jaypipes | pvo: cool, will do. | 19:03 |
comstud | agree about the config variables | 19:04 |
comstud | although it sounds like someone is working on ditching gflags | 19:04 |
comstud | replacing with what.. i dunno. | 19:05 |
* johnblue just sold my old samsung blu-ray player via craigslist .. loving the streaming netflix on the new samsung. | 19:05 | |
comstud | there's all sorts of abstraction work to do | 19:05 |
jaypipes | comstud: I'm not going to get into those battles...just suggesting grouping the mq-based config and getting rid of the boolean fake_rabbit stuff in favour of a real adapter model. | 19:06 |
jaypipes | johnblue: I've got the new samsung. awesome to play pandora straight through it. :) | 19:06 |
comstud | agreed | 19:06 |
soren | exlt: Right, I've always had a dsc I wanted to import. | 19:07 |
johnblue | jaypipes: rock on. I got the C5500 about a week ago and its been sweet. | 19:07 |
* comstud just got a ps3 | 19:07 | |
*** murteas has left #openstack | 19:08 | |
mtaylor|lunch | bytee: hey! | 19:08 |
*** silassewell has quit IRC | 19:08 | |
mtaylor|lunch | exlt: hey there | 19:08 |
mtaylor|lunch | exlt: sorry - didn't read the whole scrollback... | 19:09 |
mtaylor|lunch | exlt: you may want to merge the latest thing I just put up for merge | 19:09 |
mtaylor|lunch | exlt: should have much less diff | 19:09 |
mtaylor|lunch | exlt: bzr merge lp:~mordred/swift/release-1.0.1 | 19:09 |
*** mtaylor|lunch is now known as mtaylor | 19:09 | |
exlt | ah, I was merging from trunk | 19:10 |
johnblue | jaypipes: are ya running an HTPC? I did some personal data recovery for co-worker and got him to spring for a gigabyte H55-USB3 + RAM. Haven't fired it up yet but it has an HDMI out option. | 19:12 |
jaypipes | johnblue: hehe, not sure what that is...so, I guess no :) | 19:13 |
mtaylor | exlt: the lastest merge-upstream hasn't hit trunk yet | 19:13 |
johnblue | jay: hehe .. no worries. http://en.wikipedia.org/wiki/Home_theater_PC | 19:13 |
jaypipes | mtaylor: openstack-discuss going to be non-private? | 19:14 |
jaypipes | johnblue: ah! :) | 19:14 |
jaypipes | johnblue: negative on that one :) | 19:14 |
johnblue | :D | 19:14 |
mtaylor | jaypipes: yes. but spm isn't awake yet | 19:14 |
jaypipes | mtaylor: ah, k | 19:14 |
mtaylor | jaypipes: and they haven't given me root on launchpad yet :) | 19:14 |
jaypipes | mtaylor: oh, why not? ;0 | 19:15 |
creiht | mtaylor: https://code.launchpad.net/~redbo/swift/607186/+merge/30260 | 19:15 |
creiht | I approved that hours ago, but it never got merged | 19:15 |
mtaylor | creiht: it is missing a commit message | 19:15 |
creiht | hah | 19:15 |
* mtaylor really needs to add an option to tarmac to deal with that better | 19:16 | |
mtaylor | creiht: ok. it's on my todo list... at the moment I don't feel it's a huge value add to always require both a description and a commit message | 19:18 |
creiht | k | 19:19 |
mtaylor | creiht: I'm going to see if I can't make it cook up a default commit message if there isn't one | 19:19 |
creiht | could it pull in the description if the commit message isn't there? | 19:19 |
*** gondoi has joined #openstack | 19:20 | |
mtaylor | creiht: yeah - that's what I was thinking | 19:20 |
creiht | cool | 19:20 |
*** tupshin has joined #openstack | 19:21 | |
*** joshuamckenty has quit IRC | 19:21 | |
*** soulresin has left #openstack | 19:22 | |
*** joshuamckenty has joined #openstack | 19:23 | |
*** dendrobates is now known as dendro-afk | 19:23 | |
creiht | mtaylor: did you see the merge failure? | 19:24 |
mtaylor | yeah - fixing now | 19:24 |
creiht | cool | 19:24 |
creiht | now that I have changed his commit message, do I need to change the status for hudson to see it? | 19:24 |
mtaylor | creiht: nope. as long as it's approved, it'll get picked up | 19:26 |
creiht | yeah it just did... cool | 19:26 |
jdarcy | From reading the code, it seems like a container GET at the proxy causes the proxy to fan out to a bunch of storage nodes and collate the results. Is that accurate? | 19:26 |
jdarcy | Swift code, that is. | 19:26 |
redbo | yay! how long until it disappears? | 19:26 |
* creiht promises to not stomp on any more of redbo's commits :) | 19:27 | |
exlt | mtaylor: so it must be swift.egg-info/SOURCES.txt that pristine-tar is using to create the orig.tar.gz - since it was being ignored previously, all I was getting was bin/, swift/, and setup.py basically | 19:28 |
creiht | jdarcy: on a container read, it will return the first response that it gets... there currently is no collation | 19:29 |
*** bretpiatt has quit IRC | 19:30 | |
jdarcy | creiht: OK, but container-content-list placement is completely independent of object placement, right? | 19:30 |
creiht | there is no guarantee of order, if that is what you are asking | 19:30 |
redbo | they're completely independent | 19:31 |
creiht | and that too :) | 19:31 |
jdarcy | Context: a colleague is trying to claim that object placement based on lexicographic order is a must-have. I'm claiming that it's somewhere between "don't give a $#@!" and outright incompatible with high scale. | 19:32 |
creiht | so yeah they are completly independant of one another | 19:34 |
*** cory_ has joined #openstack | 19:34 | |
jdarcy | The use cases where concentration of similarly-named objects onto a single server would be anything but destructive seem vanishingly rare to me. | 19:34 |
creiht | jdarcy: and I would agree with you :) | 19:34 |
jdarcy | Maybe in some Hadoop cases, but that's pretty contrived at best. | 19:34 |
creiht | or if you want that, you could just change the hashing algorithm to the built in python hash :) | 19:34 |
redbo | I guess cassandra makes the in-order distribution mode work. It would hurt with how we do things. | 19:35 |
creiht | and see how things go | 19:35 |
exlt | mtaylor: there is far less extraneous stuff in the diff.gz, but there is still some: http://paste.debian.net/80929/ | 19:35 |
creiht | The problem with that type of scheme in out situation is that you don't get a very even distribution of data accross drives | 19:35 |
jdarcy | From what I can tell, Cassandra's OPP is fraught with peril and a source of many headaches for the devs. But you guys would know that better than I. | 19:35 |
jdarcy | Not to say it isn't valuable if used *very carefully*, just that it's troublesome. | 19:36 |
*** clayg_ is now known as clayg | 19:36 | |
creiht | agreed | 19:36 |
creiht | we strive for simplicity if possible | 19:37 |
jonesy | jdarcy: fwiw, I have experience in 2 env's where the whole word-based sharding thing was put into production. In one it caused hotspots, and in the other it led to amazing amounts of complexity in the data layer. | 19:37 |
jdarcy | jonesy: Yeah, that's my experience too. Which env's, if I may ask? | 19:38 |
jonesy | one I can't talk about. The other was addthis.com | 19:38 |
jdarcy | Good to know. Thanks. | 19:38 |
jonesy | note: addthis.com no longer architects things that way (jdarcy) | 19:38 |
mtaylor | exlt: interesting | 19:39 |
mtaylor | exlt: so, things in that diff are things that are in the tree but not in the tar.gz | 19:39 |
exlt | mtaylor: looking at the details, setup.py/cfg and swift.egg-info/PKG-INFO are just 1.0.2 versions listed in there | 19:39 |
mtaylor | exlt: ah. hrm. well that's because the tree has been advanced... so that would make sense | 19:40 |
exlt | right | 19:40 |
mtaylor | the others are things we should figure out how to make sure hit the tar.gz | 19:40 |
exlt | the test/* are all the full files - not in the orig.tar.gz | 19:40 |
mtaylor | and then next time we cut a tar.gz have the goal be that this diff is non-existant | 19:40 |
mtaylor | yup. we should probably include them :) | 19:41 |
exlt | well, the diff.gz should really be only debian/ ;) | 19:41 |
mtaylor | it's a feature ... to facilitate building with your changes to the trunk being kept in vc... | 19:42 |
mtaylor | exlt: OH - there's another bug... | 19:42 |
mtaylor | we need to add a debian/source/format file so that this will go into the quilt diff instead | 19:42 |
mtaylor | exlt: changes have hit trunk, btw | 19:43 |
mtaylor | exlt: at the moment the feature is pointing out to us that we've got crap in the tree that we're not shipping out in the tarball and should fix in the "upstream" tarball :) | 19:43 |
redbo | creiht: I think we could get a pretty even distribution, but the amount of data you have to redistribute every time you add or remove a node is pretty prohibitive. | 19:44 |
creiht | that is a good point | 19:44 |
creiht | and it seems like it would add a lot of complexity | 19:44 |
*** joshuamckenty has quit IRC | 19:45 | |
exlt | mtaylor: I'm working my way through the package build issues - just haven't hit format, nor watch file, etc | 19:48 |
*** aliguori has joined #openstack | 19:48 | |
mtaylor | exlt: :) | 19:49 |
mtaylor | exlt: I thought I'd already fixed the watch file. blast | 19:50 |
exlt | speaking of tarballs.. I see the download link has 'em now - sweet | 19:50 |
mtaylor | heck yes | 19:50 |
exlt | you have a watch file already? | 19:51 |
mtaylor | I thought I did. but I don't - I _do_ have a watch file for other projects already for getting tarballs from launchpad | 19:51 |
exlt | letting uscan pull the tarball makes some of the issues I was having go away | 19:51 |
*** copilot has joined #openstack | 19:52 | |
mtaylor | exlt: heh. I still use bzr bd even with uscan enabled... but I'm weird lke that | 19:52 |
*** silassewell has joined #openstack | 19:52 | |
*** wylie has joined #openstack | 19:52 | |
exlt | mtaylor: bzr bd tries uscan first, I think | 19:53 |
*** ChanServ sets mode: +v antonym | 19:55 | |
mtaylor | exlt: ok. I've got MANIFEST.in fixed so that missing files wind up in the tarball properly | 19:57 |
mtaylor | exlt: 1.0.2 shouldn't have a diff | 19:57 |
*** sirp3 is now known as s1rp | 19:58 | |
mtaylor | exlt: lp:~mordred/swift/debian-source-format | 19:59 |
creiht | %quote rax | 20:00 |
mtaylor | exlt: source format, MANFEST.in and watch | 20:00 |
creiht | heh... mischan | 20:00 |
*** jeremydei has joined #openstack | 20:01 | |
*** tobym has quit IRC | 20:03 | |
* mtaylor apologizes for the ugliness if you're merging from trunk right now re: .egginfo dirs... shouldn't happen again. | 20:06 | |
aliguori | Hi, congrats on the announcement of openstack! I had one question that I couldn't seem to find an answer to, what are the target hypervisors? Is the goal to support heterogeneous environments or to support just one hypervisor? I see libvirt and xen-api are used so I'm expecting KVM and Xen are supported, what about other things supported by libvirt (like ESX, VBox, etc.)? | 20:12 |
mtaylor | aliguori: I could be completely wrong (so don't quote me) but I believe the eventual idea is to be able to support many hypervisors | 20:13 |
*** kuadrosx has left #openstack | 20:13 | |
*** xtoddx has joined #openstack | 20:15 | |
aliguori | mtaylor, thanks, do you happen to know what works today? | 20:17 |
*** blakeyeager has quit IRC | 20:17 | |
lbieber | mtaylor: you are correct, we want to support multiple hypervisors | 20:17 |
*** thom has joined #openstack | 20:17 | |
*** eleanor has quit IRC | 20:17 | |
lbieber | initially nova supports kvm with work on support for xen coming very quickly | 20:18 |
*** enmand has quit IRC | 20:18 | |
aliguori | excellent | 20:18 |
aliguori | kvm support was the bit I was most interested in :-) | 20:19 |
*** Predominant has quit IRC | 20:22 | |
*** toddmorey has quit IRC | 20:25 | |
*** bpiatt has quit IRC | 20:26 | |
*** miclorb has joined #openstack | 20:26 | |
*** pvo has quit IRC | 20:27 | |
*** silassewell has quit IRC | 20:27 | |
*** pvo has joined #openstack | 20:27 | |
*** justinsheehy has quit IRC | 20:28 | |
*** tobym has joined #openstack | 20:29 | |
*** Predominant has joined #openstack | 20:31 | |
*** pvo has quit IRC | 20:32 | |
*** bithooki has joined #openstack | 20:32 | |
exlt | ugh.. bzr merge hosed my branch.. | 20:32 |
exlt | well, time for food | 20:33 |
*** pvo has joined #openstack | 20:34 | |
*** bretpiatt has joined #openstack | 20:37 | |
justinsb | For Nova/Compute, I'm struggling to launch instances (https://bugs.launchpad.net/nova/+bug/607355). Why do we need kernel images/ramdisks at all? Is this just a throwback to the days before hardware virtualization support? | 20:38 |
*** joshuamckenty has joined #openstack | 20:46 | |
*** JordanRinke has joined #openstack | 20:47 | |
JordanRinke | So, quick question... trying to build nova off the trunk, issue with the redis-server - anyone have a quick guide on how to replace the 1.2v from the repo with the 2.0rc2 from the site? I am a linux newb :( | 20:48 |
pvo | a linux newb trying to setup a cloud cluster! shoot for the stars man! | 20:49 |
bretpiatt | JordanRinke, what isn't working with 1.2v from trunk? | 20:49 |
creiht | haha | 20:49 |
iRTermite | pvo: only way to REALLY learn | 20:49 |
iRTermite | ;) | 20:49 |
JordanRinke | pvo: haha well, I have used ESX/RHEL but I have never used Ubuntu which this seems like it would be easiest to poke at with | 20:49 |
pvo | iRTermite: totally agree. That was meant as encouragement | 20:49 |
johnblue | nothing like getting your hands dirty. | 20:49 |
iRTermite | pvo: I felt what you were saying; just elaborating on it. | 20:50 |
JordanRinke | bretpiatt: When I attempt to run the make it is saying there is a dep on redis-server >=2.0.0rc1 from the trunk build I pulled down on | 20:50 |
JordanRinke | dpkg-checkbuilddeps: Unmet build dependencies: redis-server (>= 2:2.0.0~rc1) | 20:51 |
JordanRinke | dpkg-buildpackage: warning: Build dependencies/conflicts unsatisfied; aborting. | 20:51 |
*** oz` has quit IRC | 20:51 | |
*** joshuamckenty has quit IRC | 20:52 | |
creiht | mtaylor: current trunk has a failing unit test... how did that get past hudson? :) | 20:52 |
jdarcy | Resolving dependencies on RHEL6 was more of a pain than I expected. I'll write up some notes tonight. | 20:52 |
justinsb | bretpiatt: Nova uses redis hashes pretty extensively. I think those were only introduced in 2.0 | 20:53 |
JordanRinke | jdarcy: Would love to see that when done | 20:53 |
creiht | jdarcy: cool thanks | 20:53 |
jdarcy | Where should I send/post when I'm done? | 20:53 |
* creiht tries another checkout of code for unit tests | 20:53 | |
*** bretpiatt has quit IRC | 20:54 | |
*** joshuamckenty has joined #openstack | 20:55 | |
*** bretpiatt has joined #openstack | 20:55 | |
creiht | jdarcy: good question :) | 20:55 |
mtaylor | creiht: interesting... I will investigate :) | 20:55 |
Twitch | wiki.openstack.org? | 20:56 |
redbo | it probably doesn't have PATH_TO_TEST_XFS set | 20:56 |
jdarcy | I'll give that a shot, Twitch. | 20:56 |
jdarcy | Thanks for all the help, folks. Heading home. | 20:56 |
*** jdarcy has quit IRC | 20:56 | |
creiht | mtaylor: redbo has a good point... hudson isn't going to see it since it doesn't have the env var set | 20:57 |
mtaylor | creiht: AH | 20:58 |
redbo | that particular test should be taken outside and shot | 20:58 |
mtaylor | well... the hudson box is on ext3 - we should set up a build slave or something that _is_ running on xfs so that we can test that | 20:58 |
mtaylor | or - can we just make an XFS volume and set the env var to point at it? | 20:59 |
*** Twitch has quit IRC | 20:59 | |
exlt | I would love a bzr rebase ;) | 21:00 |
exlt | think I'm going to create a diff and then pull a fresh branch | 21:00 |
creiht | mtaylor: you can run them on ext3 if you enable xattrs in /etc/fstab | 21:00 |
*** omar has joined #openstack | 21:01 | |
*** justinsheehy has joined #openstack | 21:02 | |
mtaylor | exlt: there is a bzr rebase plugin ... although it does something different from what the git command odes | 21:02 |
holoway | mtaylor: you can always use a loopback device | 21:02 |
holoway | mtaylor: which will be great for testing | 21:02 |
creiht | yeah that too | 21:02 |
mtaylor | exlt: you can alternately do "bzr uncommit -r<last revision you want> ; bzr shelve ; bzr pull ; bzr unshelve ; bzr commit" | 21:02 |
holoway | mtaylor: check out the chef cookbooks for the SAIO for the raw commands | 21:02 |
mtaylor | holoway: got a link for the chef cookbooks? | 21:03 |
holoway | mtaylor: yep! | 21:03 |
* creiht can't vouch for weather or not the chef script still works | 21:03 | |
creiht | but it will have the loopback setup in the script | 21:03 |
holoway | mtaylor: http://c0020195.cdn1.cloudfiles.rackspacecloud.com/openstack-cookbooks.tar.gz | 21:03 |
mtaylor | holoway: sweet | 21:04 |
holoway | the install script is at http://wiki.openstack.org/GettingTheCode | 21:04 |
mtaylor | does one still have to be root to mount loopback devices? | 21:04 |
holoway | mtaylor: I think so | 21:05 |
holoway | mtaylor: you might be able to play tricks with permissions on the loopback device | 21:05 |
holoway | mtaylor: ah, just use 'users' in the mount options | 21:05 |
holoway | mtaylor: you would have an fstab entry that includes 'users', but that does noauto | 21:06 |
holoway | mtaylor: (which is required anyway, since you can't have it auto-mount at boot regardless) | 21:06 |
holoway | mtaylor: you can do that with normal filesystems as well | 21:06 |
_0x44 | exlt: If you use a mac you can bzr diff | pbcopy && bzr branch new_branch && cd to_the_new_branch && pbpaste | patch | 21:06 |
exlt | _0x44: that's very close to what I just did | 21:07 |
mtaylor | exlt: play with uncommit and shelve some time | 21:07 |
* exlt is a linux kind of guy | 21:07 | |
_0x44 | shelve doesn't work across branches. | 21:07 |
_0x44 | So shelve isn't very helpful for my workflow | 21:07 |
mtaylor | fair enough | 21:07 |
exlt | mtaylor: that's what I was going to do before going all manual on it's a__ | 21:07 |
*** oz` has joined #openstack | 21:10 | |
*** Twitch has joined #openstack | 21:11 | |
exlt | eww.. uncommit -r doesn't understand commit-parent_merge-commit-another_parent-merge, etc.. | 21:12 |
exlt | goin' manual ;) | 21:12 |
scotticus | switchin to guns | 21:14 |
mtaylor | creiht: yay! I've got a setup on hudson that can test the XFS now | 21:15 |
*** justinsheehy has quit IRC | 21:15 | |
creiht | nice | 21:15 |
mtaylor | it's not fancy and should be improved - but should work for now | 21:16 |
mtaylor | creiht: next merge should fail unless it's one fixing the xfs bug :) | 21:16 |
*** jtdowney has joined #openstack | 21:16 | |
bytee | mtaylor: howdy ho! | 21:17 |
*** bretpiatt has quit IRC | 21:18 | |
mtaylor | bytee: how're things? | 21:18 |
bytee | mtaylor: good, good. you? | 21:18 |
mtaylor | bytee: excellent | 21:19 |
bytee | mtaylor: congrats with all this open stack goodnewss. it does look exciting alright | 21:19 |
mtaylor | bytee: I seem to be at the beach - although you might not know it by my activity here today | 21:19 |
mtaylor | bytee: thanks! | 21:19 |
mtaylor | bytee: you at oscon? | 21:19 |
*** blakeyeager has joined #openstack | 21:19 | |
bytee | mtaylor: yah. enjoy the beach, sun is better than the laptop | 21:19 |
mtaylor | bytee: well... as soon as the sun actually comes out, I will enjoy it | 21:20 |
*** bretpiatt has joined #openstack | 21:23 | |
joshuamckenty | someone was asking about blueprints against nova earlier, yeah? | 21:24 |
joshuamckenty | re: redis2.0 - redis 1.2 doesn't have hashes | 21:25 |
mtaylor | creiht: yay! your fix seems to have worked | 21:28 |
mtaylor | or rather: | 21:29 |
mtaylor | redbo: yay! your fix seems to have worked | 21:29 |
*** dendro-afk is now known as dendrobates | 21:29 | |
dendrobates | joshuamckenty: We are getting all the component leads to create blueprints any chance NASA can do it for their features? | 21:30 |
joshuamckenty | heck yeah | 21:30 |
*** nickmbailey has left #openstack | 21:31 | |
joshuamckenty | I'll see if folks can get their items from this sprint written up ASAP (started today) | 21:31 |
*** justinsheehy has joined #openstack | 21:32 | |
holoway | what would you guys think about flagging some of the open issues as low-hanging fruit for new contributors? | 21:37 |
joshuamckenty | yes! | 21:37 |
joshuamckenty | we've got a long TODO list of easy stuff | 21:37 |
holoway | for example, I would love to fix a few bugs while my daughter sleeps late at night | 21:37 |
joshuamckenty | we should get those bugified in lp | 21:37 |
*** rando has quit IRC | 21:39 | |
dendrobates | I have some stuff that I want to push after I get some more sleep to ensure it's not insane. | 21:40 |
joshuamckenty | We were only halfway through the transition from internal tools to github when we switched gears and started moving to launchpad | 21:40 |
joshuamckenty | push early and often; you don't have to request a merge yet :) | 21:41 |
joshuamckenty | mtaylor: who can delete this merge proposal? https://code.launchpad.net/~vishvananda/nova/dhcplease-flags/+merge/30056 | 21:41 |
joshuamckenty | is that only vish, or only jesse, or either? | 21:41 |
soren | joshuamckenty: I can. | 21:42 |
joshuamckenty | cause they're both on planes right now | 21:42 |
soren | joshuamckenty: Shall I? | 21:42 |
joshuamckenty | yes please, I think it was landed to trunk separately | 21:42 |
joshuamckenty | http://twitter.com/jasonwhaley/statuses/18924766126 | 21:42 |
soren | (I can since it's a proposal to merge into one of my branches) | 21:42 |
joshuamckenty | ah, gotcha | 21:42 |
soren | Deleted. | 21:42 |
soren | joshuamckenty: Thanks. | 21:42 |
joshuamckenty | thank you | 21:43 |
joshuamckenty | soren: do we need to resubmit https://code.launchpad.net/~vishvananda/nova/twisted-volume/+merge/30153 as a merge against trunk? | 21:44 |
joshuamckenty | I don't think vish meant to propose it as a merge against his own branch | 21:44 |
JordanRinke | sweet, got Redis setup finally | 21:45 |
joshuamckenty | nice | 21:45 |
joshuamckenty | There's a good deb package of 2.0 built by the rdio guys | 21:45 |
joshuamckenty | that's what we use | 21:45 |
joshuamckenty | not sure where it is atm, though | 21:46 |
joshuamckenty | soren: got time for one more review? https://code.launchpad.net/~jaypipes/nova/bug606340/+merge/30133 | 21:46 |
JordanRinke | yeah | 21:46 |
JordanRinke | I had to find it in the experimental repo, and get that setup | 21:46 |
joshuamckenty | it's jaypipes mop-up to make the network tests more readable. | 21:46 |
JordanRinke | once I knew it was there / had access to update etc it was cake | 21:46 |
joshuamckenty | ah, yeah | 21:47 |
joshuamckenty | you can build it from source pretty easily, as well | 21:47 |
JordanRinke | I compiled from source and rip/replaced it initially | 21:47 |
joshuamckenty | redis builds really cleanly | 21:47 |
JordanRinke | but figured that was a smoother way to do it | 21:47 |
*** gondoi has quit IRC | 21:49 | |
joshuamckenty | mtaylor: / soren: are we planning on building deb packages from trunk via hudson as well? | 21:49 |
JordanRinke | so I have a quick text on how to update to redis 2.0 rc1, where should I post it for anyone else that is trying to quick start and hitting that compile error | 21:52 |
joshuamckenty | all: we know that the use of gflags has gotten insane. Now is a great time to be proposing a more better way. | 21:52 |
joshuamckenty | JordanRinke: add it to the wiki.openstack.org site, somewhere near to the install instructions? | 21:53 |
JordanRinke | will do | 21:53 |
joshuamckenty | awesome | 21:53 |
*** oz` has quit IRC | 21:54 | |
*** Twitch has quit IRC | 21:54 | |
*** Thweetch has joined #openstack | 21:55 | |
Thweetch | You get any sleep Josh? | 21:56 |
jk0 | hm? | 21:56 |
jk0 | oh, other Josh :) | 21:56 |
gholt | creiht: Do you know what this nonsense about swift.egg-info is or how we/I should fix it? [in swift] | 21:57 |
*** jbryce has joined #openstack | 21:58 | |
*** jtdowney_ has joined #openstack | 21:59 | |
*** Predominant has quit IRC | 21:59 | |
joshuamckenty | not much | 22:01 |
gholt | creiht: Well, I chowned some stuff and repulled/updated/discarded "conflicts" and that seems okay. But now setup.py develop doesn't work searching for WebOb=0.9.8.post1 | 22:01 |
*** pvo has left #openstack | 22:01 | |
*** pvo has joined #openstack | 22:01 | |
*** pvo has left #openstack | 22:01 | |
*** jtdowney has quit IRC | 22:01 | |
*** viewlogic has joined #openstack | 22:01 | |
*** bretpiatt has left #openstack | 22:03 | |
*** bretpiatt has joined #openstack | 22:03 | |
*** ghchinoy has joined #openstack | 22:05 | |
ghchinoy | Hi all | 22:05 |
ghchinoy | Is openstack a set of addons to eucalyptus? | 22:06 |
*** toddmorey has joined #openstack | 22:06 | |
patcito | ghchinoy, no, it another beast | 22:06 |
ghchinoy | patcito: but it's similar? a management set for compute and storage nodes that follow ec2 and s3? | 22:06 |
patcito | ghchinoy, indeed | 22:07 |
ghchinoy | Aha, ok | 22:07 |
patcito | but it's not compatible | 22:07 |
patcito | API wise | 22:07 |
creiht | gholt: I need to talk to mtaylor about the swift.egg-info stuff | 22:07 |
creiht | ideally we don't want that in the source | 22:07 |
justinsb | How can we see the merge discussion for a revision? For example, this one: http://bazaar.launchpad.net/~hudson-openstack/nova/trunk/revision/146 I think it's introduced some bugs, but I'd like to read the merge request first. | 22:07 |
ghchinoy | Is eucalyptus compatible with aws or is it also not compbatible? | 22:08 |
patcito | I think it's supposed to be, but eucalyptus doesn't scala well | 22:08 |
patcito | scale* | 22:08 |
ghchinoy | Can swift interoperate with google storage, or is that on the roadmap? | 22:09 |
ghchinoy | patcito: oh ho, i didn't know that. | 22:09 |
joshuamckenty | ghchinoy: openstack is currently ec2/s3 compatible, but I don't think that's a core feature going forward | 22:09 |
ghchinoy | joshuamckenty: ok, cool | 22:09 |
ghchinoy | are there goals/principles for the project somewhere? | 22:09 |
patcito | cool, joshuamckenty where can I read about that? | 22:09 |
joshuamckenty | openstack.org | 22:09 |
creiht | swift isn't s3 compatible api wise... it is similar | 22:09 |
patcito | I mean s3/ec2 compatibility | 22:10 |
joshuamckenty | creiht: I imagine it'll be s3-compatible in a couple of days | 22:10 |
creiht | hehe | 22:10 |
*** urandom has left #openstack | 22:10 | |
redbo | we can do that in middleware :P | 22:10 |
creiht | hah | 22:10 |
joshuamckenty | that's what she said | 22:10 |
ghchinoy | lolz | 22:10 |
*** urandom has joined #openstack | 22:10 | |
*** dizzl_ has quit IRC | 22:11 | |
ghchinoy | joshuamckenty: on the wiki? | 22:11 |
creiht | mtaylor: when you get a min, we need to talk about swift.egg-info | 22:11 |
joshuamckenty | ghchinoy: goals and principles are discussed in a lot of the press from last night, and also in: | 22:12 |
joshuamckenty | http://wiki.openstack.org/Open | 22:12 |
*** miclorb has quit IRC | 22:13 | |
ghchinoy | Ok, maybe I'm asking more about non-functional requirements like scalablity, etc | 22:13 |
ghchinoy | Not so much definitions of openness | 22:13 |
joshuamckenty | the goal is 1 million host machines, | 22:13 |
joshuamckenty | with 60 million virtual servers | 22:13 |
joshuamckenty | Not sure where that ended up in writing | 22:14 |
*** joshuamckenty has quit IRC | 22:15 | |
ghchinoy | /BasicDesignTenets has some | 22:15 |
*** toddmorey has quit IRC | 22:17 | |
*** joshuamckenty has joined #openstack | 22:18 | |
ghchinoy | /BasicDesignTenets has some, joshuamckenty, seems like it needs to be filled out more. | 22:19 |
*** jakedahn has joined #openstack | 22:19 | |
joshuamckenty | age | 22:19 |
joshuamckenty | agree | 22:19 |
soren | joshuamckenty: Well... It won't resubmit itself :D | 22:20 |
ghchinoy | Cool, I'll be back later to ask more dumb questions :) thanks joshuamckenty, patcito | 22:20 |
joshuamckenty | np | 22:21 |
dendrobates | aliguori: we really missed you at the summit | 22:21 |
aliguori | dendrobates, yeah, i didn't realize how soon it was | 22:21 |
soren | joshuamckenty: Both you and vish approved it. Just go ahead and set a commit message and set it to approved. | 22:21 |
aliguori | dendrobates, sounds like it was productive at least :-) | 22:21 |
joshuamckenty | ah, damn commit messages. K. | 22:22 |
*** Predominant has joined #openstack | 22:23 | |
*** joshuamckenty has quit IRC | 22:23 | |
soren | joshuamckenty: Yeah, in most cases it would probably just be the same as the text you put in the merge proposal. | 22:24 |
soren | Oh, he buggered off. | 22:24 |
*** devcamcar has joined #openstack | 22:24 | |
*** joshuamckenty has joined #openstack | 22:24 | |
*** adrian_otto has quit IRC | 22:24 | |
*** Manganeez has quit IRC | 22:25 | |
*** Thweetch has quit IRC | 22:26 | |
*** KnuckleSangwich has quit IRC | 22:28 | |
*** devcamcar has quit IRC | 22:28 | |
soren | 22:24 <+soren> joshuamckenty: Yeah, in most cases it would probably just be the same as the text you put in the merge proposal. | 22:28 |
joshuamckenty | ah, k | 22:28 |
*** devcamcar has joined #openstack | 22:29 | |
soren | joshuamckenty: Although having them separate lets you have a less formal style in the merge proposal (like if you really don't expect it to be merged in its current form, but are just looking for feedback). | 22:29 |
*** aspx has joined #openstack | 22:31 | |
*** devcamcar has quit IRC | 22:31 | |
aspx | how does the storage work e.g. is it similar to applogic's (3tera) | 22:31 |
*** devcamcar has joined #openstack | 22:32 | |
creiht | aspx: I'm not familiar with 3tera... is there something specific you are wanting to know? | 22:32 |
*** jakedahn has quit IRC | 22:33 | |
*** jakedahn_ has joined #openstack | 22:33 | |
aspx | 3tera uses local storage on each node that is raided across different nodes how is openstack's storage configured | 22:33 |
creiht | swift stores 3 replicas of each object distributed accross the cluster | 22:34 |
creiht | each replica being isolated depending on configuration | 22:34 |
aspx | so does it use local storage? | 22:35 |
creiht | the devices can be raided if you like, but we recommend against that | 22:35 |
creiht | In what context do you define local storage? | 22:35 |
creiht | In context to compute nodes? | 22:36 |
aspx | not san but inside the physical server itself e.g sas drive | 22:36 |
creiht | yes | 22:36 |
creiht | it is designed to be run on commodity hardware | 22:36 |
creiht | this is just blob storage though | 22:37 |
creiht | not like a filesystem like EBS, that will be a separate product | 22:37 |
aspx | awesome thank you which of the websites would be best to look at tech stats | 22:37 |
aspx | opps specs not stats | 22:37 |
creiht | aspx: well we have preliminary docs available at swift.openstack.org, but they may not have what you are looking for yet | 22:38 |
creiht | the code is at launchpad.net/swift | 22:38 |
creiht | until I get docs written for that, are there any specifics you would like to know? | 22:38 |
aspx | ok thanks for your help :-) | 22:38 |
creiht | Which may actually help me write those docs :) | 22:38 |
*** joshuamckenty has quit IRC | 22:38 | |
aspx | just the storage config we want local storage used with copies across different nodes | 22:39 |
creiht | does anyone know if OSCON presentations are on video? | 22:39 |
aspx | cause san to expensive | 22:39 |
*** rando has joined #openstack | 22:39 | |
creiht | aspx: that is by default how we usually set it up | 22:39 |
creiht | the docs have a page describing how to set a single dev vm which has the services running on one box | 22:40 |
iRTermite | creiht: the link is on the front page | 22:40 |
aspx | ok will investigate and come back with more questions | 22:40 |
creiht | cool | 22:40 |
iRTermite | don't think all are on video though | 22:40 |
creiht | iRTermite: thanks... one of our coworkers is giving a talk on the swift architecture there | 22:41 |
iRTermite | nice! | 22:41 |
*** aspx has quit IRC | 22:41 | |
creiht | I'll watch tomorrow to see if they upload video of it | 22:41 |
iRTermite | I'm certain (just like other cons) one would have to request for the session to be recorded? | 22:41 |
iRTermite | cool, I'll check that out tomorrow as well | 22:43 |
*** devcamcar has quit IRC | 22:43 | |
creiht | oh I guess it is on Wed.. http://www.oscon.com/oscon2010/public/schedule/detail/15634 | 22:44 |
*** devcamcar has joined #openstack | 22:44 | |
*** joshuamckenty has joined #openstack | 22:46 | |
*** devcamcar has quit IRC | 22:46 | |
*** miclorb_ has joined #openstack | 22:46 | |
*** devcamcar has joined #openstack | 22:47 | |
*** devcamcar_ has joined #openstack | 22:47 | |
*** devcamcar_ has left #openstack | 22:48 | |
*** devcamcar_ has joined #openstack | 22:48 | |
devcamcar_ | hey all, what is the ppa for nova? can't find it on the wiki | 22:49 |
*** urandom has quit IRC | 22:50 | |
soren | devcamcar_: https://launchpad.net/~nova-core/+archive/ppa | 22:51 |
*** devcamcar_ has quit IRC | 22:51 | |
soren | gah | 22:51 |
*** nettok has joined #openstack | 22:51 | |
soren | So some of you must know this.. If I subscribe to openstack bug mail, do I also get bug mail for its subprojects? | 22:52 |
*** JordanRinke has quit IRC | 22:52 | |
bretpiatt | creiht, more OSCON info @oscon: Speaker slides are being posted here as we get them: http://ping.fm/vxN3A #oscon | 22:53 |
creiht | cool | 22:53 |
*** Slidey has joined #openstack | 22:53 | |
*** uvirtbot has joined #openstack | 22:55 | |
*** Mr_T has joined #openstack | 22:56 | |
soren | bug #607355 | 22:58 |
uvirtbot | Launchpad bug 607355 in nova "Problems launching instances" [Undecided,New] https://launchpad.net/bugs/607355 | 22:58 |
soren | \o/ | 22:58 |
creiht | btw... if anyone is at OSCON: http://www.oscon.com/oscon2010/public/schedule/detail/15754 | 22:58 |
justinsb | soren: The new twisted-based objectstore doesn | 23:00 |
*** devcamcar has quit IRC | 23:00 | |
*** urandom has joined #openstack | 23:01 | |
*** devcamcar has joined #openstack | 23:01 | |
soren | It doesn? | 23:02 |
soren | DArn it. | 23:02 |
justinsb | soren: Sorry, the new twisted-based objectstore doesn't seem to be flushing its responses correctly for me. e.g. a curl on curl http://127.0.0.1:3333/_images will return the 1st half of the data and then hang. If I kill the daemon, I do get the rest of the data. Any ideas? The same problem seems to affect the compute API | 23:02 |
soren | I was hoping it woul | 23:02 |
soren | :) | 23:02 |
soren | :( | 23:02 |
justinsb | soren: Actually, the nova-api... I get IncompleteRead errors there when doing e.g. euca-describe-images | 23:03 |
soren | Interesting. I'll look into it. | 23:04 |
*** g-ram has joined #openstack | 23:04 | |
g-ram | evening all | 23:04 |
notmyname | howdy g-ram | 23:06 |
justinsb | soren: May have just found it... I think it's a missing call to request.finish() in render_GET in ImageResource. I suddenly noticed it was JSON, not XML so I was looking in the wrong place :-) | 23:06 |
soren | justinsb: That would do it. | 23:06 |
* soren looks | 23:06 | |
*** joshuamckenty has quit IRC | 23:06 | |
soren | justinsb: Oh, you're absolutely right. | 23:07 |
justinsb | soren: Yup, that fixed it. Put it as another comment here: https://code.launchpad.net/~soren/nova/twisted-web-s3-server/+merge/30206 | 23:07 |
justinsb | soren: A couple more comments there as I was trying to track this one down!! | 23:08 |
soren | justinsb: Add a call to request.finish() just before return server.NOT_DONE_YET | 23:08 |
soren | justinsb: That merge proposal was already approved and merged. I think it's better to file a bug in such a case. There's no way to undo the merge (per se). | 23:09 |
*** devcamcar has quit IRC | 23:09 | |
justinsb | soren: OK... I'll pop the comments into a bug | 23:09 |
*** devcamcar has joined #openstack | 23:10 | |
soren | justinsb: Great stuff. Thanks! | 23:10 |
*** jtdowney_ is now known as jtdowney | 23:12 | |
*** pvo_ has joined #openstack | 23:12 | |
justinsb | bug #607501 | 23:13 |
uvirtbot | Launchpad bug 607501 in nova "Issues in commit 146 (twisted)" [Undecided,New] https://launchpad.net/bugs/607501 | 23:13 |
justinsb | uvirtbot is cool! | 23:13 |
uvirtbot | justinsb: Error: "is" is not a valid command. | 23:13 |
ghchinoy | I'm back with more stupid questions... is openstack designed to run on a particular linux distro or is it akin to xen, as a baremetal install? | 23:14 |
soren | justinsb: Had you given it a minute, it would have announce the bug on its own. | 23:14 |
soren | justinsb: That's how cool it is :) | 23:14 |
justinsb | soren: very cool, even if it can't take a compliment :-) | 23:14 |
*** pvo_ is now known as pvo | 23:15 | |
*** devcamcar has quit IRC | 23:15 | |
*** devcamcar has joined #openstack | 23:16 | |
soren | justinsb: :) | 23:17 |
*** blakeyeager has quit IRC | 23:21 | |
uvirtbot | New bug: #607501 in nova "Issues in commit 146 (twisted)" [Undecided,New] https://launchpad.net/bugs/607501 | 23:21 |
*** nettok has quit IRC | 23:21 | |
soren | justinsb: There it goes ^ | 23:22 |
soren | justinsb: The delay is due to the fact that Launchpad doesn't send out e-mail until there's been 8 minutes of inactivity on a bug (so that people who change multiple things on a bug don't cause a bajillion e-mails to be sent to a bajillion people). | 23:23 |
*** devcamcar has quit IRC | 23:23 | |
*** rando has quit IRC | 23:24 | |
*** devcamcar has joined #openstack | 23:25 | |
*** g-ram has quit IRC | 23:25 | |
*** lbieber has quit IRC | 23:27 | |
*** spackest has joined #openstack | 23:29 | |
soren | In case anyone anyone is wondering, yes, subscribing to bugmail for openstack automatically subscribes you to nova and swift bugmail. | 23:30 |
*** g-ram has joined #openstack | 23:32 | |
*** devcamcar has quit IRC | 23:32 | |
*** devcamcar has joined #openstack | 23:33 | |
g-ram | so a question from someone who's pretty clueless at this point -- is openstack a distribution, or. . . how exactly does the installation work? | 23:36 |
soren | justinsb: Are you up for fixing bug #607501 yourself? Just shout if you need bzr/lp pointers for something. | 23:37 |
uvirtbot | Launchpad bug 607501 in nova "Issues in commit 146 (twisted)" [Undecided,New] https://launchpad.net/bugs/607501 | 23:37 |
pvo | g-ram it isn't a distribution but a software installation package | 23:38 |
g-ram | got it | 23:38 |
g-ram | so all of the nodes in a cluster would need the same packages installed I'm assuming? | 23:38 |
pvo | they might do different things, so it would depend on how you built your cluster | 23:39 |
g-ram | (also, I'm assuming at this point that openstack is distribution agnostic -- that is, it should work on any of the mainstream linux distros?) | 23:39 |
pvo | g-ram: it should but its been developed on ubuntu, iirc | 23:39 |
g-ram | great -- that's helpful; if it works on ubuntu I imagine I shouldn't have much trouble using it on debian | 23:40 |
g-ram | thanks! | 23:40 |
exlt | g-ram: if building a big storage cluster with swift, for instance, filesystem choice is key - somethat not all distros allow unfortunately | 23:40 |
exlt | missed a few characters there.. "something that" | 23:41 |
g-ram | what fs would you recommend? ext3? brtfs? zfs? | 23:41 |
notmyname | g-ram: swift requires xattr support | 23:41 |
g-ram | ok, cool | 23:42 |
exlt | we use XFS with Cloud Files - evaluated most every filesytem | 23:42 |
g-ram | I've never played with xfs but have heard good things | 23:42 |
*** bretpiatt has quit IRC | 23:43 | |
g-ram | also (sorry for so many questions) what's the difference between swift and nova? | 23:44 |
*** devcamcar has quit IRC | 23:44 | |
*** devcamcar has joined #openstack | 23:46 | |
*** matiu has joined #openstack | 23:46 | |
pvo | nova is compute controller software | 23:48 |
pvo | swift is a distributed object store | 23:48 |
pvo | think nova == ec2 | 23:48 |
pvo | and swift == s3 | 23:48 |
g-ram | got it | 23:49 |
g-ram | probably interested in nova then | 23:49 |
pvo | they can be run independently or used together for superpowers combined | 23:49 |
chilts | heh, I was wondering the same ... g-ram, thanks for quizzing :) | 23:49 |
g-ram | I'm interested in building a test cluster with a few quad core xeons to do computational fluid dynamics modeling | 23:49 |
chilts | pvo: so I guess there are also thoughts about equivalent things like SQS, SNS, SimpleDB even? | 23:50 |
pvo | chilts: yes, eventually. These two projects are the core of what we're trying to build | 23:50 |
g-ram | even on a single quad core box CFD models can take a lot of time | 23:50 |
chilts | pvo: sweet ... am liking OpenStack already ... | 23:50 |
chilts | I was gonna start my business on EC2 but actually started using RackspaceCloud instead about 4 weeks ago | 23:51 |
pvo | those other services are accessory to the core offerings | 23:51 |
chilts | I love it already and I think now I definitely made the right choice | 23:51 |
g-ram | chilts: what type of business? | 23:51 |
chilts | yeah true, but also very useful :) | 23:51 |
chilts | g-ram: online applications :) (to be pretty general) | 23:51 |
pvo | chilts: though if you want to start on one of the other services, we'd gladly accept the work. : ) | 23:51 |
chilts | so it's a very good fit | 23:51 |
g-ram | that's fair :) | 23:52 |
chilts | pvo: sounds like a winner ... though I haven't checked out the code yet, what are you guys writing it all in? | 23:52 |
pvo | python | 23:52 |
* chilts did a mock queue using Node.js | 23:52 | |
*** spackest has quit IRC | 23:52 | |
chilts | sweet, am just glad it's not Java :) | 23:52 |
g-ram | ha, java would be terrible for this type of application | 23:52 |
chilts | I'm gonna do a very simple notification system in Node.js too | 23:52 |
chilts | well, Java has it's uses - I just don't use it :D | 23:53 |
notmyname | seen on the internets: "Java is a DSL for turning XML into stack traces" | 23:53 |
chilts | hehe | 23:53 |
*** pvo has quit IRC | 23:58 | |
*** pvo has joined #openstack | 23:58 | |
justinsb | soren: I suppose I can fix bug #607501 as I have it locally. I'll just create a branch now. Sorry for delay - late lunch! | 00:00 |
uvirtbot | Launchpad bug 607501 in nova "Issues in commit 146 (twisted)" [Undecided,New] https://launchpad.net/bugs/607501 | 00:00 |
*** pengwynn has quit IRC | 00:00 | |
*** pvo has quit IRC | 00:03 | |
*** devcamcar has quit IRC | 00:03 | |
*** jbryce has quit IRC | 00:04 | |
*** hunter has quit IRC | 00:06 | |
*** Thweetch has joined #openstack | 00:06 | |
*** loglaunch has joined #openstack | 00:12 | |
*** devcamcar has joined #openstack | 00:13 | |
*** aliguori has quit IRC | 00:16 | |
*** devcamcar has quit IRC | 00:16 | |
soren | justinsb: No worries. Thanks! | 00:17 |
*** devcamcar has joined #openstack | 00:17 | |
*** lbieber has joined #openstack | 00:19 | |
uvirtbot | New bug: #607512 in nova "Authorization not checked on objectstore" [Undecided,New] https://launchpad.net/bugs/607512 | 00:21 |
*** devcamcar has quit IRC | 00:21 | |
*** vish1 has joined #openstack | 00:21 | |
exlt | just for fun :) - http://12.am/openstack/ | 00:21 |
vish1 | wow, lots of people here | 00:22 |
arcane | crazy | 00:24 |
*** dendrobates is now known as dendro-afk | 00:24 | |
exlt | oh, it'll get busier, I hope - go look at http://12.am/cassandra/ ;) | 00:25 |
*** Predominant has quit IRC | 00:26 | |
*** Thweetch has quit IRC | 00:26 | |
driftx | "in production you aren't actually going to pipe shit through it" hah, nice. | 00:27 |
exlt | oh, that is a good one, driftx | 00:28 |
*** devcamcar has joined #openstack | 00:30 | |
*** pdelgallego has quit IRC | 00:31 | |
vish1 | hoyoo devin | 00:31 |
* arcane stretches | 00:31 | |
*** vish1 is now known as vishvananda | 00:32 | |
mtaylor | w00t! I'm second on http://12.am/openstack/ | 00:33 |
mtaylor | exlt: would it be hard for you to add #drizzle ? | 00:33 |
*** Thweetch has joined #openstack | 00:33 | |
g-ram | 12.am -- that's clever | 00:34 |
g-ram | what's the am TLD for again? | 00:35 |
g-ram | guess I could google that as well | 00:35 |
* mtaylor needs that hooked in to his launchpad karma... | 00:36 | |
*** jdmaturen has quit IRC | 00:43 | |
exlt | mtaylor: there won't be any history.. got any irssi logs that go back on #drizzle? | 00:45 |
*** devcamcar has quit IRC | 00:45 | |
*** devcamcar has joined #openstack | 00:47 | |
mtaylor | exlt: I don't, but eday probably does | 00:47 |
mtaylor | eday: | 00:47 |
mtaylor | eday: | 00:47 |
mtaylor | eday: | 00:47 |
exlt | heh | 00:47 |
* mtaylor is pretty sure eday does irc via ssh/screen/irssi | 00:48 | |
exlt | same here - screen ftw | 00:48 |
* mtaylor has been meaning to learn irssi | 00:49 | |
mtaylor | problem is - I spend _soooooooo_ much time in IRC - learning something new, even if for the better, would sort of suck | 00:49 |
mtaylor | of course, this doesn't mean it's a bad idea | 00:49 |
uvirtbot | New bug: #607523 in nova "There should be a separate flag for fake_ldap" [Undecided,New] https://launchpad.net/bugs/607523 | 00:51 |
g-ram | is there a flag for real_ldap? | 00:52 |
g-ram | (sorry, my sense of humor degrades dramatically as I get tired) | 00:52 |
creiht | heh... I was worried that the auth bug earlier was for swift | 00:53 |
creiht | guess not | 00:53 |
*** lbieber has quit IRC | 00:55 | |
creiht | exlt: nice stats :) | 00:55 |
exlt | bleh - point pisg to the log file and cron - 30 seconds | 00:55 |
creiht | hehe | 00:55 |
exlt | but they are fun to look at from time to time | 00:56 |
justinsb | creiht: yes, don't worry, the auth bug is on nova's objectstore. I can't imagine anyone's running nova in production yet, which is why I set the status to public :-) | 00:57 |
creiht | hehe... well even the auth we have in swift is meant for development purposes only | 00:59 |
*** tobym has quit IRC | 00:59 | |
uvirtbot | New bug: #607527 in nova "VPN not created for auto-created project" [Undecided,New] https://launchpad.net/bugs/607527 | 01:05 |
*** johnpur has quit IRC | 01:07 | |
*** vish1 has joined #openstack | 01:07 | |
*** pvo has joined #openstack | 01:11 | |
*** devcamcar has quit IRC | 01:11 | |
*** devcamcar has joined #openstack | 01:11 | |
*** pengwynn has joined #openstack | 01:15 | |
*** jdarcy has joined #openstack | 01:22 | |
*** devcamcar has quit IRC | 01:22 | |
*** g-ram has quit IRC | 01:23 | |
*** devcamcar has joined #openstack | 01:23 | |
*** tupshin has quit IRC | 01:28 | |
notmyname | it's interesting to look at how openstack tweets have changed throughout the day. now, there are a good number in japanese. it's interesting to watch the news follow the sun | 01:31 |
*** joshuamckenty has joined #openstack | 01:37 | |
*** devcamcar has quit IRC | 01:37 | |
vish1 | cool | 01:37 |
*** devcamcar has joined #openstack | 01:38 | |
*** joshuamckenty has quit IRC | 01:38 | |
*** joshuamckenty has joined #openstack | 01:39 | |
*** devcamcar has quit IRC | 01:39 | |
*** devcamcar has joined #openstack | 01:40 | |
*** joshuamckenty has quit IRC | 01:40 | |
*** joshuamckenty has joined #openstack | 01:42 | |
*** devcamcar has quit IRC | 01:42 | |
*** jdarcy has quit IRC | 01:42 | |
*** devcamcar has joined #openstack | 01:43 | |
*** pvo has quit IRC | 01:48 | |
*** devcamcar has quit IRC | 01:48 | |
*** pvo has joined #openstack | 01:49 | |
*** devcamcar has joined #openstack | 01:49 | |
*** sander has joined #openstack | 01:53 | |
*** devcamcar has quit IRC | 01:53 | |
*** pvo has quit IRC | 01:53 | |
*** jtdowney has quit IRC | 01:56 | |
*** joshuamckenty has quit IRC | 01:56 | |
*** joshuamckenty has joined #openstack | 01:56 | |
*** tobym has joined #openstack | 01:58 | |
matiu | Japan new about it before USA :) as it was released USA midnight which is like early morning Japan :P | 01:59 |
creiht | :) | 01:59 |
chilts | yeah, I knew about it last night (or at least, _my_ last night) ... did a blog post about 15 hrs ago :D | 02:00 |
chilts | it was fun reading everything | 02:00 |
chilts | am very excited | 02:00 |
uvirtbot | New bug: #607541 in nova "ObjectStore must throw 404 when bucket does not exist (Commit 146)" [Undecided,New] https://launchpad.net/bugs/607541 | 02:01 |
notmyname | I admit I don't know much about nova, but some of these bugs look very storage focused, not compute | 02:02 |
*** joshuamckenty has quit IRC | 02:02 | |
*** joshuamckenty has joined #openstack | 02:02 | |
*** devcamcar has joined #openstack | 02:02 | |
*** poli has quit IRC | 02:03 | |
*** poli has joined #openstack | 02:03 | |
*** devcamcar has quit IRC | 02:03 | |
*** joshuamckenty has quit IRC | 02:04 | |
*** joshuamckenty has joined #openstack | 02:04 | |
*** joshuamckenty has quit IRC | 02:06 | |
*** abecc has joined #openstack | 02:10 | |
vish1 | notmyname: we have a ghetto version of storage which we need to support until nova-storage integration is complete | 02:10 |
vish1 | er openstack-storage | 02:10 |
notmyname | ah | 02:11 |
vish1 | anyone here who has installed m2crypto and redis2 successfully on mac? | 02:11 |
*** abecc has quit IRC | 02:11 | |
notmyname | I get all excited as I'm reading the bug title. "bug...storage!...that's odd, I thought we tested that...oh, it's nova" | 02:12 |
vish1 | :p | 02:12 |
*** devcamcar has joined #openstack | 02:13 | |
jk0 | hehe | 02:18 |
*** devcamcar has quit IRC | 02:18 | |
*** abecc has joined #openstack | 02:19 | |
*** devcamcar has joined #openstack | 02:19 | |
*** pvo has joined #openstack | 02:21 | |
*** devcamcar has quit IRC | 02:21 | |
*** devcamcar has joined #openstack | 02:22 | |
*** dendro-afk is now known as dendrobates | 02:24 | |
*** devcamcar has quit IRC | 02:24 | |
*** devcamcar has joined #openstack | 02:25 | |
*** jonesy_ has joined #openstack | 02:28 | |
*** devcamcar has quit IRC | 02:28 | |
*** joshuamckenty has joined #openstack | 02:30 | |
*** devcamcar has joined #openstack | 02:30 | |
*** abecc has quit IRC | 02:31 | |
*** joshuamckenty has quit IRC | 02:31 | |
*** devcamcar has quit IRC | 02:31 | |
*** joshuamckenty has joined #openstack | 02:32 | |
*** devcamcar has joined #openstack | 02:33 | |
*** nettok has joined #openstack | 02:41 | |
*** adjohn has joined #openstack | 02:41 | |
*** syah has joined #openstack | 02:44 | |
*** joshuamckenty has quit IRC | 02:44 | |
*** devcamcar has quit IRC | 02:44 | |
*** vr_mex has joined #openstack | 02:44 | |
*** joshuamckenty has joined #openstack | 02:44 | |
*** devcamcar has joined #openstack | 02:45 | |
*** joshuamckenty has quit IRC | 02:45 | |
*** joshuamckenty has joined #openstack | 02:47 | |
*** devcamcar has quit IRC | 02:47 | |
*** vr_mex has quit IRC | 02:47 | |
*** devcamcar has joined #openstack | 02:48 | |
*** joshuamckenty has quit IRC | 02:48 | |
*** jtimberman has joined #openstack | 02:49 | |
*** joshuamckenty has joined #openstack | 02:49 | |
*** abecc has joined #openstack | 02:50 | |
*** joshuamckenty has quit IRC | 02:50 | |
*** joshuamckenty has joined #openstack | 02:52 | |
*** devcamcar has quit IRC | 02:52 | |
*** devcamcar has joined #openstack | 02:53 | |
*** joshuamckenty has quit IRC | 02:53 | |
*** joshuamckenty has joined #openstack | 02:54 | |
*** devcamcar has quit IRC | 02:54 | |
uvirtbot | New bug: #607355 in nova "Problems launching instances" [Undecided,New] https://launchpad.net/bugs/607355 | 02:56 |
*** joshuamckenty has quit IRC | 02:56 | |
*** blakeyeager has joined #openstack | 02:57 | |
*** joshuamckenty has joined #openstack | 02:57 | |
*** blakeyeager has quit IRC | 02:58 | |
*** joshuamckenty has quit IRC | 02:58 | |
*** joshuamckenty has joined #openstack | 02:59 | |
*** blakeyeager has joined #openstack | 03:00 | |
*** joshuamckenty has quit IRC | 03:00 | |
*** pengwynn has quit IRC | 03:02 | |
*** joshuamckenty has joined #openstack | 03:02 | |
*** devcamcar has joined #openstack | 03:04 | |
*** jdmaturen has joined #openstack | 03:11 | |
*** joshuamckenty has quit IRC | 03:11 | |
*** devcamcar has quit IRC | 03:11 | |
pvo | anyone in the cloudcamp talk? | 03:12 |
*** joshuamckenty has joined #openstack | 03:12 | |
*** devcamcar has joined #openstack | 03:12 | |
*** dendrobates is now known as dendro-afk | 03:15 | |
*** dendro-afk is now known as dendrobates | 03:17 | |
*** devcamcar has quit IRC | 03:17 | |
*** devcamcar has joined #openstack | 03:19 | |
*** joshuamckenty has quit IRC | 03:19 | |
*** g-ram has joined #openstack | 03:19 | |
*** joshuamckenty has joined #openstack | 03:19 | |
*** jdmaturen has quit IRC | 03:19 | |
*** tanto has joined #openstack | 03:21 | |
*** toddmorey has joined #openstack | 03:23 | |
*** g-ram has quit IRC | 03:23 | |
*** g-ram has joined #openstack | 03:24 | |
*** joshuamckenty has quit IRC | 03:24 | |
*** joshuamckenty has joined #openstack | 03:24 | |
*** hugod has joined #openstack | 03:28 | |
*** devcamcar has quit IRC | 03:28 | |
*** lbieber has joined #openstack | 03:28 | |
*** cparedes has joined #openstack | 03:29 | |
*** pvo has quit IRC | 03:29 | |
*** joshuamckenty has quit IRC | 03:29 | |
*** devcamcar has joined #openstack | 03:29 | |
*** joshuamckenty has joined #openstack | 03:29 | |
*** pvo has joined #openstack | 03:29 | |
*** g-ram has quit IRC | 03:33 | |
*** joshuamckenty has quit IRC | 03:33 | |
*** pvo has quit IRC | 03:34 | |
*** joshuamckenty has joined #openstack | 03:34 | |
*** abecc has quit IRC | 03:34 | |
*** dendrobates is now known as dendro-afk | 03:40 | |
*** devcamcar has quit IRC | 03:40 | |
*** toddmorey has quit IRC | 03:42 | |
*** jakedahn_ has quit IRC | 03:42 | |
chmouel | anyone know how to get a rss feed of a launchpad repo (ie: for swift) | 03:43 |
creiht | chmouel: which part do you want rss of... commits? | 03:44 |
chmouel | yup i get the bugs by email | 03:44 |
notmyname | chmouel: feed://feeds.launchpad.net/swift/revisions.atom | 03:45 |
creiht | chmouel: you can also go here: https://code.launchpad.net/~hudson-openstack/swift/trunk | 03:46 |
chmouel | notmyname: what do you do up at this time :) kids asleep ? thanks by the way | 03:46 |
chmouel | cheers | 03:46 |
creiht | and click on the subscribe yourself to get notifications by email | 03:46 |
creiht | hehe | 03:46 |
notmyname | chmouel: ya. everyone is asleep. about to go to bed myself | 03:46 |
chmouel | ah nice http://blog.launchpad.net/bug-tracking/gmail-filters-for-launchpad-bug-email | 03:47 |
creiht | chmouel: ahh nice fi | 03:47 |
creiht | erm find | 03:48 |
*** jakedahn has joined #openstack | 03:49 | |
*** joshuamckenty has quit IRC | 03:49 | |
*** joshuamckenty has joined #openstack | 03:49 | |
*** devcamcar has joined #openstack | 03:50 | |
notmyname | what's the best way to pep8-ify a comment line that has a URL that busts the 80 char limit? | 03:54 |
*** devcamcar has quit IRC | 03:54 | |
creiht | notmyname: tinyurl? :) | 03:54 |
notmyname | hehe | 03:54 |
creiht | and yeah that type of thing can be frustrating | 03:55 |
*** pvo has joined #openstack | 03:55 | |
*** pvo has joined #openstack | 03:55 | |
chmouel | anyone tried http://github.com/pieter/git-bzr ? | 03:56 |
pvo | yep, seems to work ok | 03:56 |
chmouel | pvo: did you have to apply the patch to git and recompile? | 03:57 |
pvo | no jsut teh plugins | 03:57 |
chmouel | ah k it's for < 1.6.0 to apply the patch.. | 03:58 |
*** bpiatt has joined #openstack | 04:02 | |
*** devcamcar has joined #openstack | 04:03 | |
notmyname | gah! can't even put the URL in the docstring (less indentation). it still overflows. why does stackoverflow have to use such long urls? | 04:05 |
*** devcamcar has quit IRC | 04:05 | |
notmyname | I suppose I could remove the comment, but that seems wrong too :) | 04:05 |
*** devcamcar has joined #openstack | 04:06 | |
*** Renich has joined #openstack | 04:08 | |
Renich | congrats on the release! | 04:08 |
Renich | One question... kvm/qemu or xen??? | 04:09 |
*** bpiatt has quit IRC | 04:09 | |
chmouel | Renich: i don't know much but i believe it's kvm via libvirt and xen seems to be planned for the future | 04:10 |
chmouel | but i could be wrong as well :) | 04:10 |
*** pharkmillups has joined #openstack | 04:10 | |
*** xtoddx has quit IRC | 04:11 | |
*** abecc has joined #openstack | 04:11 | |
Renich | chmouel: I'm interested on kvm/qemu via libvirt | 04:11 |
Renich | ;) | 04:11 |
*** abecc has quit IRC | 04:11 | |
Renich | chmouel: is no longer the best... kvm/qemu has the upperhand | 04:12 |
*** adjohn_ has joined #openstack | 04:16 | |
*** Renich has left #openstack | 04:17 | |
*** devcamcar has quit IRC | 04:17 | |
*** adjohn has quit IRC | 04:17 | |
*** adjohn_ is now known as adjohn | 04:17 | |
*** pvo has quit IRC | 04:18 | |
*** devcamcar has joined #openstack | 04:18 | |
*** pvo_ has joined #openstack | 04:19 | |
*** devcamcar has quit IRC | 04:19 | |
*** devcamcar has joined #openstack | 04:21 | |
*** jdmaturen has joined #openstack | 04:21 | |
chmouel | ah yeah seems like xen does not keep up with the latest kernels as well | 04:22 |
*** joshuamckenty has quit IRC | 04:22 | |
*** joshuamckenty has joined #openstack | 04:22 | |
*** pvo_ is now known as pvo | 04:23 | |
*** joshuamckenty has quit IRC | 04:26 | |
*** devcamcar has quit IRC | 04:26 | |
*** devcamcar has joined #openstack | 04:27 | |
*** Predominant has joined #openstack | 04:32 | |
*** pvo has quit IRC | 04:39 | |
*** devcamcar has quit IRC | 04:39 | |
*** pvo has joined #openstack | 04:39 | |
*** pvo has quit IRC | 04:44 | |
*** blakeyeager has quit IRC | 04:44 | |
*** devcamcar has joined #openstack | 04:48 | |
*** Predominant has quit IRC | 04:50 | |
*** devcamcar has quit IRC | 04:50 | |
soren | Hey, NASA guys: Just glancing at: https://bugs.launchpad.net/nova/+bug/607512 | 04:50 |
uvirtbot | Launchpad bug 607512 in nova "Authorization not checked on objectstore" [Undecided,New] | 04:50 |
*** Predominant has joined #openstack | 04:50 | |
soren | We clearly do check the signature in the line just above it. | 04:51 |
soren | What's the comment really about? | 04:51 |
uvirtbot | New bug: #607388 in swift "Add graceful init.d script option" [Wishlist,Invalid] https://launchpad.net/bugs/607388 | 04:51 |
soren | Is it just a one that was never removed after the signature check was added or does it refer to the RBAC ACL check? | 04:51 |
*** devcamcar has joined #openstack | 04:51 | |
soren | Or some third option I did not consider? | 04:51 |
soren | Just trying to understand the history here. | 04:52 |
*** pharkmillups has quit IRC | 04:58 | |
*** devcamcar has quit IRC | 04:58 | |
*** poli has quit IRC | 04:59 | |
*** devcamcar has joined #openstack | 04:59 | |
*** s1rp has quit IRC | 05:01 | |
*** jonesy has quit IRC | 05:07 | |
*** jonesy_ is now known as jonesy | 05:07 | |
*** lbieber has quit IRC | 05:08 | |
*** greg-g has quit IRC | 05:08 | |
justinsb | soren: What was the motivation behind the twisted rewrite of the objectstore, if it's a stopgap measure anyway? I keep finding more and more incompatibilities... | 05:21 |
*** devcamcar has quit IRC | 05:21 | |
uvirtbot | New bug: #607572 in nova "ObjectStore now checks security (Commit 146), compute node doesn't authenticate" [Undecided,New] https://launchpad.net/bugs/607572 | 05:21 |
*** devcamcar has joined #openstack | 05:22 | |
*** tupshin has joined #openstack | 05:27 | |
*** devcamcar has quit IRC | 05:27 | |
*** devcamcar has joined #openstack | 05:28 | |
*** pvo has joined #openstack | 05:34 | |
*** devcamcar has quit IRC | 05:34 | |
*** devcamcar has joined #openstack | 05:35 | |
*** veszig has joined #openstack | 05:36 | |
*** devcamcar has quit IRC | 05:36 | |
*** devcamcar has joined #openstack | 05:37 | |
*** armmer has quit IRC | 05:59 | |
*** devcamcar has quit IRC | 05:59 | |
*** armmer has joined #openstack | 05:59 | |
*** devcamcar has joined #openstack | 06:00 | |
*** pvo has quit IRC | 06:05 | |
*** devcamcar has quit IRC | 06:05 | |
*** pvo has joined #openstack | 06:06 | |
*** devcamcar has joined #openstack | 06:06 | |
*** matiu has quit IRC | 06:09 | |
*** devcamcar has quit IRC | 06:09 | |
*** pvo has quit IRC | 06:10 | |
*** devcamcar has joined #openstack | 06:11 | |
*** Jorijn has joined #openstack | 06:15 | |
*** devcamcar has quit IRC | 06:16 | |
*** devcamcar has joined #openstack | 06:17 | |
*** pvo has joined #openstack | 06:19 | |
*** devcamcar has quit IRC | 06:19 | |
soren | justinsb: I'm not sure I understand what you mean by it being a stopgap measure? | 06:22 |
soren | justinsb: In fact, I'm completely sure I don't understand it :) | 06:23 |
soren | justinsb: ...regardless, the motivation was to get rid of tornado. | 06:23 |
*** jakedahn has quit IRC | 06:24 | |
*** nettok has quit IRC | 06:25 | |
*** Thweetch has left #openstack | 06:26 | |
*** devcamcar has joined #openstack | 06:28 | |
jdmaturen | out of curiosity, why get rid of tornado? | 06:30 |
*** devcamcar has quit IRC | 06:30 | |
jdmaturen | [feel free to not touch that question, its late and whatnot …] | 06:30 |
soren | jdmaturen: In this particular care, it simply wasn't up to the task of serving these quite big files. | 06:30 |
soren | s/care/case/ | 06:30 |
jdmaturen | ah | 06:30 |
jdmaturen | makes sense | 06:30 |
*** devcamcar has joined #openstack | 06:31 | |
soren | Besides, it didn't really buy us anything. The twisted replacement is, if I may say so myself, quite lovely. It's shorter than the version that uses tornado and actually gets the job done without nginx in front of it. | 06:32 |
*** epicurus has quit IRC | 06:33 | |
*** dendro-afk is now known as dendrobates | 06:33 | |
jdmaturen | nice | 06:36 |
*** devcamcar has quit IRC | 06:36 | |
*** devcamcar has joined #openstack | 06:37 | |
*** John|RDP has joined #openstack | 06:40 | |
*** devcamcar has quit IRC | 06:40 | |
justinsb | Well, it seems that the swift team thinks that we'll be replacing the nova objectstore with swift! | 06:41 |
*** devcamcar has joined #openstack | 06:42 | |
justinsb | e.g. vish1: "notmyname: we have a ghetto version of storage which we need to support until nova-storage integration is complete" | 06:42 |
*** rajijoom has joined #openstack | 06:42 | |
justinsb | So if we're going to rip it out anyway, seems silly to rewrite it for performance reasons | 06:42 |
justinsb | I'm just frustrated because I've spent the whole day chasing bugs that the rewrite introduced | 06:42 |
soren | justinsb: I don't think the S3 implementation will go away. | 06:43 |
justinsb | (Some of which aren't really bugs, just incompatibilities. Sometimes the rewrite is better.) | 06:43 |
soren | justinsb: | 06:43 |
soren | Whoops | 06:43 |
soren | justinsb: Swift will just be the preferred objectstore. | 06:43 |
justinsb | Isn't swift the S3 implementation? Or is that the current nova objectstore? | 06:44 |
soren | Swift is Rackspace Cloud Files. | 06:44 |
justinsb | Ah ... so we still need the nova objectstore so that it can speak S3 to e.g. the euca tools? | 06:45 |
soren | Hmmm.. Does that answer your question? | 06:45 |
soren | justinsb: I'm not entirely sure how this is going to pan out. | 06:45 |
soren | I see a couple of options. | 06:45 |
soren | The thing is, we don't want to force people who are really mostly intereested in Nova to run a full swift installation. That'd be overkill. | 06:46 |
justinsb | Fair enough. But I'm afraid that the twisted implementation isn't going to be quite so lovely once all the patches and bugfixes are in :-) | 06:46 |
soren | So Swift needs a simple mode of operation for Nova development. | 06:46 |
soren | or | 06:46 |
soren | We need a simple implementation of Swift's API inside Nova, akin to the S3 implementation. | 06:46 |
justinsb | I definitely agree that we need a low-dependency initial install of Nova | 06:47 |
soren | Depending on what we're going to do there, maybe Swift will grow an S3 frontend. | 06:47 |
*** pdelgallego has joined #openstack | 06:47 | |
soren | justinsb: Let me look at the current bugs reported. Gimme a minute. | 06:48 |
justinsb | I do think it seems very likely that someone will add a S3 frontend to Swift (if there isn't one there already). And I also see that we need our own micro-implementation | 06:48 |
*** tomu has joined #openstack | 06:48 | |
soren | justinsb: Or Swift will need a very simple mode of operation that basically does the same thing. | 06:49 |
*** devcamcar has quit IRC | 06:49 | |
*** pvo has quit IRC | 06:49 | |
*** pvo has joined #openstack | 06:50 | |
*** devcamcar has joined #openstack | 06:50 | |
justinsb | My personal feeling is that the EC2 approach for images is overly complex, and so I like the fact that we 'control our own destiny here' | 06:50 |
justinsb | It's just painful along the way!! | 06:51 |
soren | justinsb: I agree. I have a few things I need to sort of before bed, and I'll work on getting all of this fallout fixed first thing in the morning. Are we in the same timezone right now? | 06:52 |
soren | (PDT) | 06:52 |
justinsb | Yes indeed! Are you in Portland? | 06:53 |
soren | I am. | 06:53 |
justinsb | Hope OSCON is good. I think I've got past all these problems... upload now works with all the patches, and download works by setting use_s3=false | 06:53 |
soren | Ah, tricksy :) | 06:53 |
justinsb | Because it turns out the old objectstore didn't enforce security on image download!! | 06:54 |
soren | I forgot about the use_s3 option. | 06:54 |
justinsb | Anyway, on to the template | 06:54 |
soren | As a byproduct of the nginx dependency, it didn't really have to. | 06:54 |
justinsb | Trying to get it so that libvirt.xml doesn't have an image or ramdisk if one isn't specified | 06:54 |
soren | ...since image downloads were on a different port, you couldn firewall it off. | 06:54 |
*** pvo has quit IRC | 06:54 | |
soren | justinsb: That's going to be more tricky than that. | 06:55 |
soren | justinsb: ...since there are default values for kernels and ramdisks. | 06:55 |
justinsb | Yes... for the unit tests. Right now, I seem to be getting away with it (I think because the round-trip via redis makes the values empty strings rather than None) | 06:56 |
justinsb | And I'm not looking a gift horse in the mouth for nwo | 06:56 |
justinsb | nwo -> now | 06:56 |
*** dendrobates is now known as dendro-afk | 07:03 | |
*** devcamcar has quit IRC | 07:03 | |
holoway | soren: it also feels like swift is built like a CDN more than a simpler blob store | 07:04 |
holoway | soren: I haven't looked at how the publishing works in swift, but the cloudfiles stuff feels much more like a cdn than a blob store | 07:05 |
holoway | soren: to the end user | 07:05 |
*** pvo has joined #openstack | 07:05 | |
soren | holoway: How so? | 07:06 |
*** pvo has left #openstack | 07:06 | |
holoway | soren: for example, there is a publishing step in the current cloudfiles stuff | 07:06 |
holoway | soren: TTLs built in by default | 07:06 |
holoway | whereas S3 feels like a distributed blob store, and you can build a CDN on top | 07:07 |
holoway | if that makes sense | 07:07 |
soren | The data model just allows for the metadata for a CDN to be included in the primary store. If you don't CDN-enable a container, there's not much CDN-y about it. You can only access things through the cloud files API then. | 07:09 |
holoway | ah, gotcha | 07:09 |
*** pvo has joined #openstack | 07:10 | |
*** devcamcar has joined #openstack | 07:13 | |
*** pvo has quit IRC | 07:15 | |
*** devcamcar has quit IRC | 07:15 | |
*** devcamcar has joined #openstack | 07:16 | |
*** miclorb_ has quit IRC | 07:24 | |
*** king has joined #openstack | 07:31 | |
*** devcamcar has quit IRC | 07:31 | |
*** king is now known as Guest38549 | 07:31 | |
*** devcamcar has joined #openstack | 07:32 | |
*** jandresen has joined #openstack | 07:36 | |
*** barismetin has joined #openstack | 07:46 | |
*** jdmaturen has quit IRC | 07:49 | |
*** devcamcar has quit IRC | 07:49 | |
*** miclorb has joined #openstack | 07:50 | |
*** jandresen has quit IRC | 07:52 | |
*** miclorb has quit IRC | 07:56 | |
*** devcamcar has joined #openstack | 07:59 | |
*** calavera has joined #openstack | 08:08 | |
*** devcamcar has quit IRC | 08:08 | |
*** devcamcar has joined #openstack | 08:09 | |
*** jhalogen has quit IRC | 08:10 | |
*** Lartsa has left #openstack | 08:18 | |
*** jeromatron has quit IRC | 08:31 | |
*** devcamcar has quit IRC | 08:31 | |
*** patcito has quit IRC | 08:33 | |
*** devcamcar has joined #openstack | 08:40 | |
*** ewan_ has joined #openstack | 08:41 | |
*** devcamcar has quit IRC | 08:48 | |
*** ewan_ has quit IRC | 08:49 | |
*** ewan_ has joined #openstack | 08:50 | |
*** croteb has quit IRC | 08:53 | |
*** croteb has joined #openstack | 08:54 | |
*** miclorb_ has joined #openstack | 08:56 | |
*** barismetin has quit IRC | 09:01 | |
*** ttx has joined #openstack | 09:02 | |
*** adjohn has quit IRC | 09:02 | |
*** adjohn has joined #openstack | 09:03 | |
*** ewan_ has quit IRC | 09:14 | |
*** adjohn has quit IRC | 09:19 | |
*** adjohn has joined #openstack | 09:26 | |
*** barismetin has joined #openstack | 09:30 | |
*** adjohn has quit IRC | 09:35 | |
*** kim0 has joined #openstack | 09:47 | |
*** avi has quit IRC | 09:58 | |
*** jtdowney has joined #openstack | 10:03 | |
*** avi has joined #openstack | 10:13 | |
*** jonesy has quit IRC | 10:21 | |
thom | hey, the install-swift.sh process appears to be wrong - the cookbooks that are downloaded wants to pull from lp:~swift-core/swift/trunk , which doesn't exist. lp:swift/trunk works tho | 10:21 |
*** ewan_ has joined #openstack | 10:25 | |
*** bretpiatt has joined #openstack | 10:26 | |
*** z0 has joined #openstack | 10:44 | |
*** bretpiatt has quit IRC | 10:46 | |
*** calavera has quit IRC | 10:50 | |
*** calavera has joined #openstack | 10:50 | |
*** mufasa_uk has joined #openstack | 10:54 | |
*** tupshin has quit IRC | 11:01 | |
*** vish1 has quit IRC | 11:01 | |
*** urandom has quit IRC | 11:01 | |
*** tonywolf has quit IRC | 11:01 | |
*** _cerberus_ has quit IRC | 11:01 | |
*** tonywolf1 has joined #openstack | 11:01 | |
*** ttx has quit IRC | 11:01 | |
*** tupshin has joined #openstack | 11:02 | |
*** vish1 has joined #openstack | 11:03 | |
*** urandom has joined #openstack | 11:03 | |
*** _cerberus_ has joined #openstack | 11:03 | |
*** kim0 has quit IRC | 11:03 | |
*** tj_yang_ has joined #openstack | 11:05 | |
*** tj_yang__ has joined #openstack | 11:05 | |
*** jonesy has joined #openstack | 11:05 | |
*** tj_yang_ has quit IRC | 11:06 | |
*** tj_yang__ has quit IRC | 11:06 | |
*** jonesy_ has joined #openstack | 11:06 | |
*** tj_yang_ has joined #openstack | 11:06 | |
*** hogstrom has joined #openstack | 11:14 | |
*** miclorb_ has quit IRC | 11:17 | |
*** sophiap has joined #openstack | 11:21 | |
*** jtdowney has quit IRC | 11:44 | |
*** tempspace has joined #openstack | 11:47 | |
tempspace | god morning all | 11:47 |
*** vish1 has quit IRC | 11:47 | |
*** justinsheehy has quit IRC | 11:56 | |
*** kim0 has joined #openstack | 11:57 | |
*** hogstrom_ has joined #openstack | 11:57 | |
mufasa_uk | s/god/good/ | 11:57 |
*** hogstrom_ has quit IRC | 11:58 | |
*** vish1 has joined #openstack | 11:58 | |
*** hogstrom has quit IRC | 11:58 | |
*** piers has joined #openstack | 12:03 | |
*** murkk has quit IRC | 12:03 | |
jonesy | my question list is getting long :-P | 12:04 |
*** vish1 has quit IRC | 12:04 | |
creiht | thom: that recipe is a bit old now. I need to track down adam and see if he will update it | 12:07 |
*** perry753 has joined #openstack | 12:07 | |
exlt | jonesy: there are mailing lists, if irc gets a little busy and your questions need more careful consideration :) | 12:07 |
exlt | jonesy: but fire away here - instant gratification is a good thing sometimes | 12:09 |
*** soulresin has joined #openstack | 12:11 | |
creiht | soren: it may be that a simplified version of swift could be created pretty easily... let me ponder that while I eat breakfast | 12:13 |
*** CnC has joined #openstack | 12:13 | |
jonesy | exlt: I've moved from docs to the code itself, so I'm trying to answer some of my own questions and make the list smaller. | 12:16 |
jonesy | exlt: but if I can't resist the urge, I'll ask -- thanks a bunch :) | 12:17 |
exlt | resistance is futile | 12:18 |
thom | creiht: right; you could actually fix it without altering the recipe - just add "bzr_branch": "lp:swift/trunk", into swift-template.json | 12:19 |
exlt | but I'm glad you're digging in - that's awesome - nitpicking is welcomed | 12:19 |
*** jdarcy has joined #openstack | 12:20 | |
*** piers has left #openstack | 12:22 | |
jdarcy | What's the deal with all the rsync stuff in development_saio? Do I really need that? | 12:23 |
jonesy | exlt: I don't know the code well enough to really pick at it, and some of my nits are addressed in the copious TODOs in the code, which is awesome. | 12:24 |
*** jaker74 has joined #openstack | 12:29 | |
jonesy | exlt: ok, so I'm getting into code when what I really want is an answer about higher-level architecture. I've seen the (rather nice) diagrams on the wiki, but it didn't answer my question, which is basically this: does Nova deployment need to be "all or nothing"? Can I introduce it to a client by having them leverage, say, the autodeployment features of Nova without forcing them to immediately have an API server and the rest? | 12:29 |
jonesy | from looking at the code, it seems modular enough, but I haven't been able to map out all of the interdependencies. | 12:30 |
jonesy | yet :) | 12:30 |
jonesy | I'm wondering if I'd need to have them use a model where they leave existing stuff as it is and deploy new things to a cloud, or if I can have them migrate some existing services to a cloud with minimal disruption to things that depend on those services. | 12:31 |
*** Guest38549 has quit IRC | 12:31 | |
*** e-vad has joined #openstack | 12:32 | |
CnC | theres no one simple answer to that question | 12:32 |
CnC | imo | 12:32 |
h1tman | morning all | 12:32 |
CnC | evening | 12:33 |
h1tman | hows everything going with openstack? any breakthroughs overnight? | 12:34 |
jonesy | CnC: hence the code-sifting. I imagine it's possible. Question is whether it's practical. We'll see :) | 12:37 |
CnC | its more of a business question than a technical question really | 12:38 |
jonesy | it's a deployment question. It's related to how decoupled the various pieces *within* Nova are. | 12:39 |
jonesy | If all I need to do is satisfy the requirements of node.Instance, and write some wrapper code to build an autodeployment system, that's a starting point for deployment. | 12:40 |
jonesy | if I posed this question to the business folks, they'd look at me sideways and say "we hired you to answer that question" | 12:41 |
CnC | heh | 12:41 |
CnC | if you asked the question in the wrong way yes they would | 12:41 |
*** jaker74 has quit IRC | 12:42 | |
*** jaker74 has joined #openstack | 12:43 | |
jonesy | CnC: the reality is that it's not a business question. If I've posed the question in some way that makes you think it is a business question, let me know, because then I've been unclear. | 12:45 |
*** aliguori has joined #openstack | 12:45 | |
jonesy | I provide technical solutions to business problems. At the end of the day, almost all of the questions that come my way can be turned back around on the business, but nothing gets done that way. | 12:45 |
*** pengwynn has joined #openstack | 12:48 | |
*** joshuamckenty has joined #openstack | 12:51 | |
*** ericography has joined #openstack | 12:52 | |
*** snowboarder04 has joined #openstack | 12:53 | |
*** pdelgallego has quit IRC | 12:53 | |
*** sophiap has quit IRC | 12:54 | |
*** rackninja has joined #openstack | 12:56 | |
*** rackninja has left #openstack | 12:57 | |
*** doublerr has joined #openstack | 12:59 | |
*** tempspace has quit IRC | 13:00 | |
*** larsemil is now known as pelleplutt | 13:01 | |
*** pelleplutt is now known as larsemil | 13:01 | |
joshuamckenty | soren / mtaylor - do you have a strong opinion about where on the openstack.org wiki I should be sticking blueprints? | 13:04 |
*** tomu has left #openstack | 13:05 | |
*** sophiap has joined #openstack | 13:05 | |
*** pdelgallego has joined #openstack | 13:06 | |
*** joshuamckenty has quit IRC | 13:18 | |
*** perry753 has left #openstack | 13:23 | |
creiht | jdarcy: re: rsync in saio... replication uses rsync, so if you want to turn replication on, you will need the rsync configs. If you do not want to play with replication, then you can skip the rsync configuration | 13:23 |
*** tempspace has joined #openstack | 13:25 | |
*** tempspace has quit IRC | 13:29 | |
*** tempspace has joined #openstack | 13:30 | |
*** pvo has joined #openstack | 13:31 | |
*** lbieber has joined #openstack | 13:35 | |
*** pvo has quit IRC | 13:40 | |
*** pvo has joined #openstack | 13:41 | |
*** poli has joined #openstack | 13:41 | |
*** jbd has quit IRC | 13:42 | |
*** lbieber has quit IRC | 13:43 | |
*** jbd has joined #openstack | 13:43 | |
*** pvo has quit IRC | 13:45 | |
*** pvo has joined #openstack | 13:49 | |
*** termie has joined #openstack | 13:49 | |
*** jeromatron has joined #openstack | 13:50 | |
*** tempspace has quit IRC | 13:52 | |
*** tempspace has joined #openstack | 13:54 | |
*** g-ram has joined #openstack | 13:56 | |
*** ttx has joined #openstack | 13:56 | |
*** Phil_FL has joined #openstack | 14:00 | |
*** justinsheehy has joined #openstack | 14:01 | |
*** tonywolf1 has quit IRC | 14:02 | |
*** tonywolf has joined #openstack | 14:04 | |
*** mtaylor has quit IRC | 14:06 | |
*** cloud0 has joined #openstack | 14:07 | |
*** tobym has quit IRC | 14:07 | |
*** Predominant has quit IRC | 14:07 | |
*** mtaylor has joined #openstack | 14:08 | |
*** ChanServ sets mode: +v mtaylor | 14:08 | |
*** JordanRinke has joined #openstack | 14:09 | |
* JordanRinke looks around | 14:09 | |
* scotticus waves | 14:13 | |
chmouel | is conrad around ? | 14:13 |
* JordanRinke waves back | 14:13 | |
*** ewan_ has quit IRC | 14:17 | |
*** gustavomzw has joined #openstack | 14:19 | |
chmouel | anyone wants to help some user with his php on macos regards to the cloudfiles/swift PHP-API | 14:19 |
*** murkk has joined #openstack | 14:20 | |
*** jakedahn has joined #openstack | 14:21 | |
*** murkk has quit IRC | 14:22 | |
*** murkk has joined #openstack | 14:22 | |
*** jsm has joined #openstack | 14:25 | |
*** bretpiatt has joined #openstack | 14:25 | |
*** tobym has joined #openstack | 14:26 | |
*** vish1 has joined #openstack | 14:28 | |
vish1 | good morning | 14:29 |
*** com1 has joined #openstack | 14:29 | |
*** anotherjesse has joined #openstack | 14:29 | |
JordanRinke | vish1: Morning | 14:29 |
creiht | morning | 14:29 |
com1 | Greetings | 14:29 |
*** filler has joined #openstack | 14:29 | |
jsm | bacon | 14:29 |
bretpiatt | vish1, good morning, time for the OSCON Cloud Summit today! | 14:29 |
creiht | mmmmm... bacon | 14:29 |
vish1 | summit going well? | 14:30 |
com1 | Impressively active channel :) | 14:30 |
*** dendro-afk is now known as dendrobates | 14:30 | |
com1 | Any rackers here? | 14:31 |
*** dendrobates is now known as dendro-afk | 14:31 | |
*** mtaylor is now known as mtaylor|breakfas | 14:31 | |
gholt | Si | 14:31 |
creiht | com1: several | 14:31 |
*** dendro-afk is now known as dendrobates | 14:31 | |
bretpiatt | vish1, summit is 1 day, today, starts in 1.5 hours, so as of now everything has gone perfectly | 14:32 |
notmyname | com1: hi | 14:32 |
jsm | is there a good doc anywhere detailing eucalyptus vs nova? | 14:32 |
*** bretpiatt is now known as bret|away | 14:32 | |
vish1 | jsm: not really | 14:33 |
com1 | gholt: creiht: great! I am a rackspace customer and partner, and just saw the news on openstack in my inbox from lanham's release this morning. Great news! Congratulations! | 14:33 |
creiht | com1: thanks! and welcome | 14:33 |
anotherjesse | jsm: not that I know of. the major difference for me is not the features but the architecture | 14:33 |
jsm | vish1: something must be better about it.. trying to decide why rackspace uses it instead of eucalyptus . | 14:34 |
g0rdy | or OpenNebula for that matter .. | 14:34 |
jonesy_ | jsm: reportedly, the ability to scale :) | 14:35 |
jsm | right | 14:35 |
creiht | jsm: http://tinyurl.com/24swha5 is an interesting article that seems to touch on it a bit | 14:35 |
com1 | creiht: Thanks! i've also spent a lot of time with the not-so-openstack, microsoft virtualization and platforms (things like DDC) and eager to see how things develop and ways they may fit in. | 14:36 |
anotherjesse | jsm: eucalyptus's queue, db, user system, ... is all part of the cloud controller java process | 14:36 |
anotherjesse | instead of having external components (datastore/queues/users) that can scale independently | 14:36 |
jsm | Interesting, when I was working for the government, I was told that the only license the government could release software with was Public Domain | 14:37 |
jsm | or completely closed :) | 14:38 |
anotherjesse | jsm: open source + govt is complicated ... we have spent over a year working on getting apache approved | 14:38 |
com1 | I've also been following a conversation on various cloud platforms, is ubuntu's enterprise cloud somehow tied into eucalyptus? | 14:38 |
chmouel | com1: yes uec use eucalyptus | 14:38 |
jsm | I won't be surprised if there is some anti-competition lawsuit.. | 14:39 |
anotherjesse | jsm: how does apache license not allow companies to use it? | 14:39 |
anotherjesse | for either open or closed products with apache code | 14:39 |
jsm | anotherjesse: i'm not validating the lawsuits :).. i'm just saying | 14:39 |
jsm | people are sue happy | 14:39 |
jsm | anotherjesse: iirc, the argument used to be, if the software was created with government tax dollars, then it shouldn't be restricted to any license. | 14:41 |
jsm | i'd be interested to read how this project got around that... i would have loved to release some open source stuff while i was working for the gov't.. maybe the process should be detailed and documented so other agencies can do the same. | 14:42 |
gholt | :) | 14:42 |
*** Predominant has joined #openstack | 14:43 | |
*** sirp1 has joined #openstack | 14:43 | |
jsm | anotherjesse: is there someone I can contact wrt to this? | 14:44 |
jonesy_ | jsm: I'd think that argument is easy to defeat when the assumption that only Americans will use it is thrown out :) | 14:45 |
*** tgodar has joined #openstack | 14:47 | |
anotherjesse | jsm: use - jesse.d.andrews@nasa.gov - we are working on fixing the process at nasa... | 14:48 |
anotherjesse | it is a many step process | 14:48 |
jsm | anotherjesse: thks | 14:49 |
*** g-ram has quit IRC | 14:50 | |
uvirtbot | New bug: #607068 in swift "swift-1.0.0.tar.gz is missing doc directory" [High,Fix released] https://launchpad.net/bugs/607068 | 14:52 |
uvirtbot | New bug: #607074 in swift "setup.py find_packages typo" [Undecided,Fix released] https://launchpad.net/bugs/607074 | 14:52 |
uvirtbot | New bug: #607186 in swift "proxy should return 500 on exception" [Medium,Fix released] https://launchpad.net/bugs/607186 | 14:52 |
creiht | anotherjesse: there was some talk over the evening about the built in object store for nova | 14:52 |
anotherjesse | do we have scrollback? | 14:52 |
creiht | If there was a simple striped down version of swift, would that help? | 14:53 |
creiht | not sure if we have a logger | 14:53 |
*** jeromatron has quit IRC | 14:53 | |
*** jonesy_ has left #openstack | 14:54 | |
*** jhalogen has joined #openstack | 14:54 | |
*** poli has quit IRC | 14:54 | |
*** poli has joined #openstack | 14:54 | |
notmyname | creiht: I've got the channel logged | 14:55 |
creiht | mtaylor|breakfas: We need a channel logger :) | 14:55 |
creiht | so someone could go to say irc.openstack.org and see channel logs | 14:56 |
anotherjesse | creiht: I think so - we were debating just updating the objectstore in nova to have the same API | 14:56 |
*** tempspace has quit IRC | 14:58 | |
anotherjesse | if there is swift-mini that would be hawt though | 14:58 |
creiht | anotherjesse: I'm contemplating something like that | 14:59 |
creiht | let me think about it a little more | 15:00 |
anotherjesse | creiht: would be nice to remove the objectstore since it was the minimal naive implementation to support uploading/registered bundled images - minimal because we knew a raid backed objectstore was wasted effort | 15:03 |
creiht | anotherjesse: would it be an issue if it still required the swift source? | 15:04 |
*** lbieber has joined #openstack | 15:05 | |
*** WaterDog has joined #openstack | 15:05 | |
jonesy | looking for a testbed for my openstack playground. Considered putting it on my existing rackspace cloud ;-p | 15:07 |
creiht | yo dawg, I heard you like clouds | 15:07 |
soulresin | put a cloud in yo cloud | 15:08 |
anotherjesse | jonesy: it should work with 2 modifications - you need to use QEMU not KVM (modify the libvirt template) and use flat_networking | 15:08 |
JordanRinke | soulresin: would that create a double rainbow? | 15:08 |
jonesy | heh -- I'm pretty sure I'm just gonna repurpose some physical hardware somewhere. Or find a client with hardware I can use (I have *one* that should have lots of it) | 15:08 |
com1 | anotherjesse: might it possibly work within hyper-v (xen) too? | 15:09 |
soulresin | doesn't everyone use a cloud for their desktop? | 15:09 |
anotherjesse | com1: yes - there are patches for that already | 15:09 |
jonesy | soulresin: does Xnest count? ;-) | 15:09 |
exlt | soulresin: my kbd would fall through the vapor.. | 15:09 |
anotherjesse | com1: also if you are using kvm & amd you can do kvm inside of kvm | 15:10 |
anotherjesse | soulresin: we run cloud on our linux desktops all the time during dev :) | 15:10 |
soulresin | :) | 15:10 |
com1 | anotherjesse: very nice, we were on amd, now we're on some shiny new dell servers with nehalem procs ... from your experience, do you have any preference? | 15:11 |
anotherjesse | com1: soren has done amd+nested kvm --- our production hardware is various vendors - mostly westmere | 15:13 |
*** jakedahn has quit IRC | 15:14 | |
WaterDog | Is there any benefit to swift vs Gluster? | 15:14 |
creiht | Where's devin when you need him? :) | 15:16 |
creiht | WaterDog: I personally haven't used glusterfs | 15:16 |
anotherjesse | WaterDog: at least you didn't say lustre | 15:17 |
creiht | anotherjesse: were you guys using glusterfs or lusterfs? | 15:17 |
WaterDog | I'm fairly certain lustre would lead to an early death for me. | 15:17 |
creiht | hehe | 15:17 |
anotherjesse | creiht: lustre - which is great for the supercomputer - which operates in batch mode - but is not that good for "online storage" | 15:17 |
creiht | k | 15:18 |
*** pvo_ has joined #openstack | 15:20 | |
creiht | WaterDog: Well the first thing is that the docs seem to indicate scaling to an upper bound of scaling to severla petabytes, where we are designing swift to scale to 100's of petabytes | 15:20 |
creiht | WaterDog: Gluster also seems to work more around filesystem semantics, while swift is more blob storage like S3 | 15:21 |
* creiht notes that he is making these assumptions just from what he reads online | 15:21 | |
anotherjesse | WaterDog: glusterfs is very interesting for certain usecases - but yeah what creiht said - at nasa the earth scientists come to use with pilots for nebula that run in the multi-petabytes... some of the data is irreplaceable without relaunching spacecrafts | 15:22 |
anotherjesse | WaterDog: so swift is really interesting to us because we can modify the user interface so users can say that this data is really important and keep > 3 copies | 15:23 |
*** pvo has quit IRC | 15:24 | |
*** pvo_ is now known as pvo | 15:24 | |
WaterDog | Oh, I totally get the limitations around filesystems, etc. It is just a more familiar tradition (to me) | 15:24 |
jdarcy | Is there any particular reason the scripts and config files included inline in developer-SAIO can't be checked in separately in a samples directory? | 15:25 |
creiht | jdarcy: We are thinking about something like that | 15:25 |
jdarcy | creiht: Cool. In the RHEL instructions, should I refer to them in the doc for now, and update later? | 15:26 |
*** jeromatron has joined #openstack | 15:27 | |
creiht | hrm | 15:27 |
creiht | yeah I would just refer to them in the doc for now | 15:27 |
jdarcy | FYI, since I see that folks are discussing Lustre/GlusterFS vs. Swift above, I was just in a meeting where I was pushing both GlusterFS and S3/Swift for different uses. | 15:27 |
jdarcy | I have nothing good to say about Lustre, other than that a full-time team of developers and admins can make it work well given enough time. | 15:28 |
creiht | seems like GlusterFS is much more like something one would use for block storage (like EBS) | 15:28 |
* jdarcy spent two years trying to make the damn thing usable at SiCortex. | 15:29 | |
creiht | heh | 15:29 |
*** jhalogen has quit IRC | 15:29 | |
*** Thweetch has joined #openstack | 15:29 | |
jdarcy | I think GlusterFS is a good way to share and distribute things that need to be accessed *as files*, such as VM images, and also as a shared filesystem directly visible within instances. | 15:29 |
*** alonswartz has joined #openstack | 15:31 | |
com1 | Jdarcy, forgive my ignorance on the subject, but do you happen to know what azure uses for its highly available block storage, and how it compares to ebs? | 15:31 |
com1 | jdarcy: i saw your benchmarks on the various cloud providers on your blog (which was excellent btw), and thought you may have some insight | 15:32 |
jdarcy | com1: I'm pretty sure it's natively developed, and I haven't seen any details or comparisons. | 15:32 |
pvo | jdarcy: we have plans to release our vm image caching app which will pull images out of various locations and drop them onto the hosts. | 15:33 |
pvo | may not be a shared fs | 15:33 |
jdarcy | For distributed block storage, you can either go the simple route with *nbd or images within a distributed filesystem, or the more sophisticated route with something like sheepdog or vinzvault. | 15:33 |
*** Predominant has quit IRC | 15:34 | |
*** pvo has quit IRC | 15:34 | |
jdarcy | pvo: Yeah, we're working on something similar at $dayjob, though it's more of a push model than pull so we can deal with "dumb" destination repositories where we have no pull capability. | 15:35 |
com1 | jdarcy: yes, i may do some testing, as i was quite interested from the performance aspect | 15:35 |
jdarcy | pvo: Unfortunately there's some jockeying between you guys and us (Red Hat) over that. | 15:35 |
com1 | jdarcy: why is that? | 15:36 |
jdarcy | com1: Both trying to establish leadership/influence/whatever over the APIs, that kind of thing. | 15:37 |
*** jakedahn has joined #openstack | 15:37 | |
jdarcy | com1: Nothing horrible, just healthy competition. | 15:37 |
*** christiekoehler has joined #openstack | 15:37 | |
com1 | Yes, was about to say it sounds healthy :) | 15:37 |
jdarcy | I only say "unfortunately" because it puts me in a difficult position personally. | 15:38 |
com1 | indeed | 15:38 |
com1 | jdarcy: i imagine at certain *large* companies you might even find that internally between different divisions | 15:39 |
creiht | com1: that never happens... what are you talking about? :) | 15:40 |
jdarcy | com1: Yep. Funniest was when I was at a small company trying to play referee between three IBM divisions. | 15:40 |
*** murkk has quit IRC | 15:40 | |
com1 | creiht: oh...nothing :) | 15:41 |
*** murkk has joined #openstack | 15:41 | |
com1 | jdarcy: politics reigns... slows everybody down | 15:41 |
vish1 | just proposed a merge of massive auth refactor | 15:43 |
vish1 | feedback welcome | 15:43 |
*** alonswartz has quit IRC | 15:44 | |
vish1 | https://code.launchpad.net/~vishvananda/nova/clean-auth/+merge/30431 | 15:44 |
jdarcy | creiht: Where the SAIO instructions say to create /srv/1/node/sdb1, /srv/2/node/sdb2 etc. shouldn't those all be .../sdb1 instead? | 15:45 |
creiht | we use different numbers so it makes it a little easier to debug if there are issues | 15:45 |
com1 | Very interested to see this announcement as i've seen similiar happenings on the telephony side of things with voxeo (standards based, voicexml leader) & their cloud service tropo for which they've open sourced the key parts (moho)-unlike competitors | 15:46 |
creiht | if something is going to sdb2, you know it was on srv/2 for example | 15:46 |
*** dendrobates is now known as dendro-afk | 15:46 | |
*** calavera has quit IRC | 15:46 | |
jdarcy | creight: So "sdb2" is what would show up in logs etc.? Does that last path component really have a direct connection to the device name on a remote node? | 15:48 |
creiht | jdarcy: we do that just for development | 15:48 |
*** hunter has joined #openstack | 15:48 | |
creiht | in a production setup, we wouldn't do that | 15:48 |
jdarcy | I'm just writing a script to do all the mkdirs, and want to embed the proper conventions. | 15:49 |
creiht | k | 15:49 |
*** jakedahn has quit IRC | 15:50 | |
*** jakedahn has joined #openstack | 15:56 | |
*** omar has quit IRC | 15:58 | |
*** jakedahn has quit IRC | 15:58 | |
*** omar has joined #openstack | 15:59 | |
*** blakeyeager has joined #openstack | 16:01 | |
*** pvo has joined #openstack | 16:01 | |
com1 | blakeyeager: hi! another racker I recognise from the cloud! | 16:03 |
blakeyeager | howdy | 16:04 |
blakeyeager | com1? who is that? | 16:04 |
com1 | blakeyeager: greetings...just a rackspace customer/partner! thanks for helping to bring windows (especially r2) to the cloud earlier this year, works superbly! | 16:06 |
blakeyeager | awesome, glad it is working out for you guys, and glad to see our customers/partners getting involved | 16:07 |
*** lbieber has quit IRC | 16:07 | |
*** anotherjesse has quit IRC | 16:09 | |
com1 | blakeyeager: thanks, eaglerly following the new developments, especially as a bit of our strategy relates windows virtualization in managed, and they're all very complementary. | 16:09 |
com1 | I'm especially interested in geographic redundancy (we were in the uk once, have some customers in hk, in dallas ourselves, and see cloud is now in chicago too...so very interested to see what openstack might enable on a geographic level :) | 16:10 |
*** lbieber has joined #openstack | 16:12 | |
*** rajijoom has quit IRC | 16:13 | |
*** jdmaturen has joined #openstack | 16:14 | |
*** hunter has quit IRC | 16:15 | |
*** jdmaturen has quit IRC | 16:15 | |
*** jdmaturen has joined #openstack | 16:15 | |
*** dendro-afk is now known as dendrobates | 16:18 | |
*** ttx has quit IRC | 16:19 | |
*** vish1 has quit IRC | 16:26 | |
*** brianh has joined #openstack | 16:27 | |
*** brianh is now known as Guest46458 | 16:27 | |
*** Guest46458 is now known as RandomBrian | 16:27 | |
creiht | btw... If anyone is in the Texas area, some of the swift devs will be at (and likely giving talks) at pytexas next month | 16:27 |
creiht | http://pytexas.org/ | 16:28 |
*** mufasa_uk has quit IRC | 16:28 | |
*** phildoc has joined #openstack | 16:28 | |
*** jakedahn has joined #openstack | 16:28 | |
*** adrian_otto has joined #openstack | 16:29 | |
*** jdmaturen is now known as sku | 16:30 | |
*** sku is now known as jdmaturen | 16:30 | |
*** NeilNeely has joined #openstack | 16:31 | |
*** bret|awa1 has joined #openstack | 16:31 | |
*** rando has joined #openstack | 16:31 | |
*** zagi1 has joined #openstack | 16:31 | |
com1 | creiht: it's a little way from here, but wouldn't mind visiting (the general) texas area again :) | 16:33 |
*** tgodar_ has joined #openstack | 16:33 | |
*** tgodar has quit IRC | 16:33 | |
*** tgodar_ is now known as tgodar | 16:33 | |
*** jodom has joined #openstack | 16:33 | |
*** jakedahn has quit IRC | 16:34 | |
*** bret|away has quit IRC | 16:34 | |
*** barismetin has quit IRC | 16:36 | |
*** jdmaturen is now known as jdmaturen1 | 16:37 | |
*** sladercn has joined #openstack | 16:37 | |
com1 | Just reading through the swift docs...which are refreshingly well put together. | 16:37 |
*** jdmaturen1 is now known as jdmaturen | 16:37 | |
*** enmand has joined #openstack | 16:42 | |
*** tobym has quit IRC | 16:47 | |
*** com1 has quit IRC | 16:48 | |
*** mtaylor|breakfas is now known as mtaylor | 16:49 | |
*** mindgap has joined #openstack | 16:51 | |
*** devcamcar_ has joined #openstack | 16:51 | |
chmouel | yeah kudo to creiht | 16:52 |
*** com1 has joined #openstack | 16:52 | |
*** tupshin has quit IRC | 16:52 | |
chmouel | and others ... | 16:53 |
*** murkk has quit IRC | 16:54 | |
*** devcamcar_ has quit IRC | 16:54 | |
*** bretpiatt has joined #openstack | 16:54 | |
*** mindgap has quit IRC | 16:58 | |
*** HollyRain has joined #openstack | 17:01 | |
HollyRain | thanks for opening it | 17:02 |
*** Predominant has joined #openstack | 17:03 | |
*** jhalogen has joined #openstack | 17:03 | |
HollyRain | but I believe that Go would be there a better choice | 17:04 |
*** abecc has joined #openstack | 17:04 | |
HollyRain | Go has a lot better performance and a lesser consume of memory, a great concurrency and it's really simple | 17:05 |
creiht | HollyRain: Everyone has their favorite language :) | 17:05 |
HollyRain | creiht: yes, right but I think that since that I saw that was used twisted for openstack | 17:06 |
bretpiatt | HollyRain, thanks for having interest in what we've opened, Go is new and looks like a great language but it doesn't yet have the talent pool that python does in infrastructure orchestration programming | 17:06 |
creiht | And this isn't the place for language wars :) | 17:06 |
*** greenisus has joined #openstack | 17:07 | |
*** jakedahn has joined #openstack | 17:07 | |
*** apocrathia has joined #openstack | 17:14 | |
apocrathia | hey guys. | 17:14 |
bretpiatt | good day | 17:14 |
apocrathia | i had a few questions about the project if anyone has a minute. | 17:14 |
bretpiatt | ask away | 17:15 |
apocrathia | okay. | 17:15 |
apocrathia | i run a small colo, and right now we're using citrix xenserver to do all of out vps's. is virtualization a part of what you're trying to accomplish openstack, or is it just services instead of vms | 17:16 |
apocrathia | we've tried using openqrm in the past to launch a cloud-based service, and it didn't really suit our needs. | 17:16 |
bretpiatt | openstack compute is the orchestration management around the hypervisor | 17:16 |
*** tupshin has joined #openstack | 17:16 | |
apocrathia | so we could provision vms through xenserver, and openstack would manage all of the processing distribution? | 17:17 |
bretpiatt | you would provision the vms through the openstack apis, it will deploy them to the host machines and start them up | 17:18 |
apocrathia | this really seems like a promising project, i'm just thinking ahead into migration from what we have. | 17:18 |
com1 | bretpiatt: do you have plans to be able to support multiple hypervisors? | 17:18 |
pvo | com1: sure do | 17:18 |
apocrathia | okay, so is there a management node required, or does the entire system function as one. like, say i have 10 hardware nodes, does one need to be a management console? | 17:19 |
pvo | right know it supports KVM but work is already being done to support XenServer | 17:19 |
bretpiatt | thanks, we're really excited about it, for a starting point take a look through http://nova.openstack.org and http://wiki.openstack.org to get more background on what it does and how it works | 17:19 |
bretpiatt | com1, yes, compute will support many hypervisors as while they all provide vms, they also service workloads differently | 17:19 |
bretpiatt | apocrathia, it is designed to have management and orchestration nodes, right now it isn't a single server simple to setup orchestration system, we started with solving the "run at scale" problem, we're going to work on ease of deployment over time | 17:21 |
com1 | pvo: Excellent, and i know it's probably far out of the ballpark (and probably not *exactly* the spirit of the project), but hyper-v too? | 17:21 |
bretpiatt | com1, we'd love to have hyper-v support, it needs somebody committed to making the code from our abstraction layer through to hyper-v supported and maintained | 17:22 |
pvo | ^^ +1 | 17:23 |
uvirtbot | pvo: Error: "^" is not a valid command. | 17:23 |
pvo | oh yes it is | 17:23 |
soulresin | heh | 17:23 |
apocrathia | bretpiatt, so with 2 different types of nodes, what is the failover for such a system. right now, I have 10 hardware nodes all running xenserver independently. once this project gets released, how could I go about deploying it to my current hardware. Assuming there is support for the xen hypervisor built in at that point, migration should be simple, but what about how openstack is going to use my actual hardware. will I simply install a stock version | 17:25 |
apocrathia | ubuntu server or centos, compile openstack, and add it into the cloud , letting openstack take care of the rest? | 17:25 |
com1 | bretpiatt: i've been doing a lot of work, almost exclusively in the windows virtualization domain (admittedly on a small scale, but close to the hypervisor) and such a layer would be a dream | 17:26 |
com1 | i'd be seriously considering helping to build it myself | 17:26 |
*** johnbergoon_ has joined #openstack | 17:26 | |
*** kim0 has quit IRC | 17:27 | |
*** jakedahn has quit IRC | 17:28 | |
zagi1 | apocrathia, what are you trying to accomplish by adding OpenStack to your 10 hardware nodes :)? | 17:30 |
apocrathia | right now, our xenserver setup isn't exactly high available, we're looking for a foss means of implementing a highly available virtualization platform, be it through provisioning virtual services in a cloud, or virtual machines | 17:31 |
zagi1 | from the sound of it you're not using central storage, and so what you have wtih your Xen systems is a VPS platform, OpenStack is meant more for building a Cloud architecture | 17:31 |
*** pvo has quit IRC | 17:31 | |
WaterDog | The difference between a cloud and a vps platform is central storage? =D | 17:32 |
zagi1 | no but its certainly important | 17:32 |
apocrathia | we're trying to implement an iscsi powered san for central storage right now | 17:32 |
*** Predominant has quit IRC | 17:32 | |
WaterDog | hmm.. not sure I agree there, not to mince words about what defines one vs the other though | 17:32 |
apocrathia | essentially, local storage should be used for swap imo. | 17:33 |
zagi1 | you may be right, central storage may not be a requirement, it would just suck to have the vm go down and lose all of your info ;) | 17:33 |
bretpiatt | The difference is more about being able to manage the instances and move images around, you don't have to have central storage to do that, you "might" need central storage for live migration, you do need it for live failover | 17:33 |
WaterDog | More ram should be used for swap (IMO) | 17:33 |
zagi1 | you could always provision another instance and rebuild everything into it, but that would be painful for a lot of applications | 17:34 |
WaterDog | Well I agree that it's a good idea to use central storage where appropriate my only sticking point was the assertion that a 'cloud' without central storage is a VPS | 17:34 |
com1 | zagi1: or use a periodic snapshot, automatic restore scenario for less than live migration | 17:35 |
bretpiatt | zagi1, the host machine can go down but it doesn't have to have all of the instances vanish, how durable you want the vms to be is part of the configuration | 17:35 |
WaterDog | but anyway like i said, words meet mincing... | 17:35 |
zagi1 | com1: so long as the snapshot is not stored on the same local hypervisor, but then with local storage you have a long restore period if you have a 50GB image | 17:35 |
apocrathia | we're really trying to set up something that we can use all of our resources as one, with the ability for virtual machines and services to process across multiple systems. | 17:35 |
zagi1 | bretpiatt: if you are using local storage, than any hypervisor failure will take all vms offline with it | 17:36 |
zagi1 | apocrathia: that is not possible :) | 17:36 |
WaterDog | Well, I don't think you'll find a 'cloud' solution that lets a single VM run on 2 nodes simultaneously | 17:36 |
zagi1 | you are still confined to the physical resources of 1 hypervisor | 17:36 |
WaterDog | you could use load balancing to spread the load to multiple copies of the same VM | 17:37 |
zagi1 | unless your DB is inside the VM, and then you need to reconcile that :) | 17:37 |
apocrathia | it's more about h.a. than load balancing. | 17:37 |
com1 | zagi1: yes, it's quite long, but if it's automatic the pain is reduced with substantially less cost | 17:37 |
WaterDog | zagi1: you could cluster the DBs on the vms across nodes | 17:38 |
WaterDog | might be a nightmare | 17:38 |
WaterDog | but you could do it | 17:38 |
adrian_otto | apocrathia: You might want to have a look at Remus http://bit.ly/2JTEA0 | 17:38 |
adrian_otto | and use a shared storage solution for the data | 17:39 |
*** gein has joined #openstack | 17:40 | |
*** novas0x2a has quit IRC | 17:42 | |
com1 | I have a few thoughts, what would be your guidelines performance-wise for using swift as shared storage? | 17:45 |
*** bretpiatt has quit IRC | 17:45 | |
*** sladercn has left #openstack | 17:46 | |
*** tobym has joined #openstack | 17:47 | |
*** dendrobates is now known as dendro-afk | 17:47 | |
com1 | In the same DC it seems the throughput isn't bad at all...I haven't tested it extensively as data gets hit harder? Are there any interesting use cases where its limits have really been stretched, but it still holds up for the application? | 17:48 |
*** HollyRain has left #openstack | 17:48 | |
*** devcamcar has joined #openstack | 17:56 | |
*** bretpiatt has joined #openstack | 17:57 | |
*** hunter has joined #openstack | 17:57 | |
*** hunter has quit IRC | 17:57 | |
*** anotherjesse has joined #openstack | 18:01 | |
*** hunter has joined #openstack | 18:03 | |
*** bobo_b has joined #openstack | 18:03 | |
*** tomjones has joined #openstack | 18:03 | |
*** pvo has joined #openstack | 18:10 | |
*** pvo has joined #openstack | 18:10 | |
*** rnirmal has joined #openstack | 18:13 | |
*** gustavomzw has quit IRC | 18:14 | |
*** gustavomzw has joined #openstack | 18:15 | |
*** jeromatron has left #openstack | 18:15 | |
*** vish1 has joined #openstack | 18:15 | |
*** bretpiatt has quit IRC | 18:16 | |
*** bretpiatt has joined #openstack | 18:17 | |
devcamcar | hoyoo | 18:18 |
devcamcar | bretpiatt: just pinged you in email about the python lib | 18:18 |
anotherjesse | devcamcar: devin - welcome to the IRCs ... for when you are at nasa http://webchat.freenode.net/ works | 18:19 |
devcamcar | hah, really? no irc? | 18:20 |
devcamcar | hiiiiilarious. | 18:20 |
anotherjesse | devcamcar: surprised? | 18:20 |
devcamcar | anotherjesse: this is me not looking surprised | 18:20 |
_0x44 | devcamcar: Your not surprised face looks a lot like your surprised face. | 18:21 |
devcamcar | _0x44: yes, its quite subtle | 18:21 |
adrian_otto | com1: i have tested swift extensively from a performance perspective. You will notice writes will slow down then you have a very large number of objects. From my experience it starts to matter in the 1 million object per container range. The workaround there is simply to distribute your objects into more containers. | 18:23 |
adrian_otto | you will also notice that with ridiculous number of objects in your containers that replication may take longer. | 18:24 |
adrian_otto | but I did cram something like 800 million objects into a single account, over a span of a couple of days. | 18:24 |
*** murkk has joined #openstack | 18:24 | |
adrian_otto | actually, no, it was more than that. I crossed the two billion object mark. | 18:25 |
adrian_otto | also, the more write concurrency you place on a single container, the slower the request processing will get. | 18:28 |
adrian_otto | I found that concurrencies of about 10 concurrent connections constantly doing writes was about the performance sweet spot | 18:28 |
adrian_otto | but I I sprayed the writes across numerous containers, I was able to scale out the write performance horizontally | 18:29 |
adrian_otto | I hope that helps | 18:29 |
*** dendro-afk is now known as dendrobates | 18:29 | |
*** rdw has quit IRC | 18:29 | |
g0rdy | adrian_otto: do you have that published anywhere? I'd like to tote that info to some decision makers - | 18:31 |
adrian_otto | at the time it was an internal study, but I'll see what I have that's suitable for publication | 18:32 |
devcamcar | adrian_otto: the write concurrency can be limited by sqlite internals. i am dealing with similar issues as we prepare to roll out for nebula | 18:34 |
g0rdy | thank you, performance data is beneficial :) | 18:35 |
devcamcar | adrian_otto: our internal test case is for a customer that will use a half billion images | 18:35 |
adrian_otto | devcamcar: images probably are mostly read (not mostly write) correct? In that case, you probably don't even need to worry a whole lot about spreading those into a bunch of different containers. | 18:38 |
devcamcar | adrian_otto: yes, it is somewhat academic, but in our case the images are the result of a largescale processing batch | 18:38 |
adrian_otto | swift also employs a read cache, so your read concurrency is not really much of a concern. Where you could get snagged is trying to focus way too much write concurrency into a single container. | 18:39 |
devcamcar | sometimes there are issues with the batch, color correction, etc. and they are run multiple times sometimes | 18:39 |
devcamcar | so being able to write out and test and throw away 30T of data and try again is important | 18:39 |
devcamcar | but yes, the solution is to write to as may containers as you reasonably can | 18:39 |
adrian_otto | ok, so if you plan to batch them, and want batch processing as fast as possible, then it would not hurt to find a way to divide those objects such that each container holds about a million of them. | 18:40 |
*** novas0x2a has joined #openstack | 18:40 | |
anotherjesse | devcamcar: purging an entire container that isn't empty would probably help with that | 18:40 |
devcamcar | adrian_otto: sounds reasonable | 18:40 |
adrian_otto | anotherjesse: I don't think that's currently supported | 18:41 |
anotherjesse | adrian_otto: yeppers | 18:41 |
*** bretpiatt has quit IRC | 18:42 | |
adrian_otto | anotherjesse: but I could think of some ways to adapt swift so that it could be much better at that particular trick. | 18:42 |
*** bretpiatt has joined #openstack | 18:43 | |
adrian_otto | adding the concept of hierarchy would help, but tI expect that would come with some costs as well. Maybe using a filesystem per container would be another way to do that. You could drop the whole filesystem when you want to do a "delete this full container" operation. | 18:44 |
adrian_otto | (along with the related database files) | 18:45 |
anotherjesse | adrian_otto / devcamcar I don't think the mars data sets need the LIST funcitonality - eg, we would sacrifice LIST against 1B objects .. | 18:45 |
adrian_otto | that would probably need to be combined with a filesystem grow/shrink capability too | 18:45 |
devcamcar | yes, we've discussed adding an "opaque" flag to a container that disables listing | 18:46 |
adrian_otto | devcamcar: what benefits does that offer? | 18:47 |
adrian_otto | would you still need a SQlite file for each container then? | 18:48 |
devcamcar | adrian_otto: in our case there's no need to ever list the objects once they are there. the data is stored at zoom levels as tiles, similar to google maps, etc, so the data is accessed algorithimically | 18:48 |
devcamcar | good question, i assume the sqlite data needs to be there for auditing | 18:48 |
adrian_otto | I can certainly see that not _using_ list has benefits, but I'm wondering what additional benefit you'd see by disabling it. | 18:49 |
devcamcar | adrian_otto: metaphorically it'd be like putting the safety on a gun :) | 18:50 |
adrian_otto | fair enough. | 18:50 |
adrian_otto | but it will only show you 10,00 results at a time anyway | 18:50 |
notmyname | adrian_otto: I'm curious what you mean by a read cache in swift? can you tell me more about that? | 18:50 |
adrian_otto | so it's essentially got a safety already | 18:50 |
adrian_otto | notmyname: it uses memcached so subsequent readers of the same object can get it from cache, right? | 18:51 |
adrian_otto | I have not verified that by inspecting the source, but that's the only thing I can think of that would explain the performance behavior I observed while testing it. | 18:52 |
adrian_otto | and I do see a memcached here running on each node of my swift install | 18:52 |
notmyname | there is no read cache in swift | 18:52 |
notmyname | also, replication isn't really dependent on container size | 18:53 |
*** xtoddx has joined #openstack | 18:53 | |
devcamcar | xtoddx: hoyoo | 18:53 |
xtoddx | devcamcar: yoyo | 18:54 |
notmyname | PUTs will slow down on large containers. that is a function of sqlite. recommended practice is to go wide, like you said | 18:54 |
notmyname | we have investigated sqlite's recent WAL feature and are hoping that it will give some good performance gains | 18:55 |
notmyname | devcamcar: there is no advantage in the current design to disable listings. simply don't do a GET on containers, and you don't do listings | 18:56 |
redbo | The performance would be very different if you got rid of container listings. But then it's just a DHT that handles large objects hopefully well. | 18:56 |
notmyname | I mean, I suppose you could rip out the whole container part of the system... | 18:56 |
redbo | yeah, if you could flag a container not to insert any data | 18:57 |
notmyname | adrian_otto: memcache is used for things like auth credentials and account/container existence. | 18:57 |
adrian_otto | I see, thanks for the clarity. | 18:58 |
adrian_otto | are you using the Linux page cache at all? I suppose that might also explain my observations. | 18:59 |
creiht | adrian_otto: which observations are you referring to, then I could probably tell you what you are observing? :) | 18:59 |
creiht | are you talking about read throughput from multiple GETs to the same object? | 19:00 |
adrian_otto | initial reads are not as fast as subsequent reads, and the speed of the subsequent reads, while substantially faster, did not appear to be sensitive to system abuse like writes were sensitive to it. | 19:00 |
devcamcar | notmyname: cool, sounds like we won't need to worry about disabling listings then | 19:01 |
adrian_otto | when doing a GET on an object | 19:01 |
uvirtbot | New bug: #607912 in swift "swift-auth-create-account lets you create multiple users with same params but only single token exists causing problems on reset" [Undecided,New] https://launchpad.net/bugs/607912 | 19:01 |
redbo | oh yeah | 19:01 |
adrian_otto | also, when doing subsequent GET's on an object the performance was very consistent | 19:01 |
*** soulresin has left #openstack | 19:02 | |
devcamcar | creiht: greetings! thanks again for the help last week getting our test lab up to speed. | 19:02 |
creiht | adrian_otto: that is either due to linux's built in cache, or the caching on the raid controller | 19:02 |
creiht | devcamcar: hey man! How goes the testing? | 19:02 |
notmyname | devcamcar: if you want to disable listings, you can take out the @public decorator in the proxy server that is on the ContainerController GET method | 19:02 |
adrian_otto | where the performance of a first read on a file varied at about the same standard deviation that writes did. | 19:02 |
creiht | notmyname: It is more than just disabling listing, but disabling contianer updates on writes, etc. | 19:02 |
notmyname | well that would be step 2 :-) | 19:03 |
creiht | hehe | 19:03 |
devcamcar | creiht: we are in final phases of hardware reqs gathering for ordering a few columns to run cloud files on nebula. do you guys have recommendations about # of spindles per box? | 19:03 |
creiht | devcamcar: we talked a little bit about it friday | 19:03 |
creiht | We are currently doing 24 per box | 19:03 |
devcamcar | creiht: we're looking at 30 right now | 19:03 |
creiht | nice | 19:03 |
creiht | the only downside with the more spindles per server, is that if you have a node go down, that is more data that has to be replicated | 19:04 |
*** pvo has quit IRC | 19:04 | |
devcamcar | creiht: we're looking at rolling out about 800T in about 4-6 weeks | 19:04 |
*** bretpiatt has quit IRC | 19:05 | |
notmyname | but that is a concern to be weighed against the size of the network connected to the storage node | 19:05 |
creiht | I'm not sure what type of upper bound that would be | 19:05 |
creiht | yeah... and since you guys are on all 10g that is much less of a concern | 19:05 |
redbo | maaan... I want all 10G networking. | 19:06 |
*** ChanServ sets mode: +v dendrobates | 19:06 | |
creiht | :) | 19:06 |
devcamcar | redbo: as nasa you get 10G networking but no irc :) | 19:06 |
anotherjesse | redbo: they cost 10x as much ... and break ... be careful what you wish for | 19:06 |
creiht | hah | 19:06 |
devcamcar | all about trade offs... and strange ones | 19:06 |
redbo | with 10g networking I'll make my own irc network | 19:06 |
pandemicsyn | even at 10G replicating a 30 drive box (say with 2T drives) is gonna take a bit | 19:07 |
devcamcar | pandemiscsyn: yes 2T drives all around | 19:07 |
*** hunter has quit IRC | 19:07 | |
creiht | devcamcar: btw pandemicsyn is one of our ops guys | 19:07 |
devcamcar | pandemicsyn: i'm a dev on the nebula team, nice to meet you <3 | 19:08 |
creiht | devcamcar: so about 14 boxes? | 19:08 |
pandemicsyn | devcamcar: o/ , nice to meet ya | 19:08 |
anotherjesse | creiht: we have a few different clusters we are testing on - the cluster I look forward to testing is 24 - 12 2T disks, 96GB ram, 24 cores | 19:09 |
creiht | nice | 19:10 |
anotherjesse | creiht: but that cluster is not going to be pure swift - a mix of swift and nova | 19:10 |
anotherjesse | but first a pure swift just for testing | 19:10 |
devcamcar | creiht / anotherjesse: that is what we'll be testing on next, but won't be what we're rolling out on in about 2 months | 19:10 |
creiht | ?"? | 19:10 |
uvirtbot | New bug: #607920 in nova "Basic quota system" [Undecided,New] https://launchpad.net/bugs/607920 | 19:10 |
devcamcar | creiht: its new hardware, we're going to load it up with no raid, and purely use it for swift testing, then tear it down and rebuild it as compute/volumes | 19:11 |
*** novas0x2a has quit IRC | 19:11 | |
devcamcar | don't think we'll have too much trouble hitting 5k writes/sec with that set up | 19:11 |
creiht | sorry... accidently hit keys while transporting the laptop | 19:11 |
creiht | So it *feels* a little top heavy on the device to node ratio for how many nodes you would have | 19:12 |
anotherjesse | devcamcar: have we tested read? | 19:12 |
creiht | but that is what testing for right? :) | 19:12 |
creiht | anotherjesse: your current dev cluster has raid, and we saw some perf issues with it | 19:13 |
creiht | It would be nice to un-raid that and do testing to just be academically complete :) | 19:13 |
*** dendrobates is now known as dendro-afk | 19:14 | |
devcamcar | creiht: we'd like to and may get to still | 19:14 |
creiht | cool | 19:15 |
devcamcar | anotherjesse: we haven't scientifically tested read yet | 19:15 |
*** tempspace has joined #openstack | 19:15 | |
creiht | devcamcar: it will also be interesting to see what the performance looks like when you guys get your cluster set up | 19:18 |
creiht | it may be good enough that we don't have to make any changes to the container :) | 19:18 |
*** kim0 has joined #openstack | 19:19 | |
*** jakedahn has joined #openstack | 19:19 | |
*** londo has joined #openstack | 19:19 | |
*** tempspace has quit IRC | 19:19 | |
redbo | Stand back... I'm going to use science. | 19:20 |
jdarcy | Did webob move some stuff to webob.request between 0.9.6 and 0.9.8? | 19:23 |
creiht | jdarcy: not sure | 19:25 |
devcamcar | creiht: we'll keep you guys posted on our progress | 19:26 |
jdarcy | "from webob.request import Request" (in auth.py) gets "ImportError: No module named request" | 19:26 |
jdarcy | Looking in /usr/lib/python2.6/site-packages/webob I see no signs of a request sub-module. | 19:26 |
*** antzs has joined #openstack | 19:27 | |
creiht | jdarcy: http://pythonpaste.org/webob/news.html#id3 | 19:27 |
creiht | Split __init__.py into four modules: request, response, descriptors and datetime_utils. | 19:27 |
*** novas0x2a has joined #openstack | 19:27 | |
*** antzs has quit IRC | 19:28 | |
creiht | jdarcy: I highly recommend an updated version of webob for swift | 19:28 |
jdarcy | Yep, I see it now. Working around... | 19:28 |
*** chuhnk has quit IRC | 19:29 | |
devcamcar | we're going to start work soon on integrating the rackspace python api into nova and working to obsolete euca2ools | 19:29 |
devcamcar | and create our own suite of nova-* command line tools | 19:30 |
jdarcy | Working now. Looks like I'll need to document that in the RHEL instructions. Thanks, creiht. | 19:30 |
*** rdw has joined #openstack | 19:30 | |
*** dendro-afk is now known as dendrobates | 19:31 | |
creiht | For those interested in swift, the swift docs have been updated to now include a short getting started section and an architectural overview | 19:32 |
creiht | rdw: welcome! :) | 19:32 |
anotherjesse | creiht: how often does http://swift.openstack.org/ get rebuilt? | 19:32 |
creiht | anotherjesse: docs are rebuilt on every merge | 19:32 |
creiht | via hudson | 19:33 |
* creiht thanks mtaylor for setting all that up | 19:33 | |
rdw | hi :) | 19:33 |
*** abecc has quit IRC | 19:33 | |
*** devcamcar has quit IRC | 19:36 | |
*** RandalSchwartz has joined #openstack | 19:37 | |
* RandalSchwartz waves at the channel | 19:38 | |
RandalSchwartz | Rick Clark confirmed for FLOSS Weekly interview tomorrow. | 19:38 |
anotherjesse | there is a bug with | 19:40 |
rdw | ....newlines in irc | 19:40 |
rdw | :P | 19:40 |
anotherjesse | yeah - sorry | 19:41 |
creiht | mtaylor: when you are around, and have some time, I would like to talk about the swift.egg-info dir | 19:41 |
*** Predominant has joined #openstack | 19:42 | |
*** ewan_ has joined #openstack | 19:43 | |
*** joshuamckenty has joined #openstack | 19:44 | |
*** dizzl_ has joined #openstack | 19:45 | |
* jaypipes waves back at RandalSchwartz | 19:50 | |
*** LiamMac has joined #openstack | 19:54 | |
*** sophiap has quit IRC | 19:56 | |
*** devcamcar has joined #openstack | 19:57 | |
*** cglee has joined #openstack | 20:07 | |
*** tsal has joined #openstack | 20:10 | |
tsal | Howdy. | 20:10 |
notmyname | hi | 20:10 |
*** pvo has joined #openstack | 20:10 | |
*** pvo has joined #openstack | 20:10 | |
tsal | Mainly here to check things out, as I am a python developer and a systems engineer looking to push a cloud solution at work. :) | 20:12 |
*** devcamcar has quit IRC | 20:12 | |
notmyname | tsal: let us know if you have any questions, from tech details to points to mention to your boss | 20:13 |
*** dendrobates is now known as dendro-afk | 20:14 | |
*** mshadle has joined #openstack | 20:14 | |
tsal | Will do. I have done some twisted development so I may see about contributing, too. | 20:14 |
*** murkk has quit IRC | 20:15 | |
notmyname | awesome | 20:15 |
tsal | Just don't know how often I'll be on irc during the day as it's blocked. I'm using my android for this as it is. | 20:17 |
*** Predominant has quit IRC | 20:17 | |
anotherjesse | tsal: try webchat.freenode.net | 20:18 |
notmyname | have you considered a job at nasa? ;-) | 20:18 |
tsal | Lol, yes, but no degree so I'll have a hard time | 20:19 |
scotticus | get a slice, ssh in, irc from there :) | 20:20 |
tsal | Webchat is blocked too :/ | 20:20 |
notmyname | we were talking earlier about how nasa has a great netowrk but doesn't allow irc access | 20:20 |
*** jakedahn has quit IRC | 20:20 | |
joshuamckenty | try mibbit | 20:20 |
joshuamckenty | used to work | 20:20 |
tsal | Ah, this client is functional enough for me. | 20:21 |
creiht | tsal: ConnectBot? | 20:21 |
creiht | well.. maybe not... I use that to log into a slice and then use screen :) | 20:21 |
*** devcamcar has joined #openstack | 20:22 | |
*** abecc has joined #openstack | 20:22 | |
tsal | Yaairc | 20:22 |
*** Predominant has joined #openstack | 20:22 | |
mtaylor | creiht: yes. I'm not really here right now- but I'll grab you when I actually am, k? | 20:23 |
sander | Whats the major advantages between openstack and eucalyptus ? | 20:23 |
creiht | mtaylor: sounds good... thanks! | 20:23 |
aliguori | are there any demo appliances for openstack? | 20:23 |
aliguori | like pre-canned virtual machines that can be used to build an openstack cluster on a single box? | 20:24 |
*** vish1 has quit IRC | 20:24 | |
aliguori | ovirt did something like that and it was pretty useful | 20:24 |
tsal | I have several irc capable hosts I can use but external ssh is blocked too. I can probably get around that but it's not worth my job. ;) | 20:24 |
creiht | hah | 20:24 |
Thweetch | Im pretty sure someone who has deployed it onto a cloud server could backup the image export it to cloudfiles and offer it as a download | 20:24 |
PiotrSikora | mtaylor | 20:25 |
PiotrSikora | erm.. | 20:25 |
PiotrSikora | that should be PM ;P | 20:25 |
creiht | aliguori: I don't think there is anything official yet, but that would be very nice indeed | 20:25 |
Thweetch | If I get the chance to roll it out tonight and noone has yet, ill dump a cs image and load it onto the wiki | 20:26 |
*** tsal is now known as tsal-mobile | 20:26 | |
aliguori | that would be neat | 20:27 |
*** sophiap has joined #openstack | 20:28 | |
*** scrllock_ has joined #openstack | 20:28 | |
*** anotherjesse has quit IRC | 20:29 | |
aliguori | are there any architectural docs available? I've found some very high level docs, and lots of very low level docs, but I'm trying to understand how some of the basic workflows operate. For instance, does nova implement persistent or empheral storage, is persistent storage stored in the Object Store, how are nodes chosen for provision, etc.? | 20:29 |
*** enmand has quit IRC | 20:30 | |
*** dendro-afk is now known as dendrobates | 20:32 | |
*** ewan_ has quit IRC | 20:34 | |
devcamcar | aliguori: check out http://wiki.openstack.org/Overview | 20:39 |
devcamcar | it may be too high level but hopefully its a good start | 20:39 |
aliguori | yeah | 20:40 |
aliguori | devcamcar, thanks, I've seen that | 20:40 |
*** jakedahn has joined #openstack | 20:40 | |
aliguori | devcamcar, so does nova have a persistent storage model? | 20:40 |
*** binaryWarrior has joined #openstack | 20:41 | |
aliguori | i'm assuming the use of an image cache implies ephemeral storage | 20:41 |
aliguori | IOW, is there an EBS equivalent | 20:41 |
devcamcar | aliguori: yes, nova has persistent storage, similar to amazon EBS | 20:41 |
aliguori | devcamcar, is Nova integrated with Object Storage yet or is that a separate project right now? | 20:42 |
devcamcar | aliguori: nova was originally built with its own object store, though it is very simple | 20:42 |
aliguori | yeah, I was just looking at it :-) | 20:42 |
devcamcar | it is currently used internally by nova to store images/kernels | 20:42 |
devcamcar | and some other bits | 20:42 |
devcamcar | we are debating moving all of that to cloudfiles, though we don't force a hard dependency on cloudfiles | 20:43 |
devcamcar | so we'll probably support both in the future | 20:43 |
dhdenny | aliguori: this link may also be helpful http://nova.openstack.org/volume.html, if you haven't seen it already | 20:43 |
*** hunter has joined #openstack | 20:43 | |
aliguori | devcamcar, okay, last question :-) when selecting a cluster controller to provision to, is it basically just round robin? | 20:43 |
aliguori | or is there something more sophisticated? | 20:43 |
aliguori | dhdenny, ah, aoe is an interesting choice | 20:44 |
*** vish1 has joined #openstack | 20:44 | |
aliguori | dhdenny, so does that fundamentally limit the size of a cluster to a single subnet? | 20:45 |
dhdenny | aliguori: i should note that i'm just an intrigued rackspace employee, who has no direct involvement in the project. :) i'll have to defer to the experts | 20:45 |
aliguori | heh, okay :-) | 20:46 |
devcamcar | aliguori: its pretty simple right now but we're going to make it more sophisticated in the future and allow compute nodes to be dedicated to certain instance types (so you could have a node of tiny only instances for example) | 20:46 |
devcamcar | once a node is at capacity it starts ignoring run requests from the queue | 20:47 |
justinsb | Dumb KVM question: why does the network device appear as eth3 in the guest (and not eth0)? Does it always come as eth3? How does this get determined? | 20:47 |
aliguori | justinsb, that's your distro | 20:47 |
jdarcy | Everything seems set up OK, running functest now. | 20:47 |
creiht | jdarcy: woot | 20:48 |
aliguori | a lot of distros remember network devices and give them persistent names. so if the image was installed with two network devices, and then you launch the image with just one (that happens to have a different mac from the previous two), it'll be eth2 | 20:48 |
justinsb | aliguori: Running Ubuntu Lucid. Trying to figure out how to handle this for templating /etc/network/interfaces on initial build. Any idea where the mapping is stored or how I could override it? | 20:48 |
*** vish1 has quit IRC | 20:48 | |
aliguori | justinsb, /etc/udev/rules.d/70-persistent-net.rules | 20:50 |
aliguori | if you just rm -f that file, you'll get eth0 next time you boot | 20:50 |
justinsb | aliguori: Sweet - thank you! Do you know if that's the same for most distros? | 20:51 |
aliguori | justinsb, the rule name is sometimes different and some distros depending on their age don't do it | 20:51 |
*** murkk has joined #openstack | 20:51 | |
justinsb | aliguori: Thanks! I guess I'll just delete it if it's there for now... | 20:52 |
*** murkk has quit IRC | 20:52 | |
jdarcy | Is there a description somewhere of what this "EBS-like" storage in Nova is like? | 20:52 |
*** jodom has quit IRC | 20:52 | |
*** murkk has joined #openstack | 20:52 | |
*** xfrogman5 has joined #openstack | 20:53 | |
jdarcy | BTW, hi Anthony. You probably don't remember me, but RicW introduced us when you were visiting here at RH Westford. | 20:53 |
*** jonesy has quit IRC | 20:53 | |
aliguori | jdarcy, yup, I remember you :-) | 20:53 |
devcamcar | justinb: for ubuntu i recommend always killing 70-persistent-net.rules before bundling | 20:53 |
devcamcar | makes life simpler | 20:53 |
*** vish1 has joined #openstack | 20:54 | |
aliguori | jdarcy, how's it going? | 20:54 |
jdarcy | aliguori: Not bad. I think we'll see some acceleration in the cloud-storage area shortly. | 20:54 |
*** toddmorey has joined #openstack | 20:55 | |
*** apocrathia has quit IRC | 20:55 | |
jdarcy | Meanwhile, I'm getting Swift to run on RHEL6. Fun. | 20:55 |
aliguori | yeah, I'm very interested in these storage systems.. | 20:56 |
*** xfrogman5 has quit IRC | 20:56 | |
*** rando has quit IRC | 20:59 | |
*** devcamcar has quit IRC | 20:59 | |
*** cglee has quit IRC | 20:59 | |
*** vish1 has quit IRC | 21:00 | |
tsal-mobile | jdarcy, since that is our likely target platform, do you mind emailing me any gotchas you run into? | 21:00 |
*** devcamcar has joined #openstack | 21:00 | |
jdarcy | tsal-mobile: Check out http://wiki.openstack.org/RhelInstructions for some tips. It's not linked anywhere else yet, but it might be helpful. | 21:01 |
tsal-mobile | Awesome, thanks! | 21:01 |
*** justinsheehy has quit IRC | 21:01 | |
tsal-mobile | Gotta run! See you all around! | 21:05 |
*** tsal-mobile has quit IRC | 21:05 | |
*** oz` has joined #openstack | 21:05 | |
*** Phil_FL has quit IRC | 21:07 | |
*** devcamcar has quit IRC | 21:07 | |
jdarcy | How long should these functional tests take to run? | 21:10 |
creiht | jdarcy: not very long | 21:11 |
creiht | have you seen any . ? | 21:11 |
creiht | (it runs like a unit test) | 21:11 |
*** vish1 has joined #openstack | 21:12 | |
jdarcy | Yeah, about a hundred dots so far. | 21:12 |
joshuamckenty | no 'E' ? | 21:12 |
jdarcy | No 'E' but the machines all seem idle. | 21:12 |
creiht | hrm | 21:12 |
joshuamckenty | there's a lot of waiting | 21:12 |
* creiht goes to run the functests | 21:13 | |
*** sophiap has quit IRC | 21:13 | |
*** joshuamckenty has quit IRC | 21:14 | |
creiht | jdarcy: you might check either /var/log/syslog or /var/log/messages to see if there are any errors | 21:14 |
*** pvo has quit IRC | 21:15 | |
creiht | I just ran them on my vm: 141 tests in 83.485s | 21:15 |
jdarcy | I just got a bunch of rsync errors in /var/log/messages. Hm. | 21:15 |
creiht | hrm | 21:16 |
creiht | I don't think the func tests directly test replication | 21:17 |
*** devcamcar has joined #openstack | 21:17 | |
*** ewan_ has joined #openstack | 21:18 | |
creiht | jdarcy: the unit tests all pass right? | 21:19 |
*** devcamcar has quit IRC | 21:19 | |
*** pvo has joined #openstack | 21:19 | |
*** hunter has quit IRC | 21:19 | |
*** devcamcar has joined #openstack | 21:20 | |
*** jtdowney has joined #openstack | 21:21 | |
jdarcy | Let me re-check. | 21:21 |
jdarcy | Ran 289 tests in 13.929s | 21:21 |
jdarcy | Looks OK at that level. | 21:21 |
creiht | hrm | 21:22 |
jdarcy | Is there any way to make the tests more verbose, see where it's bogging down? | 21:22 |
* creiht looks | 21:22 | |
ghchinoy | Saw that tweet about using http://cappuccino.org for the web control panel - neat | 21:23 |
chmouel | nosetest -v ? | 21:23 |
chmouel | nosetests | 21:23 |
glange | jdarcy: you ran 289 tests in 14 seconds? | 21:23 |
jdarcy | That's what it said. | 21:23 |
glange | did it actually take 14 seconds? | 21:24 |
creiht | jsm: yeah run nosetests -v from the tests/functional dir | 21:24 |
jdarcy | Seemed about right. | 21:24 |
*** sophiap has joined #openstack | 21:24 | |
creiht | it will tell you at least what test it hangs on | 21:24 |
glange | and that's not fast enough ? | 21:24 |
creiht | glange: the functional tests hang at the end | 21:24 |
creiht | the above was for the unit tests | 21:24 |
glange | oh | 21:24 |
* creiht blames any functests issues on glange :) | 21:25 | |
* glange dodges | 21:25 | |
*** sophiap__ has joined #openstack | 21:29 | |
*** devcamcar has quit IRC | 21:29 | |
*** ewan_ has quit IRC | 21:29 | |
*** zagi1 has quit IRC | 21:30 | |
justinsb | aliguori, devcamcar: Thanks for your help. Networking config now works in my experimental branch http://bazaar.launchpad.net/~justin-fathomdb/nova/raw-disk-images/revision/154 | 21:30 |
*** johnbergoon_ has quit IRC | 21:30 | |
aliguori | np | 21:30 |
*** pengwynn has quit IRC | 21:30 | |
*** devcamcar has joined #openstack | 21:31 | |
*** sophiap has quit IRC | 21:32 | |
*** devcamcar has quit IRC | 21:32 | |
*** sophiap__ is now known as sophiap | 21:32 | |
*** devcamcar has joined #openstack | 21:33 | |
*** justinsheehy has joined #openstack | 21:35 | |
*** devcamcar has quit IRC | 21:35 | |
*** arcane has quit IRC | 21:35 | |
*** jsm has quit IRC | 21:36 | |
jdarcy | nosetest -v is making progress, but slowly. I have to head home; I'll get back to this tonight/tomorrow. | 21:36 |
gholt | If you're getting rsync errors in the logs, I wonder if that means you're running all the daemons all at once and the poor box just can't keep up? | 21:37 |
creiht | possible | 21:38 |
*** justinsheehy has quit IRC | 21:39 | |
jdarcy | gholt: I am seeing a pretty high context-switch rate on the host. I'll just let this run and see what happens. | 21:39 |
*** jakedahn has quit IRC | 21:40 | |
*** justinsheehy has joined #openstack | 21:42 | |
*** Thweetch has quit IRC | 21:43 | |
*** devcamcar has joined #openstack | 21:44 | |
*** seshupasam has joined #openstack | 21:45 | |
*** arcane has joined #openstack | 21:45 | |
*** seshupasam has left #openstack | 21:48 | |
*** devcamcar has quit IRC | 21:48 | |
*** devcamcar has joined #openstack | 21:49 | |
*** LiamMac has quit IRC | 21:51 | |
*** devcamcar has quit IRC | 21:51 | |
*** devcamcar has joined #openstack | 21:52 | |
*** jdmaturen has quit IRC | 21:53 | |
*** jdmaturen has joined #openstack | 21:54 | |
*** nettok has joined #openstack | 21:54 | |
*** jdmaturen has quit IRC | 21:54 | |
*** vish1 has quit IRC | 21:54 | |
*** jdmaturen has joined #openstack | 21:55 | |
*** tomjones has quit IRC | 21:58 | |
*** enmand has joined #openstack | 21:59 | |
*** vish1 has joined #openstack | 22:02 | |
*** NeilNeely has quit IRC | 22:04 | |
*** binaryWarrior has quit IRC | 22:06 | |
*** binaryWarrior has joined #openstack | 22:06 | |
*** iRTermite has quit IRC | 22:07 | |
rdw | on unit tests taking surprisingly long times: last time this happened to me, it was because my system did not have enough entropy | 22:07 |
rdw | launching a process on Linux blocks until it gets ~128 bytes of entropy, so if you have forky tests and a slowly-replenishing entropy pool (which Debian's is by default for some reason), you will be fucked, man | 22:08 |
*** abecc has quit IRC | 22:08 | |
*** kim0 has quit IRC | 22:14 | |
*** dizzl_ has quit IRC | 22:18 | |
justinsb | Anyone know how to force the guest to 'look for a disk' I've just attached. I'm trying to get iSCSI support working; I have an Ubuntu guest I built using vmbuilder; I've attached the disk using virsh attach-disk on vdb; I've modprobe-d acpiphp; udev is running, but it doesn't show up... | 22:19 |
*** sophiap has quit IRC | 22:20 | |
*** tobym has quit IRC | 22:24 | |
*** novas0x2a has quit IRC | 22:35 | |
*** novas0x2a has joined #openstack | 22:38 | |
*** abecc has joined #openstack | 22:39 | |
*** novas0x2a has quit IRC | 22:50 | |
*** nettok has quit IRC | 22:54 | |
*** oz` has quit IRC | 22:55 | |
*** eshkinkot has joined #openstack | 22:57 | |
*** jonesy has joined #openstack | 22:57 | |
*** silassewell has joined #openstack | 22:57 | |
*** eshkinkot has left #openstack | 22:58 | |
*** rnirmal has quit IRC | 23:00 | |
*** joshuamckenty has joined #openstack | 23:02 | |
*** novas0x2a has joined #openstack | 23:02 | |
*** pdelgallego has quit IRC | 23:03 | |
*** lew has joined #openstack | 23:06 | |
*** jdarcy_ has joined #openstack | 23:07 | |
*** lew has left #openstack | 23:07 | |
*** miclorb_ has joined #openstack | 23:07 | |
*** kirkland has quit IRC | 23:13 | |
*** tobym has joined #openstack | 23:13 | |
*** toddmorey has quit IRC | 23:14 | |
*** JordanRinke has quit IRC | 23:18 | |
*** sophiap has joined #openstack | 23:18 | |
*** kirkland has joined #openstack | 23:19 | |
*** jdarcy_ has quit IRC | 23:20 | |
*** lbieber has quit IRC | 23:25 | |
*** jakedahn has joined #openstack | 23:26 | |
*** dendrobates is now known as dendro-afk | 23:27 | |
*** silassewell has quit IRC | 23:30 | |
*** sophiap has quit IRC | 23:31 | |
*** sophiap has joined #openstack | 23:36 | |
*** jakedahn has quit IRC | 23:36 | |
*** blakeyeager has quit IRC | 23:39 | |
*** jakedahn has joined #openstack | 23:41 | |
*** jakedahn has quit IRC | 23:45 | |
*** novas0x2a has quit IRC | 23:46 | |
*** toddmorey has joined #openstack | 23:50 | |
*** jmckenty has joined #openstack | 23:50 | |
*** joshuamckenty has quit IRC | 23:52 | |
*** jmckenty is now known as joshuamckenty | 23:52 | |
*** dendro-afk is now known as dendrobates | 23:52 | |
*** lbieber has joined #openstack | 23:53 | |
*** jtdowney has quit IRC | 23:53 | |
*** binaryWarrior has quit IRC | 23:55 | |
*** lbieber has quit IRC | 23:55 | |
*** abecc has quit IRC | 23:57 | |
*** kirkland has quit IRC | 23:57 | |
*** kirkland has joined #openstack | 23:58 | |
holoway | btw, I hate you all for giving me a reason to buy a python book :) | 00:00 |
*** tj_yang_ has quit IRC | 00:01 | |
*** lbieber has joined #openstack | 00:01 | |
pvo | holoway: http://diveintopython.org/toc/index.html | 00:02 |
holoway | pvo: thanks! | 00:03 |
*** jbd has quit IRC | 00:04 | |
pvo | holoway: you're welcome : ) | 00:05 |
holoway | pvo: and I'm kidding - more languages, more better | 00:06 |
creiht | hah | 00:06 |
holoway | (what's another language between friends?) | 00:06 |
joshuamckenty | that's what she said | 00:09 |
*** dendrobates is now known as dendro-afk | 00:09 | |
joshuamckenty | body language? | 00:09 |
_0x44 | joshuamckenty: You can't say that. | 00:10 |
joshuamckenty | :) | 00:10 |
*** silassewell has joined #openstack | 00:15 | |
*** Predominant has quit IRC | 00:18 | |
*** abecc has joined #openstack | 00:18 | |
*** lbieber has quit IRC | 00:21 | |
*** com1 has quit IRC | 00:24 | |
*** christiekoehler has quit IRC | 00:30 | |
*** silassewell has quit IRC | 00:35 | |
*** Predominant has joined #openstack | 00:42 | |
*** sophiap has quit IRC | 00:42 | |
*** sophiap has joined #openstack | 00:44 | |
*** sophiap has quit IRC | 00:45 | |
*** murkk_ has joined #openstack | 01:03 | |
*** pvo has quit IRC | 01:04 | |
*** murkk has quit IRC | 01:07 | |
*** murkk_ is now known as murkk | 01:07 | |
*** novas0x2a has joined #openstack | 01:13 | |
*** joshuamckenty has quit IRC | 01:15 | |
*** adjohn has joined #openstack | 01:16 | |
*** aliguori has quit IRC | 01:28 | |
*** dmd17 has joined #openstack | 01:31 | |
*** jakedahn has joined #openstack | 01:38 | |
*** mtaylor has quit IRC | 01:42 | |
*** gustavomzw has left #openstack | 01:42 | |
*** silassewell has joined #openstack | 01:43 | |
*** bret|awa1 has quit IRC | 01:43 | |
*** sophiap has joined #openstack | 01:49 | |
*** mtaylor has joined #openstack | 01:54 | |
*** ChanServ sets mode: +v mtaylor | 01:54 | |
*** sophiap has quit IRC | 01:54 | |
*** novas0x2a1 has joined #openstack | 01:55 | |
*** novas0x2a has quit IRC | 01:57 | |
*** rdw has quit IRC | 01:57 | |
*** vish1 has quit IRC | 01:58 | |
*** vish1 has joined #openstack | 02:00 | |
*** sophiap has joined #openstack | 02:03 | |
*** jdmaturen has quit IRC | 02:03 | |
*** abecc has quit IRC | 02:06 | |
vish1 | bzr geniuses: how does one associate a branch with a bug? | 02:26 |
*** jdmaturen has joined #openstack | 02:28 | |
*** jtdowney has joined #openstack | 02:30 | |
*** anotherjesse has joined #openstack | 02:30 | |
anotherjesse | soren - when do you use blueprints instead of bugs? (I'm thinking about writing about making the vpn system more generic) | 02:31 |
*** joshuamckenty has joined #openstack | 02:34 | |
eday | anotherjesse: we should be using blueprints for plans like that (like the quota "bug" today too). usually bugs in LP are for existing bugs that are found | 02:35 |
*** binaryWarrior has joined #openstack | 02:37 | |
*** binaryWarrior has quit IRC | 02:40 | |
soren | anotherjesse: It's a matter of taste, really. In Ubuntu it's more obvious since a blueprint often involves creating a new package or involves work on a multitude of packages. | 02:40 |
*** joey_ has joined #openstack | 02:40 | |
*** lbieber has joined #openstack | 02:40 | |
soren | anotherjesse: Using the same heuristic in Nova, it depends on how holistic the "issue" is. | 02:41 |
soren | vish1: Two ways: | 02:41 |
soren | vish1: You can either -- after the fact -- link a bug to a branch.. | 02:42 |
soren | vish1: ..or.. | 02:42 |
soren | vish1: as you commit, you pass "--fixes lp:bugnumber" to bzr. | 02:42 |
*** tupshin has quit IRC | 02:42 | |
soren | vish1: That embeds a reference to the bug in the bzr commit metadata which launchpad in turn notices and creates the link. | 02:42 |
vish1 | how do you do it after the fact? | 02:45 |
anotherjesse | vish1: you have a lot of patches outstanding - which ones would you like reviewed most? | 02:48 |
vish1 | cert-fix is tiny | 02:49 |
vish1 | twisted-volume is medium | 02:49 |
soren | vish1: Open the bug page on Launchpad.. | 02:49 |
vish1 | clean-auth is huge | 02:50 |
soren | vish1: On the right, the second box from the top.. | 02:50 |
soren | vish1: click "Link to related branch".. | 02:50 |
vish1 | so go in order of small to big? | 02:50 |
anotherjesse | soren: why has https://code.launchpad.net/~jaypipes/nova/bug606340/+merge/30133 not been merged by hudson? | 02:50 |
soren | vish1: and input the ~vishvananda/nova/whatever thing. | 02:50 |
soren | anotherjesse: I was just about to say something about that :) | 02:50 |
soren | anotherjesse: It's because there's no commit message set. | 02:51 |
soren | anotherjesse: I was about to just set it to Jay's initial text from the merge proposal, but I wasn't sure if it was still accurate after the revisions. | 02:51 |
anotherjesse | i'll set | 02:51 |
vish1 | ah cool thnx | 02:51 |
anotherjesse | the merge queue is nice | 02:52 |
soren | I'm not sure how Jay managed the "dependent branch" thing. | 02:52 |
soren | That's pretty neat. | 02:52 |
*** silassewell has quit IRC | 02:53 | |
*** joshuamckenty has quit IRC | 02:54 | |
*** patcito has joined #openstack | 02:55 | |
*** blakeyeager has joined #openstack | 02:55 | |
*** tupshin has joined #openstack | 02:55 | |
*** blakeyeager has quit IRC | 02:55 | |
*** binaryWarrior has joined #openstack | 02:56 | |
soren | It might just be a matter of pushing both of the branches and Launchpad will work out the relationship. Maybe. Perhaps. I don't know. | 02:58 |
creiht | so my first attempt at hacking a swift-lite was a bit of faail | 02:58 |
creiht | :/ | 02:58 |
vish1 | swift-lite is swift - all functionality? | 02:59 |
creiht | I was trying to use wsgi_intercept to patch the different backend services together into one simple server, but it turns out it isn't quite robust enough to handle it | 02:59 |
anotherjesse | I was just thinking of taking objectstore as it exists and use the same api | 02:59 |
creiht | vish1: I was trying to make a simple version of swift that you guys could use for swift (based on the current swift code) | 02:59 |
anotherjesse | keeping it simple and unscalable | 03:00 |
anotherjesse | ;) | 03:00 |
vish1 | ah nice, as an out of the box replacement for objectstore | 03:00 |
creiht | interface would be the same, but would be only one replica, no replication, etc. | 03:00 |
creiht | yeah | 03:00 |
*** abecc has joined #openstack | 03:00 | |
creiht | I thought it might be easy, but turns out, not so much | 03:00 |
eday | soren: during the merge proposal request you can specify a dependent branch | 03:02 |
*** binaryWarrior has quit IRC | 03:02 | |
*** novas0x2a1 has quit IRC | 03:03 | |
soren | eday: Wow. I've never noticed that. That's cool. | 03:03 |
*** murkk has quit IRC | 03:03 | |
*** murkk has joined #openstack | 03:04 | |
eday | soren: yup :) | 03:04 |
soren | vish1: ^^ that's probably what you want for those two branches you wanted to merge where one included the other. | 03:04 |
*** nettok has joined #openstack | 03:04 | |
* soren considers graphing the number of people in this channel | 03:05 | |
vish1 | soren: cool, i'll do that next time | 03:08 |
*** anotherjesse has quit IRC | 03:08 | |
*** binaryWarrior has joined #openstack | 03:09 | |
*** silassewell has joined #openstack | 03:10 | |
*** dmd17_ has joined #openstack | 03:16 | |
*** dmd17 has quit IRC | 03:16 | |
*** dmd17_ is now known as dmd17 | 03:16 | |
pandemicsyn | gholt: so swift-init shutdown = graceful ? | 03:25 |
notmyname | soren: http://12.am/openstack/ (via exlt) | 03:25 |
notmyname | pandemicsyn: yes | 03:26 |
creiht | It's like karma, but in IRC | 03:27 |
pandemicsyn | dang, gholt must have subconciously anticipated my feature request today | 03:27 |
pandemicsyn | i made him open - https://bugs.launchpad.net/swift/+bug/607388 for me today | 03:28 |
uvirtbot | Launchpad bug 607388 in swift "Add graceful init.d script option" [Wishlist,Invalid] | 03:28 |
notmyname | pandemicsyn: looking at the code, shutdown will send SIGHUP to ['account-server', 'container-server','object-server', 'proxy-server', 'ring-server', 'services-server','auth-server'] and SIGTERM to the rest | 03:29 |
*** jtdowney has quit IRC | 03:31 | |
soren | notmyname: Awesome. | 03:33 |
notmyname | don't thank me. exlt made it :-) | 03:34 |
*** vish1 has quit IRC | 03:34 | |
*** vish1 has joined #openstack | 03:40 | |
*** Predominant has quit IRC | 03:45 | |
*** Predominant has joined #openstack | 03:45 | |
*** novas0x2a has joined #openstack | 03:54 | |
*** abecc_ has joined #openstack | 03:56 | |
*** Predom[osx] has joined #openstack | 03:57 | |
*** Predominant has quit IRC | 03:59 | |
*** Predom[osx] is now known as Predominant | 03:59 | |
*** abecc has quit IRC | 03:59 | |
*** abecc_ is now known as abecc | 03:59 | |
*** Predominant has quit IRC | 04:04 | |
*** Predominant has joined #openstack | 04:04 | |
*** vish1 has quit IRC | 04:05 | |
*** rajijoom has joined #openstack | 04:06 | |
*** xtoddx has left #openstack | 04:09 | |
*** abecc_ has joined #openstack | 04:12 | |
*** jeromatron has joined #openstack | 04:13 | |
*** abecc has quit IRC | 04:14 | |
*** abecc_ is now known as abecc | 04:14 | |
*** dmd17 has quit IRC | 04:19 | |
*** binaryWarrior has quit IRC | 04:19 | |
*** adjohn has quit IRC | 04:20 | |
*** bobo_b has quit IRC | 04:22 | |
*** bobo_b has joined #openstack | 04:23 | |
*** bobo_b has quit IRC | 04:23 | |
*** silassewell has quit IRC | 04:25 | |
*** holoway has quit IRC | 04:28 | |
*** jeromatron has quit IRC | 04:29 | |
*** holoway has joined #openstack | 04:30 | |
holoway | hey, in https://bugs.launchpad.net/swift/+bug/607912 | 04:33 |
uvirtbot | Launchpad bug 607912 in swift "swift-auth-create-account lets you create multiple users with same params but only single token exists causing problems on reset" [Undecided,New] | 04:33 |
holoway | is the fix that we should check to see if the user already exists, and if so, do nothing? | 04:33 |
*** holoway has quit IRC | 04:42 | |
*** holoway has joined #openstack | 04:44 | |
*** toddmorey has quit IRC | 04:45 | |
*** lbieber has quit IRC | 04:48 | |
*** sirp1 has quit IRC | 05:00 | |
*** dmd17 has joined #openstack | 05:01 | |
*** adjohn has joined #openstack | 05:01 | |
*** wylie has quit IRC | 05:02 | |
*** rajijoom has quit IRC | 05:12 | |
*** abecc has quit IRC | 05:20 | |
*** abecc has joined #openstack | 05:22 | |
*** murkk has quit IRC | 05:22 | |
*** spackest has joined #openstack | 05:26 | |
*** abecc_ has joined #openstack | 05:26 | |
spackest | how goes the buzz? | 05:27 |
spackest | folks loving openstack? | 05:27 |
*** abecc has quit IRC | 05:27 | |
*** abecc_ is now known as abecc | 05:27 | |
spackest | we're pretty excited about trying it out | 05:27 |
RandalSchwartz | Simon Phipps and I are interviewing Rick Clark for FLOSS Weekly tomorrow | 05:28 |
RandalSchwartz | if you're around the hallway at 9:30am, you can get into the background of the video. :) | 05:28 |
spackest | is that RandalSchwartz of perl fame? | 05:28 |
RandalSchwartz | indeed | 05:28 |
RandalSchwartz | and FLOSS Weekly | 05:28 |
spackest | we've met at oscon a time or two | 05:28 |
*** rajijoom has joined #openstack | 05:28 | |
RandalSchwartz | cool | 05:28 |
spackest | are you at rackspace? | 05:28 |
RandalSchwartz | no | 05:29 |
spackest | stonehedge? | 05:29 |
spackest | still doing your own thing? | 05:29 |
RandalSchwartz | stonehenge | 05:29 |
chilts | yay, more people to thank ... thanks for all you've done over the years RandalSchwartz | 05:29 |
RandalSchwartz | yes | 05:29 |
* RandalSchwartz bows | 05:29 | |
* chilts likes randomly bumping into people | 05:29 | |
chilts | heh | 05:29 |
* RandalSchwartz hits head on desk | 05:29 | |
RandalSchwartz | ouch! | 05:29 |
spackest | I am a little unsure about the current level of both the object store and compute stuff | 05:30 |
*** joey_ has quit IRC | 05:31 | |
spackest | is the object store released? and does it mirror what rs uses in production? | 05:31 |
chilts | yeah, I think at the moment it's a little big and scary to have a play, but in a week or so I'll have more time :) | 05:31 |
chilts | spackest: I don't think it currently mirrors what RS have | 05:31 |
chilts | I think they have plans to migrate to it later in the year | 05:31 |
chilts | for both compute and storage | 05:31 |
spackest | storage is rather mature though, and compute less so? | 05:32 |
spackest | feels to me like someone has to start an opennstack commercial company, somehow blessed by openstack | 05:33 |
holoway | spackest: I believe swift, the storage part, is basically what they have in production | 05:34 |
spackest | I mean eucalyptus was getting some 100,000,000 valuation, something with the openstack pull could hit there, right? | 05:34 |
chilts | I think the plan is that many cloud server companies can use it (after all, they had 25 companies at their meeting the week before last) | 05:34 |
holoway | spackest: at least that's what I understood at the summit | 05:34 |
holoway | spackest: and nova is in use @ nasa | 05:34 |
chilts | holoway: is that the sumit I was just referring to (the week before last)? | 05:34 |
holoway | but there is work to be odne on moth | 05:34 |
holoway | chilts: yep | 05:34 |
holoway | wow, that was "done on both" | 05:35 |
chilts | holoway: cool, which company were you representing? | 05:35 |
holoway | chilts: Opscode - I'm CTO (we build Chef) | 05:35 |
chilts | sweet, I'll go check it out | 05:35 |
chilts | ah Chef, yeah, I've read about that | 05:35 |
chilts | looks nice | 05:35 |
holoway | chilts: I think it's pretty sweet, but I'm biased | 05:36 |
RandalSchwartz | Chef - on my list for FLOSS Weekly | 05:36 |
chilts | holoway: yep, fair enough | 05:36 |
holoway | nobody thinks they have an ugly baby, you know? :) | 05:36 |
holoway | RandalSchwartz: I'll be there Thursday to give a talk, and Friday as well | 05:36 |
holoway | (where there == OSCON) | 05:37 |
RandalSchwartz | yeah, I'm pretty swamped this week | 05:37 |
RandalSchwartz | I have to retype the entire Learning Perl course into Keynote before next tuesday | 05:37 |
RandalSchwartz | starting tomorrow. :) | 05:37 |
holoway | RandalSchwartz: good luck! | 05:37 |
holoway | RandalSchwartz: that book taught me how to program | 05:38 |
RandalSchwartz | as well as interviewing Rick Clark | 05:38 |
spackest | so let's say I work for a rather large company and we would potentially like to be included in the fun | 05:38 |
*** joshuamckenty has joined #openstack | 05:38 | |
*** Predominant has quit IRC | 05:39 | |
spackest | is there an application process or the like? | 05:39 |
*** abecc has quit IRC | 05:39 | |
*** abecc has joined #openstack | 05:40 | |
holoway | spackest: you can start contributing without doing anything at all other than signing the CLA | 05:40 |
*** Predominant has joined #openstack | 05:40 | |
holoway | spackest: you can talk to Bret Piatt and other similar folks @ Rackspace about getting more on board in the participants page knid of way | 05:41 |
*** joshuamckenty has quit IRC | 05:43 | |
*** dendro-afk is now known as dendrobates | 05:44 | |
*** joshuamckenty has joined #openstack | 05:49 | |
spackest | thanks, will try to have my people :) talk to him | 05:49 |
*** joshuamckenty has quit IRC | 05:55 | |
*** spackest1 has joined #openstack | 05:59 | |
*** spackest has quit IRC | 06:01 | |
*** jdmaturen has quit IRC | 06:03 | |
*** abecc has quit IRC | 06:16 | |
*** abecc has joined #openstack | 06:17 | |
*** abecc has quit IRC | 06:22 | |
*** spackest1 has quit IRC | 06:24 | |
*** abecc has joined #openstack | 06:26 | |
*** rdw has joined #openstack | 06:37 | |
*** jdmaturen has joined #openstack | 06:44 | |
dendrobates | spackest: just go to wiki.openstack.org and start looking around. | 07:03 |
*** abecc has quit IRC | 07:07 | |
dendrobates | mtaylor: have you had a chance to submit your launchpad patch that would allow us to integrate CLA signing? | 07:13 |
*** jakedahn has quit IRC | 07:16 | |
*** ttx has joined #openstack | 07:21 | |
*** flamingoA has joined #openstack | 07:26 | |
*** bret|away has joined #openstack | 07:37 | |
*** pdelgallego has joined #openstack | 07:38 | |
*** ttx has quit IRC | 07:38 | |
*** mindgap has joined #openstack | 07:51 | |
*** adrian_otto has quit IRC | 07:51 | |
*** nettok has quit IRC | 07:53 | |
*** dendrobates is now known as dendro-afk | 07:54 | |
*** termie has quit IRC | 07:55 | |
*** termie has joined #openstack | 07:56 | |
*** termie has quit IRC | 07:56 | |
*** adrian_otto has joined #openstack | 07:58 | |
*** barismetin has joined #openstack | 07:58 | |
*** termie has joined #openstack | 07:58 | |
*** termie has joined #openstack | 07:58 | |
*** kim0 has joined #openstack | 08:02 | |
*** adrian_otto has quit IRC | 08:14 | |
*** ewan_ has joined #openstack | 08:15 | |
*** ewan_ has left #openstack | 08:17 | |
*** ewan_ has joined #openstack | 08:19 | |
*** adrian_otto has joined #openstack | 08:20 | |
*** miclorb_ has quit IRC | 08:20 | |
*** benoitc has joined #openstack | 08:29 | |
*** bret|away has quit IRC | 08:31 | |
*** mindgap has quit IRC | 08:31 | |
*** jdmaturen has quit IRC | 08:31 | |
*** termie has quit IRC | 08:35 | |
*** poli has quit IRC | 08:35 | |
*** cparedes has quit IRC | 08:35 | |
*** soren has quit IRC | 08:35 | |
*** termie has joined #openstack | 08:36 | |
*** poli has joined #openstack | 08:36 | |
*** cparedes has joined #openstack | 08:36 | |
*** soren has joined #openstack | 08:36 | |
*** niven.freenode.net sets mode: +v soren | 08:36 | |
*** patcito has quit IRC | 08:44 | |
*** hugod has quit IRC | 08:45 | |
*** hugod has joined #openstack | 08:49 | |
*** jhalogen has quit IRC | 09:26 | |
*** ttx has joined #openstack | 09:32 | |
*** dmd17_ has joined #openstack | 09:33 | |
*** adjohn_ has joined #openstack | 09:33 | |
*** Daviey has joined #openstack | 09:34 | |
*** dmd17 has quit IRC | 09:36 | |
*** dmd17_ is now known as dmd17 | 09:36 | |
*** adjohn has quit IRC | 09:36 | |
*** adjohn_ is now known as adjohn | 09:36 | |
*** dmd17 has quit IRC | 09:52 | |
*** bahadir has joined #openstack | 09:57 | |
bahadir | So openstack runs on which host OS? I assume it is debian or ubuntu from the installation commands. But I couldn't find any details in the documentation. | 10:11 |
*** [1]johnblue has joined #openstack | 10:22 | |
*** johnblue has quit IRC | 10:25 | |
*** [1]johnblue is now known as johnblue | 10:25 | |
*** calavera has joined #openstack | 10:27 | |
*** jonesy has quit IRC | 10:41 | |
*** gustavomzw has joined #openstack | 10:44 | |
*** tupshin1 has joined #openstack | 10:53 | |
*** tupshin has quit IRC | 10:54 | |
*** ttx has quit IRC | 11:01 | |
*** sophiap_ has joined #openstack | 11:01 | |
*** jtdowney has joined #openstack | 11:02 | |
*** sophiap__ has joined #openstack | 11:02 | |
*** sophiap has quit IRC | 11:05 | |
*** sophiap__ is now known as sophiap | 11:05 | |
*** sophiap_ has quit IRC | 11:06 | |
*** kim0 has quit IRC | 11:09 | |
*** omar has quit IRC | 11:13 | |
PiotrSikora | bahadir: which piece of the software? | 11:16 |
PiotrSikora | swift is mostly python, so it should run on almost every OS | 11:16 |
*** enmand has quit IRC | 11:19 | |
*** jonesy has joined #openstack | 11:20 | |
*** jonesy_ has joined #openstack | 11:21 | |
bahadir | I am mainly interested in the compute part | 11:33 |
bahadir | In the intro it says "OpenStack Compute is software for automatically creating and managing large groups of virtual private servers." | 11:35 |
bahadir | but I don't see any explanation how it does that. XMV, virtualbox, xen ?? | 11:35 |
bahadir | this part should be OS specific I guess | 11:36 |
*** justinsheehy has quit IRC | 11:40 | |
*** kim0 has joined #openstack | 11:42 | |
*** miclorb_ has joined #openstack | 11:46 | |
*** vish1 has joined #openstack | 11:53 | |
*** vish1 has quit IRC | 12:04 | |
*** pcrews has joined #openstack | 12:13 | |
*** vish1 has joined #openstack | 12:18 | |
termie | bahadir: compute uses libvirt | 12:20 |
termie | vish1: yay irc | 12:21 |
*** ker2x has joined #openstack | 12:21 | |
notmyname | bahadir: swift currently targets (or we at rackspace) target ubuntu 10.04 | 12:21 |
ker2x | friendly greetings \o/ | 12:21 |
notmyname | chilts: spackest: swift is what we run in production for Rackspace Cloud Files (ok, there are a few packaging differences, but it is our production code) | 12:22 |
notmyname | ker2x: hi | 12:23 |
ker2x | i'm exploring the website right now :) | 12:23 |
notmyname | great. just ask if you have any questions | 12:24 |
ker2x | i will. thank you :) | 12:24 |
* notmyname is out for a bit. just checking in to see what people talked about overnight | 12:26 | |
*** justinsheehy has joined #openstack | 12:29 | |
*** miclorb_ has quit IRC | 12:32 | |
*** rajijoom has quit IRC | 12:34 | |
*** dabo has joined #openstack | 12:50 | |
*** ttx has joined #openstack | 13:01 | |
*** vish1 has quit IRC | 13:06 | |
*** vish1 has joined #openstack | 13:10 | |
*** vish1 has quit IRC | 13:14 | |
*** larsemil has quit IRC | 13:16 | |
*** redbo has quit IRC | 13:16 | |
*** redbo has joined #openstack | 13:17 | |
*** ChanServ sets mode: +v redbo | 13:17 | |
*** larsemil has joined #openstack | 13:17 | |
*** bd_ has quit IRC | 13:17 | |
*** bd_ has joined #openstack | 13:17 | |
*** lbieber has joined #openstack | 13:18 | |
*** vish1 has joined #openstack | 13:19 | |
*** lbieber has quit IRC | 13:21 | |
*** lbieber has joined #openstack | 13:22 | |
*** _0x44 has quit IRC | 13:22 | |
*** lbieber has quit IRC | 13:22 | |
*** _0x44 has joined #openstack | 13:22 | |
*** ChanServ sets mode: +v _0x44 | 13:22 | |
*** vish1 has quit IRC | 13:23 | |
*** notmyname has quit IRC | 13:23 | |
*** notmyname has joined #openstack | 13:24 | |
*** ChanServ sets mode: +v notmyname | 13:24 | |
*** dendro-afk has quit IRC | 13:24 | |
*** vish1 has joined #openstack | 13:25 | |
*** RandalSchwartz has quit IRC | 13:25 | |
*** pquerna_ has joined #openstack | 13:25 | |
*** anotherjesse has joined #openstack | 13:27 | |
*** pquerna has quit IRC | 13:27 | |
*** jeromatron has joined #openstack | 13:27 | |
*** dendro-afk has joined #openstack | 13:28 | |
*** cory_ has quit IRC | 13:29 | |
*** cory_ has joined #openstack | 13:29 | |
*** z0 has quit IRC | 13:33 | |
*** Daviey has quit IRC | 13:35 | |
*** RandalSchwartz has joined #openstack | 13:37 | |
*** RandalSchwartz has joined #openstack | 13:37 | |
*** anotherjesse has quit IRC | 13:39 | |
*** vish1 has quit IRC | 13:39 | |
*** binaryWarrior has joined #openstack | 13:39 | |
*** Daviey has joined #openstack | 13:41 | |
*** sophiap has quit IRC | 13:42 | |
*** tgodar has quit IRC | 13:44 | |
*** anotherjesse has joined #openstack | 13:47 | |
chmouel | PiotrSikora: run almost every OS is a long shot.. there is a lot of Unix/Linux specifics in the code and performance could be bad with wrong FS | 13:53 |
*** vish1 has joined #openstack | 13:53 | |
*** sophiap has joined #openstack | 13:53 | |
*** z0 has joined #openstack | 13:58 | |
*** binaryWarrior has quit IRC | 14:00 | |
*** binaryWarrior has joined #openstack | 14:00 | |
jdarcy | FYI, I got the Swift functional tests to run on RHEL6 last night. | 14:01 |
jdarcy | I fixed a couple of self-inflicted problems, but testRangedGets was still hanging (timeouts). | 14:02 |
*** pvo has joined #openstack | 14:02 | |
jdarcy | I did a "bzr pull" to get the last couple of days' updates and it went away, but I don't see any bug reports for it. | 14:02 |
*** flamingoA has quit IRC | 14:03 | |
*** anotherjesse has quit IRC | 14:05 | |
*** vish1 has quit IRC | 14:06 | |
*** flamingoA has joined #openstack | 14:10 | |
*** lbieber has joined #openstack | 14:10 | |
*** adjohn has quit IRC | 14:10 | |
*** vish1 has joined #openstack | 14:12 | |
*** barismetin has left #openstack | 14:13 | |
*** silassewell has joined #openstack | 14:14 | |
*** anotherjesse has joined #openstack | 14:15 | |
*** pcrews is now known as pcrews_bbl | 14:15 | |
*** johnpur has joined #openstack | 14:17 | |
*** ChanServ sets mode: +v johnpur | 14:17 | |
*** flamingoA has quit IRC | 14:17 | |
*** tobym has quit IRC | 14:17 | |
*** vish1 has quit IRC | 14:17 | |
*** anotherjesse has quit IRC | 14:18 | |
*** alexh_ has joined #openstack | 14:19 | |
*** jeromatron has left #openstack | 14:20 | |
*** flamingoA has joined #openstack | 14:25 | |
*** bret|away has joined #openstack | 14:27 | |
*** bret|away has quit IRC | 14:31 | |
*** jhalogen has joined #openstack | 14:32 | |
*** hunter has joined #openstack | 14:34 | |
*** vish1 has joined #openstack | 14:36 | |
*** Thweetch has joined #openstack | 14:37 | |
*** lbieber has quit IRC | 14:39 | |
*** tobym has joined #openstack | 14:40 | |
*** bahadir has quit IRC | 14:41 | |
*** sirp2 has joined #openstack | 14:45 | |
*** flamingoA has quit IRC | 14:50 | |
*** jdarcy has quit IRC | 14:50 | |
*** jdarcy has joined #openstack | 14:51 | |
*** JordanRinke has joined #openstack | 14:51 | |
creiht | jdarcy: good to hear you got the functional tests running | 14:57 |
creiht | and to mirror notmyname's response, swift is the exact same code that we run at Rackspace for CloudFiles | 14:57 |
creiht | We are currently running it on two large production clusters | 14:58 |
*** flamingoA has joined #openstack | 14:58 | |
jdarcy | Yeah, if nothing else this will enable to run more intensive tests of our CloudFiles support e.g. in Deltacloud. | 14:58 |
creiht | jdarcy: great! | 14:58 |
jdarcy | Seems like a win/win to me. :) | 14:58 |
creiht | also a reminder for all of those at OSCON today and who are interested in swift, check out Will Reese's talk on how we built swift | 14:59 |
*** omar has joined #openstack | 14:59 | |
creiht | http://www.oscon.com/oscon2010/public/schedule/detail/15634 | 14:59 |
*** dec3pti01 has left #openstack | 15:01 | |
*** gcc has joined #openstack | 15:03 | |
*** aliguori has joined #openstack | 15:05 | |
*** btorch has joined #openstack | 15:09 | |
*** xfrogman5 has joined #openstack | 15:18 | |
*** hunter has quit IRC | 15:18 | |
johnblue | creiht: Is Will Reese's talk going to be recorded in some fashion for review later? | 15:19 |
creiht | johnblue: I hope so, but I am not sure | 15:20 |
*** xfrogman5 has quit IRC | 15:20 | |
*** bret|away has joined #openstack | 15:20 | |
creiht | as soon as I find out, I will post the info here | 15:20 |
*** xfrogman5 has joined #openstack | 15:20 | |
johnblue | cool. Thanks. | 15:22 |
*** anm_ has joined #openstack | 15:24 | |
*** dendro-afk is now known as dendrobates | 15:27 | |
*** dendrobates has joined #openstack | 15:27 | |
*** seshupasam has joined #openstack | 15:27 | |
*** gcc has quit IRC | 15:27 | |
jaypipes | dendrobates: morning poison frog. | 15:27 |
dendrobates | jaypipes: morning. happy to see you decided to stay with us. | 15:28 |
jaypipes | dendrobates: :) | 15:28 |
*** seshupasam has left #openstack | 15:29 | |
* jaypipes wonders whether 67 tabs in Firefox is a few too many... | 15:29 | |
*** gcc has joined #openstack | 15:29 | |
*** xfrogman5 has quit IRC | 15:29 | |
dendrobates | if you can see any text on them, it's too few | 15:30 |
dendrobates | I'm off to oscon. I'll be at the booth talking openstack, please come say hi. | 15:30 |
*** xfrogman5 has joined #openstack | 15:33 | |
*** calavera has quit IRC | 15:36 | |
*** jdmaturen has joined #openstack | 15:38 | |
*** pvo has quit IRC | 15:42 | |
*** pvo has joined #openstack | 15:42 | |
*** pvo has joined #openstack | 15:42 | |
*** pcrews_bbl is now known as pcrews | 15:45 | |
jdarcy | jaypipes: 67 tabs in Firefox is 67 too many, for a Chrome user. ;) | 15:46 |
JordanRinke | oh snap | 15:46 |
pvo | oh its gonna be that kinda party, eh? | 15:48 |
anm_ | in the nova setup wiki, there's the statement: "Vendored python libaries (don’t require any installation)" and in the list is Torando, but when trying to start nova-api, i get: | 15:48 |
anm_ | from tornado import httpserver | 15:48 |
anm_ | ImportError: No module named tornado | 15:49 |
anm_ | is there somehing I nee to do to get the Tornado python libs on Ubuntu 10.04 server? | 15:49 |
anm_ | easy_install maybe? | 15:49 |
*** dendrobates is now known as dendro-afk | 15:53 | |
binaryWarrior | http://www.tornadoweb.org/ | 15:54 |
binaryWarrior | install instructions are on there as well as the ubuntu pre-requiosites | 15:54 |
*** ker2x has quit IRC | 15:54 | |
anm_ | binaryWarrior: thanks, i got past that and the other modules twisted, boto, and ipy, and now when starting nova-api, get: | 15:56 |
anm_ | from gflags import * | 15:56 |
anm_ | ImportError: No module named gflags | 15:56 |
binaryWarrior | did you run the pre-requsite first | 15:57 |
binaryWarrior | sudo apt-get install python-dev python-pycurl python-simplejson | 15:57 |
binaryWarrior | if you happen to have no gflags though, looks liek this guy compiled it | 15:58 |
binaryWarrior | http://compbrain.net/archives/28 | 15:58 |
anm_ | i did the prereqs from the wiki, but it differs from what you say a re the prereqs | 15:59 |
anm_ | apt-get install -y python-setuptools python-dev python-pycurl python-m2crypto | 15:59 |
anm_ | should I create a wiki account and add this info? | 16:00 |
binaryWarrior | try it first | 16:00 |
anm_ | is it time to start getting the wiki up to speed? | 16:00 |
anm_ | ok | 16:00 |
_0x44 | http://code.google.com/p/python-gflags/ | 16:01 |
anm_ | perfect, I may start adding stuff like this to the wiki | 16:02 |
*** hunter has joined #openstack | 16:02 | |
anm_ | I'm a small hosting company that has looked at EVERY cloud framework that exists, and this one shows promise | 16:03 |
anm_ | solely based on the backers | 16:03 |
*** kim0 has quit IRC | 16:04 | |
anm_ | We tried OpenQRM for a few weeks, that was two weeks I'll never get back :) | 16:04 |
vish1 | anm_ are you using the PPA? | 16:05 |
anm_ | PPA? | 16:05 |
*** ttx has quit IRC | 16:05 | |
vish1 | hmm did we lose the instructions on installing from the ppa that were on the wiki? | 16:07 |
anm_ | vish1: no, I'm installing from scratch | 16:07 |
*** dendro-afk is now known as dendrobates | 16:07 | |
anm_ | maybe, the wiki is very slim on info from the view of an openstack newb | 16:07 |
anm_ | searching nova on luanchpad.net returns three PPAs | 16:09 |
antonym | anm_: http://etherpad.openstack.org/HackDay-Slices there are some instructions i think justinsb wrote up in the orange, i haven't had a chance to try it out yet | 16:09 |
anm_ | one called "Nova", and one called "Nova Packages" | 16:09 |
_0x44 | http://wiki.openstack.org/NovaInstallFestInstructions | 16:10 |
_0x44 | Summit install day instructions. | 16:10 |
antonym | yeah, those are from the packges, the one i posted is supposed to be starting from source | 16:10 |
_0x44 | They're probably still relevant. | 16:10 |
vish1 | especially for twisted | 16:10 |
antonym | i'll paste justin's over to the wiki | 16:10 |
vish1 | you probably need to use soren's ppa release of twisted unless you want to patch it yourself | 16:11 |
_0x44 | That reminds me that I'll need to do that. | 16:11 |
anm_ | awesome, lots of chatter about OpenStack right now, came out of nowhere! | 16:12 |
anm_ | opinions on production quality? | 16:12 |
anm_ | as a hosting company, I love contributing to projects that I use in house, so if this works, i'll be pretty active | 16:13 |
creiht | anm_: For the storage side, swift is the same code that we run currently on Cloud Files | 16:13 |
creiht | (at Rackspace) | 16:13 |
anm_ | cool, thanks | 16:14 |
justinsb | My instructions on getting nova going from source are a bit of a mess... If you copy them into the wiki, send me a link and I can update. The code is moving quite fast (e.g. objectstore now uses twisted, nginx no longer required) | 16:14 |
jonesy_ | awesome. I was hoping more howto docs would surface today. Time constraints (projects) are keeping me from spending time playing with openstack. This will speed things up tremendously. Thanks guys! | 16:14 |
anm_ | i have many terabytes of iSCSI storage, and no good way to make it available save for openfiler, so swift is next | 16:15 |
* jonesy_ wishes for more r&d in his role | 16:15 | |
jdarcy | jonesy_: Should RhelInstructions be among those howtos? That's specific to Swift, but still. | 16:15 |
*** hunter has quit IRC | 16:15 | |
jonesy_ | jdarcy: the more the merrier. I only just stumbled upon the installfest instructions. | 16:16 |
jonesy_ | I'm testing each component individually to start, so instructions for swift are great too. | 16:16 |
jonesy_ | SAIO looks like it should be good enough to get started with. | 16:16 |
creiht | jonesy_: if you run into any issues setting it up, let me know | 16:17 |
jonesy_ | but like I said, more docs are always better (assuming they actually work and all that) ;-P | 16:17 |
jonesy_ | creiht: will do, thanks. | 16:17 |
jonesy_ | creiht: are you an engineer on some part/all parts of openstack? | 16:18 |
notmyname | jonesy_: several of the swift and nova devs are in here | 16:18 |
creiht | I work for rackspace on CloudFiles, and thusly on swift | 16:18 |
jonesy_ | I know, but I don't know who's who. | 16:19 |
notmyname | hint: look for the "+" on the names in the channel :-) | 16:19 |
jdarcy | jonesy_: SAIO is a bit Ubuntu-specific, though. I think the standard punishment for running Ubuntu around here (Red Hat) is a hundred hours of Fedora-community service. | 16:19 |
jonesy_ | notmyname: ah - I had that pane hidden. | 16:19 |
creiht | haha | 16:19 |
jonesy_ | good call | 16:19 |
antonym | jonesy_: http://wiki.openstack.org/People | 16:19 |
antonym | there's a few of us there | 16:19 |
pandemicsyn | maybe we need a whos who in the world of openstack wiki page | 16:19 |
jonesy_ | jdarcy: I was an *nix admin in a former life. I don't usually care what distro docs are written for. | 16:19 |
antonym | we should probably just expand that list | 16:20 |
jonesy_ | jdarcy: thought I recognized the handle from somewhere. You in #trilug? | 16:20 |
jdarcy | jonesy_: Yeah, I wouldn't either except when things like an old version of python-webob in the repo trip me up. | 16:20 |
jonesy_ | jdarcy: sure, but that's only a problem on redhat | 16:20 |
jonesy_ | ZING! | 16:20 |
jonesy_ | ;-P | 16:20 |
creiht | We use ubuntu internally, and thus the reason for the focus on the docs. I would much like to make instructions for other distros as well | 16:20 |
jdarcy | jonesy_: Don't know what #trilug is, but we could have met before. | 16:21 |
notmyname | jonesy_: http://wiki.openstack.org/RhelInstructions | 16:21 |
jdarcy | creiht: Yeah, perfectly understandable. I don't think Deltacloud generally writes up Gentoo install instructions either. ;) | 16:21 |
jonesy_ | jdarcy: #trilug is the research triangle (NC) LUG, which holds meetings at RH HQ there. | 16:21 |
creiht | lol | 16:21 |
jdarcy | jonesy_: Ah, no, I'm in Westford MA. | 16:21 |
pandemicsyn | antonym: ah cool, didn't even see that page | 16:21 |
jonesy_ | ah | 16:21 |
antonym | yeah, we just need to expand it up a bit | 16:22 |
jonesy_ | antonym: where's that linked from? Is there some page that aggregates howto docs like that? | 16:24 |
*** adrian_otto1 has joined #openstack | 16:24 | |
antonym | i'm putting together a pointer page on the front page right now :) | 16:25 |
jonesy_ | sweet. | 16:25 |
antonym | i don't think there's a place to tell how to get stuff installed yet on the front page | 16:25 |
antonym | you can find it via the recent changes right now i guess | 16:25 |
*** adrian_otto has quit IRC | 16:26 | |
*** jdmaturen has joined #openstack | 16:28 | |
*** jdmaturen has joined #openstack | 16:29 | |
antonym | http://wiki.openstack.org/InstallInstructions linked off the main page now | 16:31 |
*** veszig has quit IRC | 16:38 | |
*** dendrobates is now known as dendro-afk | 16:38 | |
*** ewan_ has quit IRC | 16:39 | |
jaypipes | jdarcy: heh, went to lunch. but yeah, everyone keeps trying to get me to switch to chrome...I will eventually I think. just lazy... | 16:42 |
gustavomzw | gustavo123 | 16:42 |
gustavomzw | sorry wrong window | 16:43 |
*** pquerna_ has quit IRC | 16:43 | |
*** pquerna_ has joined #openstack | 16:43 | |
*** pquerna_ is now known as pquerna | 16:43 | |
* jaypipes writes down gustavomzw's password... | 16:43 | |
gustavomzw | :) | 16:43 |
*** vish1 has quit IRC | 16:43 | |
silassewell | haha | 16:43 |
*** veszig has joined #openstack | 16:43 | |
_0x44 | gustavomzw: Don't worry, when you type hunter2 all we see is ******* :P | 16:44 |
jaypipes | _0x44: whatup dude? how's things? | 16:44 |
jaypipes | _0x44: getting used to bzr yet? or still a pain in your ass? ;) | 16:45 |
_0x44 | jaypipes: I installed git-bzr so I wouldn't have to worry about that anymore. | 16:45 |
jaypipes | _0x44: rock. | 16:45 |
_0x44 | jaypipes: My latest bane is eldritch non-json json parsing in BASH. | 16:45 |
_0x44 | :) | 16:45 |
jaypipes | hehe | 16:45 |
jaypipes | :) | 16:45 |
thom | _0x44: does it work reasonably? (git-bzr_ | 16:45 |
_0x44 | thom: For everything I've tried so far, it does. | 16:45 |
silassewell | did anyone use bzr before this project was released? | 16:46 |
_0x44 | thom: I have non-openstack work I have to do before the 1st, or RS business people's heads will spin and we'll need an old priest and a young priest. | 16:46 |
_0x44 | silassewell: jaypipes, soren, eday, and lbieber | 16:46 |
thom | _0x44: heh :) | 16:46 |
silassewell | gotcha, I was kinda thinking it was brought over with the project lead or something | 16:47 |
_0x44 | silassewell: Oh, dendro-afk also but he hardly counts. | 16:47 |
_0x44 | :D | 16:47 |
jaypipes | lol | 16:47 |
jdarcy | JSON parsing in bash? Nah, not gonna ask. | 16:48 |
jaypipes | hehe | 16:48 |
jaypipes | me neither. | 16:48 |
jk0 | :D | 16:49 |
_0x44 | jdarcy: The JSON parsing is in C, but it's exporting it as NOT-JSON. | 16:49 |
_0x44 | It's... not pretty. | 16:49 |
_0x44 | It's not nearly as ugly as the original JSON parser in BASH, that only didn't work because BASH doesn't like nested arrays and hashes. | 16:50 |
*** tupshin1 has quit IRC | 16:50 | |
*** bret|away has quit IRC | 16:52 | |
*** greenisus has quit IRC | 16:53 | |
*** jbryce has joined #openstack | 16:54 | |
*** tupshin has joined #openstack | 16:58 | |
*** jakedahn has joined #openstack | 17:02 | |
soren | _0x44: I still don't understand what you need that for. You only need to support crap like that when you don't know what data you're looking for and just want to parse random stuff. | 17:10 |
_0x44 | soren: Because we don't know what data we're looking for. The number of IPs assigned to a slice is not fixed. | 17:10 |
soren | _0x44: So? You know the structure. Querying is way easier than parsing. | 17:11 |
soren | _0x44: If you know the structure, all you really need are loops and grep. :) | 17:11 |
*** silassewell has quit IRC | 17:12 | |
_0x44 | I'm using loops and set/sed, which is why it's not nearly as scary as the original bash json parser was. | 17:12 |
soren | I can imagine :) | 17:13 |
*** rajijoom has joined #openstack | 17:19 | |
*** Predominant has quit IRC | 17:20 | |
jonesy_ | hey I have a question: was Python chosen because NASA was already using it, or RS was already using it? Were both teams coincidentally using it? | 17:23 |
jonesy_ | just curious. :) | 17:23 |
jbryce | both teams were using it | 17:24 |
pvo | cloudservers team is rails back-ended but we're moving to python | 17:25 |
PiotrSikora | any of the swift guys around? | 17:26 |
jonesy_ | pvo: you eyeballing python frameworks to mirror the rails experience, or rolling your own? I guess you just mean rails was used for the API server? | 17:27 |
*** rektide has joined #openstack | 17:28 | |
_0x44 | jonesy_: The CloudServers/Slicehost backend is Rails end to end. | 17:28 |
pvo | we've transitioning to the nova arch | 17:28 |
*** silassewell has joined #openstack | 17:28 | |
pvo | jonesy_: its rails almost front to back. | 17:28 |
notmyname | PiotrSikora: what's uo? | 17:28 |
pvo | we've/we're | 17:28 |
adrian_otto1 | the Cloud Servers API is not a Rails system, but the part that touches the physical hosts is. | 17:29 |
*** tomjones has joined #openstack | 17:29 | |
PiotrSikora | notmyname: i've got a question regarding data in node's sqlite3 database... | 17:29 |
pvo | slicehost is rails... rackspace api is different, that is true | 17:29 |
PiotrSikora | notmyname: what data is in there? only about single node or about few nodes? | 17:30 |
pvo | adrian_otto1: it is more than just the part that touches the hosts | 17:30 |
PiotrSikora | notmyname: and how many times that sqlite3 database is replicated in the ring? | 17:30 |
notmyname | PiotrSikora: there is one db for each container and each account | 17:30 |
notmyname | there are 3 replica's of each throughout the system | 17:30 |
*** adrian_otto1 is now known as adrian_otto | 17:30 | |
notmyname | and the contents are all about the contents of the container or account | 17:31 |
notmyname | like the listing, the size, etc | 17:31 |
adrian_otto | pvo: my point is that it's not an all Rails system, it's a combination of things. | 17:31 |
PiotrSikora | notmyname: does this go in pair with object replicas? or does it go to random nodes? | 17:31 |
PiotrSikora | ie. if db from node A is replicated on node E and F | 17:32 |
PiotrSikora | then is all data from node A also on node E and F | 17:32 |
PiotrSikora | or is it all random? | 17:32 |
notmyname | PiotrSikora: the accounts and contianers have rings, just like the objects | 17:32 |
notmyname | where is it stored is based on that partitioner | 17:32 |
notmyname | (it's called The Ring only for historical reasons. It's not actually a ring anymore. "partitioner" is better word) | 17:33 |
pvo | adrian_otto: right.. thats why I said its "amost rails front to back".. the interesting parts I think are currently rails that are going to be transitioning to nova | 17:33 |
PiotrSikora | notmyname: thank you very much :) | 17:33 |
notmyname | PiotrSikora: we added some archetecture notes yesterday. http://swift.openstack.org/overview_architecture.html | 17:34 |
g0rdy | Aw.. The Ring sounded much nice, take the lead in terminology | 17:34 |
g0rdy | ;) | 17:34 |
_0x44 | adrian_otto: The parts that compare to Nova are entirely rails. The customer-facing parts not-specific to cloud servers are whatever they use for the customer facing fiddly bits. | 17:35 |
PiotrSikora | notmyname: nice, thx :) | 17:35 |
PiotrSikora | notmyname: but if you want to ditch "The Ring" you should probably stop using this name everywhere :P | 17:36 |
notmyname | I doubt we will get rid of the name of it | 17:36 |
notmyname | it hasn't been a "ring" for probably over 6 months :-) | 17:36 |
adrian_otto | PiotrSikora: +1 on dropping that name | 17:36 |
PiotrSikora | well, project is open for 2 days to the public | 17:36 |
notmyname | it does the same thing as a consistent hashing ring (where the name comes from). it's just handles our needs a little better. | 17:37 |
notmyname | so, conceptually, it fills the same role as a hash ring in other systems. | 17:37 |
*** jbryce has quit IRC | 17:38 | |
justinsb | Just cleaned up the notes on running nova from source in the wiki: http://wiki.openstack.org/NovaInstallFestInstructions. I think it should still have big warning flags all over it though, because nova is a-changing pretty fast right now... | 17:39 |
RandalSchwartz | just finished recording FLOSS Weekly with Rick Clark | 17:40 |
RandalSchwartz | should go live later today | 17:40 |
RandalSchwartz | audio *and* video | 17:40 |
*** lbieber has joined #openstack | 17:42 | |
jonesy_ | the new arch overview doc for swift is great. However, swift now kinda reminds me of pvfs a little ;-P | 17:43 |
notmyname | I have to admit, I'm not familiar with pvfs. just pulled up the web page | 17:45 |
*** omar has quit IRC | 17:45 | |
*** omar has joined #openstack | 17:45 | |
jonesy_ | the wheel hasn't been reinvented, but there's a bit of a pattern between filesystems like pvfs, lustre, etc that were designed for HPC (at least I think they were. That's the only context I've used them in) | 17:45 |
adrian_otto | there is a typo under Proxy server "they are streamed directly through the proxy server to of from the user" should be "to or from the user" | 17:46 |
jonesy_ | pvfs certainly has no proxy server, and v1 couldn't deal with.... well... v1 just kinda sucked. | 17:46 |
jonesy_ | I imagine (but lemme know if I'm wrong) that swift is designed such that the pieces are loosely coupled, which is not the case for any of the HPC fs's I know of. | 17:48 |
jaypipes | justinsb: yo, ping! | 17:48 |
notmyname | thanks adrian_otto | 17:48 |
*** pcrews is now known as pcrews_bbl | 17:48 | |
notmyname | jonesy_: the pieces are very loosely coupled | 17:48 |
*** brejoc has joined #openstack | 17:49 | |
notmyname | I know there is a slide in Will Reese's talk to day at oscon about that (different deployment options). I hope those slides (or a recording) are posted later | 17:49 |
brejoc | hi everyone | 17:49 |
* notmyname is trying to type and eat pizza. sorry for typos and slowness | 17:50 | |
justinsb | jaypipes: Hi! | 17:50 |
*** scrllock_ has left #openstack | 17:50 | |
justinsb | jaypipes: On that unit test... I don't think objectstore_unittest.py checks the HTTP side, just the internals | 17:50 |
justinsb | jaypipes: But I think the smoketest should fail, because euca-upload-bundle fails | 17:51 |
justinsb | jaypipes: (presuming the bucket hasn't already been created) | 17:51 |
jaypipes | justinsb: disregard smoketests for right now... :) devcamcar and I are working on cleaning it up... | 17:51 |
jaypipes | justinsb: the unittest does check raised exceptions, though. | 17:51 |
jaypipes | justinsb: badly, but it does check it... | 17:51 |
notmyname | actually, it looks like Will is giving his talk at oscon right now | 17:52 |
jaypipes | justinsb: lemme throw some code together for a test for the bug in question. I'll post the code on the merge prop. | 17:52 |
jaypipes | justinsb: also, see my note on https://bugs.launchpad.net/nova/+bug/607572 ... | 17:52 |
uvirtbot | Launchpad bug 607572 in nova "ObjectStore now checks security (Commit 146), compute node doesn't authenticate" [Undecided,New] | 17:52 |
justinsb | jaypipes: That would be great; I think the unit tests that exist test the inner 'model' classes, not the webserver class. | 17:52 |
justinsb | jaypipes: And I only started learning Python on Thursday, so this is all very new to me :-) | 17:53 |
jaypipes | justinsb: no worries, mate! :) | 17:53 |
jaypipes | justinsb: what I'll do is pull your branch, code in some unittests for your work, and push up a branch to LP for you to see and then merge into your branch before going into trunk... | 17:54 |
justinsb | jaypipes: Re comment on 607572, the ObjectStore now checks that an authentication header is present. Before the twisted rewrite, I don't think it did. | 17:54 |
*** bret|away has joined #openstack | 17:54 | |
justinsb | jaypipes: I don't think it actually verifies that the header is actually correct :-) | 17:54 |
jaypipes | justinsb: so...is one of those bugs invalid now? | 17:54 |
justinsb | jaypipes: I don't think so... 607512 says you can make up Authentication headers without knowing the secret; 607572 says that the compute node doesn't pass any header at all | 17:55 |
justinsb | jaypipes: I think that's a good summary ... I'll pop that into the bug report | 17:56 |
jaypipes | justinsb: k. I was just confused because the title of one is "Objectstore doesn't check security" and the other is " | 17:56 |
jaypipes | ObjectStore now checks security" :) | 17:56 |
creiht | I think the ring name should stay, since someone could use a "consistently hashed ring" if they wanted to | 17:57 |
justinsb | jaypipes: Yeah, sorry! I'll try to change the titles to make it clearer | 17:57 |
creiht | and the idea of a ring is pretty well known | 17:57 |
jaypipes | justinsb: cool! :) | 17:57 |
notmyname | adrian_otto: changed pushed. should be merged and docs rebuilt soon | 17:57 |
adrian_otto | oh, thanks. Sounds like a lot of fanfare for a one word change :-) | 17:58 |
notmyname | hehe | 17:59 |
creiht | notmyname needed something to do anyways :) | 17:59 |
notmyname | one letter then | 17:59 |
notmyname | ya. I solved all the other problems this morning and was looking for something to do | 17:59 |
adrian_otto | well if it were a semicolon, we'd be getting excited. | 17:59 |
jonesy_ | some terminology confusion in this first paragraph of the swift docs. http://swift.openstack.org/overview_architecture.html#the-ring | 18:00 |
notmyname | this is a python project. we frown on semicolons (even in docs!) | 18:00 |
mshadle | security is overrated | 18:00 |
adrian_otto | exactly!! | 18:00 |
jonesy_ | it says a ring is a mapping, and then says it consists of partitions. I'm pretty sure I can figure out what's going on there, but it's kinda confusing if you're coming in cold. | 18:00 |
notmyname | jonesy_: that's why were here! | 18:01 |
jonesy_ | :) | 18:01 |
justinsb | Any nova developers good on twisted and want to take on a TODO: https://code.launchpad.net/~justin-fathomdb/nova/bug607541/+merge/30357 ? TODO is for a better way to convert exceptions into 404/403 HTTP response codes | 18:01 |
notmyname | but seriously, we want to make them readable. the trick is that the ring is a mapping. of partitions. | 18:02 |
jonesy_ | sorry, notmyname, I'm kind of a documentation freak I guess. This was one area where the SAIO was clearer than the docs. | 18:02 |
notmyname | you may be creiht's new best friend | 18:02 |
jonesy_ | notmyname: maybe the idea to be conveyed is that what the Ring represents within the larger deployment is a little different than what it *consists of*. | 18:03 |
jonesy_ | it consists of partitions, but is used in the context of finding things :) | 18:03 |
jonesy_ | so it represents a map, but consists of partitions. | 18:03 |
jonesy_ | of course, that could be totally wrong. | 18:03 |
*** vish1 has joined #openstack | 18:03 | |
notmyname | no, that sounds pretty close | 18:03 |
jonesy_ | good, then I wasn't too tripped up by the verbage. :-D | 18:04 |
notmyname | jonesy_: so what you're saying is that the docs tell you what the ring is, but not what it's there for | 18:04 |
notmyname | at least, in the context of the system as a whole | 18:05 |
jonesy_ | notmyname: even more confusing, the docs conflate the two notions of representation and construction. | 18:05 |
jonesy_ | ....just in the first paragraph! | 18:05 |
jonesy_ | ;-P | 18:05 |
notmyname | hey! we didn't have those docs 2 days ago! creiht worked very hard on it :-) | 18:06 |
jonesy_ | notmyname: and believe me, I'm appreciative. They rock compared to what was there just yesterday, and are miles ahead of lots of projects that have nothing. | 18:07 |
*** mynnx has joined #openstack | 18:07 | |
jonesy_ | pfft. Did I say projects? I meant corporations, open source projects, and entire governmental institutions. :) | 18:07 |
notmyname | hehe | 18:07 |
jonesy_ | anyway, I'm not trying to criticize the docs as much as make sure your docs make you all look like rock stars :) | 18:08 |
g0rdy | jonesy_: does that include universities ? ;) | 18:08 |
jonesy_ | g0rdy: the ones I've had experience with... yeah, mostly. | 18:09 |
notmyname | no, no. I understand. | 18:09 |
*** vish1 has quit IRC | 18:09 | |
notmyname | and always, patches are welcome :-) | 18:09 |
jonesy_ | notmyname: for the docs? Do they use the same mechanism for patches as everything else? | 18:10 |
notmyname | ya. currently the docs are in the same source tree | 18:10 |
jonesy_ | I believe I have finally reached the day when I can no longer avoid Launchpad. | 18:10 |
*** xtoddx has joined #openstack | 18:10 | |
*** pvo has quit IRC | 18:11 | |
creiht | jonesy_: thanks for the feedback | 18:14 |
creiht | and yes, the docs are ReStructured text files in the source tree | 18:15 |
jonesy_ | creiht: I'm writing a patch. | 18:15 |
creiht | awesome | 18:15 |
creiht | I'm working on deployment docs at the moment | 18:15 |
jonesy_ | well, you haven't seen the patch yet. | 18:15 |
jonesy_ | awesome. | 18:15 |
creiht | hah | 18:15 |
jonesy_ | :) | 18:15 |
creiht | those might take me a little longer though | 18:15 |
creiht | meeting time :) | 18:15 |
jonesy_ | progress is progress. | 18:15 |
jonesy_ | enjoy | 18:15 |
*** vish1 has joined #openstack | 18:17 | |
chmouel | FLOSS weekly ? | 18:18 |
adrian_otto | So, some thoughts on http://swift.openstack.org/overview_architecture.html. What this page really talks about are the foundational components of the system. When describing the architecture, it may be better to describe the overall design, its purpose, and the intended use case(s) for the system. The architecture page may actually be more helpful as a narrative rather than specific descriptions of each component that implements the archit | 18:18 |
adrian_otto | Something that covers design concepts that were employed to reach specific objectives. | 18:20 |
uvirtbot | New bug: #608386 in nova "Bucket.delete() throws NotAuthorized exception when Bucket is not empty" [Low,New] https://launchpad.net/bugs/608386 | 18:21 |
*** lbieber has quit IRC | 18:23 | |
RandalSchwartz | FLOSS Weekly, twit.tv/floss | 18:24 |
RandalSchwartz | today's show, openstack, with rick clark | 18:24 |
RandalSchwartz | should be pushed in a few hours | 18:24 |
jtdowney | RandalSchwartz: looking forward to it | 18:25 |
justinsb | I've created a launchpad blueprint for tracking abstracting out the Redis dependency: https://blueprints.launchpad.net/nova/+spec/datastore-abstraction Do I link this to a page on the wiki? It doesn't seem that launchpad actually lets me type anything other than metadata!! | 18:30 |
*** pvo has joined #openstack | 18:32 | |
jonesy_ | a question (notmyname?) -- in the context of this sentence (from the swift docs) would 'cluster' == 'zone'? --> "The partitions of the ring are equally divided among all the devices in the | 18:33 |
jonesy_ | cluster." | 18:33 |
notmyname | no. that should stay as cluster | 18:33 |
*** lbieber has joined #openstack | 18:34 | |
jonesy_ | hm. So a cluster contains zones, contains replicas of partitions, which live on devices, right? | 18:34 |
notmyname | the partitions are evenly distributed throughout the entire swift installation (whether a single machine or multi-dc) | 18:34 |
*** alekibango has joined #openstack | 18:34 | |
jonesy_ | oh right. I'm discounting the single-machine use case. | 18:35 |
notmyname | I think your statement is right | 18:35 |
jonesy_ | cool. | 18:36 |
jonesy_ | thanks | 18:36 |
*** ewan_ has joined #openstack | 18:38 | |
*** vish1 has quit IRC | 18:38 | |
*** xfrogman5 has quit IRC | 18:39 | |
jaypipes | justinsb: somewhat related.. https://code.launchpad.net/~jaypipes/nova/bug608386/+merge/30568 | 18:39 |
*** hunter has joined #openstack | 18:40 | |
*** bret|away has quit IRC | 18:41 | |
*** bretpiatt has joined #openstack | 18:41 | |
*** vish1 has joined #openstack | 18:44 | |
justinsb | jaypipes: Good catch on NotAuthorized/NotEmpty. I'm not sure how S3 surfaces NotEmpty. Currently, if you don't wrap it explicitly, it'll be a 500 error from Twisted. | 18:45 |
holoway | hey guys, on bug 607912 | 18:45 |
uvirtbot | Launchpad bug 607912 in swift "swift-auth-create-account lets you create multiple users with same params but only single token exists causing problems on reset" [Undecided,New] https://launchpad.net/bugs/607912 | 18:45 |
justinsb | jaypipes: Also you need to be careful that you're throwing the exception within the main handler function and not the continuation, otherwise I haven't figured out how to wrap it | 18:45 |
holoway | is the fix to have the create user method return false if the user exists? | 18:45 |
justinsb | jaypipes: No idea how to do that in twisted - hence the TODO that I'm looking for somebody to take ownership of :-) | 18:45 |
holoway | I have code and tests that do just that, but nobody was around to confirm it was even the right thing :) | 18:46 |
*** pvo has quit IRC | 18:46 | |
jaypipes | justinsb: I'll look into that... | 18:47 |
justinsb | jaypipes: As an aside, do I create the content for launchpad blueprint designs on the openstack.org wiki ? | 18:47 |
creiht | holoway: I think you would want to check for existence, and if it is there, then return true | 18:47 |
holoway | creiht: if we that's the case, we'll want to confirm that all the arugments are identical, yes? | 18:47 |
creiht | hrm | 18:47 |
* jaypipes would prefer raising a UserExists exception and enable the caller to decide action.. | 18:47 | |
holoway | creiht: ie: if you said "create me a user on this account with password foo", and it was actually bar, that would be a failure | 18:47 |
*** xfrogman5 has joined #openstack | 18:47 | |
creiht | right | 18:48 |
holoway | jaypipes: that sound smart | 18:48 |
holoway | s/sound/sounds/g | 18:48 |
jaypipes | creiht: thoughts on raising an exception instead? | 18:48 |
* creiht thinks | 18:48 | |
holoway | jaypipes: well, that code returns false on error right now | 18:48 |
holoway | jaypipes: so there is probably code that already exists that relies on that | 18:49 |
creiht | some of that is a byproduct of older systems | 18:49 |
anm_ | hey, how do I get a wiki account setup on http://nova.openstack.org to start adding to the docs? is that the "official" nova related docs? there are bits and pieces scattered all over the place | 18:49 |
* creiht goes to look at that code | 18:50 | |
holoway | http://gist.github.com/484931 | 18:50 |
holoway | is my changes to swift/auth/server.py | 18:50 |
holoway | look at line 269 for the added select | 18:50 |
*** Predominant has joined #openstack | 18:51 | |
creiht | so looking at the code, the api endpoint is a REST call to PUT the account | 18:52 |
jaypipes | anm_: that is auto-generated documentation | 18:52 |
creiht | so if someone tries to put a new account/user it should replace the current account info with whatever was sent | 18:52 |
creiht | holoway: looking now | 18:52 |
anm_ | jaypipes: ok, where is the official user contributed docs? | 18:53 |
jaypipes | anm_: wiki.openstack.org is the wiki docs | 18:53 |
anm_ | ok | 18:53 |
jaypipes | anm_: still very much in progress ;) | 18:53 |
anm_ | jaypipes: i understand, Im interested in contributing! | 18:53 |
*** vish1 has quit IRC | 18:53 | |
jaypipes | anm_: awesome! | 18:53 |
holoway | creiht: yeah, I had that question for you actually | 18:54 |
holoway | creiht: at the moment we always PUT | 18:54 |
*** doublerr has quit IRC | 18:54 | |
jaypipes | anm_: let us know if you run into any issues or need assistance...happy to help. | 18:54 |
*** doublerr has joined #openstack | 18:54 | |
holoway | creiht: and it does feel like RESTfully, that would mean we should just replace | 18:54 |
creiht | holoway: also keep in mind, the intention is for this to be just a dev auth server | 18:54 |
holoway | creiht: that's what I figured | 18:55 |
holoway | creiht: replace would be easy | 18:55 |
creiht | yeah... and would probably prevent most issues | 18:55 |
creiht | if someone created the same account again, they wouldn't run into any issues | 18:55 |
holoway | creiht: right | 18:55 |
holoway | and the intention is clear | 18:55 |
creiht | This would also allow them to change their password if they needed to | 18:55 |
anm_ | ok, running through python dependencies right now, and trying to keep up with which ones are missing | 18:55 |
creiht | yup | 18:55 |
holoway | "give me this user in dev with this set of attributes" | 18:55 |
jaypipes | holoway, creiht: Adam Jacobs aassigned himself to that bug...might want to coordinate with him :) | 18:55 |
holoway | jaypipes: that is me | 18:56 |
creiht | :) | 18:56 |
anm_ | says redis is missing now | 18:56 |
holoway | :) | 18:56 |
jaypipes | holoway: ha! ok then! :) | 18:56 |
holoway | (and there is no S in Jacob :) | 18:56 |
jaypipes | holoway: oops, my bad! | 18:56 |
holoway | jaypipes: from Opscode, we met in Austin | 18:56 |
jaypipes | holoway: yes, I know you ;) didn't realize you were holoway :) | 18:56 |
holoway | it does tend to throw people off | 18:57 |
holoway | creiht: so, I'm +1 on replace | 18:57 |
jaypipes | holoway: and I didn't want y'all to get out of sync | 18:57 |
creiht | sounds good to me as well | 18:57 |
holoway | creiht: want me to just do it? | 18:57 |
creiht | holoway: by all means :) | 18:57 |
holoway | okay | 18:57 |
silassewell | so my understanding is that rackspace will be switching over to nova eventually, anyone know if they're switching to kvm as well (just curious)? | 18:58 |
*** vish1 has joined #openstack | 18:58 | |
*** Predominant has quit IRC | 18:59 | |
*** lbieber has quit IRC | 19:00 | |
alekibango | hello. i am new here and i am also interested in kvm hosting | 19:00 |
creiht | silassewell: The vm will be pluggable, so that RS will stay with Xen | 19:00 |
alekibango | creiht: so writing kvm support should be easy? | 19:01 |
creiht | alekibango: kvm support is already there since that is what NASA uses | 19:01 |
alekibango | good :) | 19:01 |
anm_ | I dont know if it helps, since Rackspace is around here, but I'm a small startup data center/hosting company, and have resources to volunteer to the project, as long as my end state results in a usable OpenStack cluster! | 19:01 |
*** novas0x2a has quit IRC | 19:01 | |
anm_ | I have 10 dell servers, 28tb iSCSI strage appliance, and 300mb of bandwidth | 19:02 |
creiht | anm_: all help is welcome :) | 19:02 |
silassewell | creiht: that's what I figured, I was just curious, I haven't heard of any big kvm deployments (probably not as up on the subjects as others though) | 19:02 |
anm_ | dedicated to building a cloud infrastructure | 19:02 |
*** bretpiatt has quit IRC | 19:02 | |
creiht | holoway: btw... have you tried the chef scripts lately for the SAIO install? I've had a couple of reports of them not working | 19:03 |
holoway | creiht: I'm about to try them myself again | 19:03 |
vish1 | silassewell: nasa nebula uses kvm | 19:03 |
creiht | cool | 19:03 |
holoway | creiht: so if they b0rk, I'll fix 'em | 19:03 |
holoway | creiht: wow, this is going to be a very small fix | 19:04 |
*** jbd has joined #openstack | 19:05 | |
creiht | holoway: don't forget a unit test :) | 19:06 |
holoway | creiht: but of course! | 19:06 |
notmyname | anm_: are you talking about that hardware for a public test cluster or something? or as in "i can test stuff on more than one machine"? | 19:06 |
silassewell | vish1: yeah, I should have said that was the first big deployment I'd heard of, although I didn't see how big their setup is (probably in a press release I read to quickly) | 19:07 |
anm_ | either, right now I'm running xenserver on all nodes, and for testing, I can create VMs for use by the project, and I can also migrate VMs off and rebuild hw nodes as ubuntu for use as well | 19:08 |
anm_ | I'd probably only volunteer their use to the project right now, not hosting publicly accessible clusters, but may want to do that soon | 19:09 |
silassewell | vish1: so assuming 8-core boxes that would be 1875 servers, that's pretty good | 19:10 |
*** pcrews_bbl is now known as pcrews | 19:11 | |
anm_ | also, these are 8 core boxes, but only 8gb ram | 19:11 |
*** carsonm has quit IRC | 19:13 | |
*** Thweetch has left #openstack | 19:14 | |
*** vish1 has quit IRC | 19:20 | |
holoway | creiht: ah, the upstream swift branch in launchpad has changed | 19:20 |
holoway | creiht: is why that cookbookk is failing | 19:20 |
holoway | I'll publish a fix | 19:21 |
*** Predominant has joined #openstack | 19:21 | |
creiht | holoway: ahh... thanks | 19:21 |
holoway | creiht: perhaps it would be wise to just stick it in the source? | 19:21 |
holoway | creiht: so then the steps would be "check out the source on your box, run the SAIO installer, and be done" | 19:21 |
holoway | creiht: which seems like it would suit the dev workflow better | 19:21 |
creiht | hrm | 19:21 |
holoway | creiht: you know you want some ruby infecting your python project, man | 19:22 |
holoway | :) | 19:22 |
creiht | but that also would mean that we would have to maintain the chef script in the source tree :) | 19:22 |
holoway | creiht: I'm going to do that anyway | 19:22 |
creiht | hehe | 19:22 |
holoway | I don't care either way, happy to do whatever works | 19:22 |
creiht | holoway: I think we would be fine with that | 19:23 |
creiht | maybe in a /contrib directory | 19:23 |
holoway | creiht: exactly | 19:24 |
creiht | sounds good to me | 19:24 |
holoway | creiht: I'll file a bug/ticket thingy | 19:24 |
holoway | creiht: and a merge request | 19:24 |
creiht | great | 19:25 |
redbo | And we could do a "quick start with chef" page in the docs or something. I don't want to lose the verbose instructions, because understanding the process is important do doing a real installation. | 19:25 |
holoway | redbo: yep | 19:25 |
creiht | indeed | 19:25 |
holoway | redbo: we'll be building cookbooks for real installation too | 19:25 |
holoway | redbo: just haven't gotten around to it yet | 19:25 |
creiht | I think that is the most that I have ever seen redbo write in IRC before :) | 19:25 |
holoway | redbo: I have a saying about automating this kind of stuff - it runs as root, so you better know what it's doing | 19:26 |
holoway | redbo: doesn't absolve you, just speeds you on your way :) | 19:26 |
creiht | holoway: If you don't mind, it also might be worthwhile adding a blurb at the top of the SAIO instructions pointing to the chef script | 19:26 |
holoway | creiht: I will | 19:26 |
creiht | holoway: and thanks for the contributions! | 19:26 |
holoway | creiht: of course | 19:26 |
*** vish1 has joined #openstack | 19:28 | |
holoway | creiht: it's my first few lines of python, so thank me if you don't have to re-write it all | 19:29 |
* holoway grins | 19:29 | |
creiht | hah! | 19:29 |
holoway | creiht: so that create user function returns an url | 19:29 |
* creiht will be happy if there are no end's :) | 19:29 | |
holoway | creiht: and the other tests look for a matching hash? | 19:29 |
holoway | creiht: thus far! | 19:30 |
holoway | creiht: I'll try and sneak one in | 19:30 |
holoway | creiht: basically the account_hash should change, right? | 19:30 |
creiht | the url that it returns is the storage url for that account | 19:30 |
pandemicsyn | holoway: first few lines of python eh, before you know it you'll have ported Chef to python | 19:30 |
creiht | we don't want the account_hash to change | 19:31 |
holoway | pandemicsyn: we've wanted a python DSL forever | 19:31 |
*** hunter has quit IRC | 19:31 | |
holoway | pandemicsyn: the integration points are there to hook one in | 19:31 |
holoway | pandemicsyn: we just need someone whose python-fu is strong enough that they know what idiomatic pythonized chef would look like | 19:31 |
creiht | the account hash is based off of the account and username | 19:32 |
* jonesy_ wrings hands | 19:32 | |
creiht | the only thing that should change is the password | 19:32 |
creiht | if the username changes, that is a whole new account all together | 19:32 |
holoway | creiht: cool - so if we call the hash generation function we'll get the same result | 19:33 |
creiht | I know that might be kind of confusing, but is due to legacy stuff | 19:33 |
creiht | holoway: if you did it right, it should :) | 19:33 |
holoway | creiht: because the fix for the SQL is just 'INSERT OR REPLACE' | 19:33 |
jonesy_ | I thought about doing that with puppet, holoway. Chef might be fun to do too! | 19:33 |
holoway | jonesy_: doing it with Chef is just serializing objects to JSON | 19:33 |
redbo | no, the hash is just a UUID | 19:33 |
creiht | ahh | 19:34 |
creiht | yeah what redbo said :) | 19:34 |
holoway | jonesy_: I have a working example of doing it with Perl inside a cookbook, and our VP of eng has a working lsip example | 19:34 |
holoway | redbo: ah, so we need to be sure not to call that function if we don't need it | 19:34 |
holoway | so it's not as easy as INSERT OR REPLACE | 19:34 |
redbo | or you could try an update, and if that doesn't update any rows, do an insert | 19:34 |
jonesy_ | Oh man. I went to Python as a means of replacing perl. | 19:34 |
holoway | it needs to be select ? update : insert | 19:34 |
jonesy_ | you used perl to replace... ruby? | 19:34 |
creiht | holoway: yeah | 19:34 |
holoway | jeremydei: not to replace, to extend | 19:35 |
jonesy_ | oh | 19:35 |
jonesy_ | I think I missed some context. | 19:35 |
jonesy_ | my bad. | 19:35 |
holoway | jonesy_: http://search.cpan.org/~holoway/Chef-0.01/lib/Chef.pm | 19:35 |
jonesy_ | oh, gotcha. | 19:35 |
jonesy_ | I was talking about rewriting Puppet in Python. What kind of ruby project starts with P anyway? ;-P | 19:36 |
holoway | jonesy_: gotcha - I'm saying use Chef as a library, and get the DSL in Python, because that's what you care about | 19:36 |
jonesy_ | I was doing it as an academic exercise because its creator sort of arrogantly criticized my evangelizing Python's use in system administration. | 19:36 |
holoway | jonesy_: no need to replicate all the logic in the providers, or dealing with crazy use cases | 19:37 |
jonesy_ | yeah, understood. | 19:37 |
*** ewan_ has quit IRC | 19:37 | |
holoway | jonesy_: yeah, I'm all about pragmatism | 19:37 |
holoway | jonesy_: and Chef is as well, by virtue of that | 19:37 |
holoway | jonesy_: I think it's a bug you can't write Chef in python if that's what makes you happy | 19:37 |
holoway | jonesy_: even though I don't want it at all :) | 19:37 |
jonesy_ | heh | 19:38 |
jonesy_ | glad to have met a rubyist who hasn't taken vows and put on funny clothes and stuff. | 19:38 |
jonesy_ | :) | 19:39 |
holoway | creiht: the password is stored in plain text, yes? | 19:39 |
creiht | yes | 19:39 |
notmyname | https://twitter.com/DevCamCar/status/19100202548 | 19:39 |
creiht | even more reason to be only a dev server! :) | 19:40 |
creiht | notmyname: I wonder how many yeats of openstack experience they are looking for? :) | 19:40 |
notmyname | heh | 19:40 |
creiht | /yeats/years | 19:40 |
redbo | Just 1 yeat | 19:40 |
*** Adrian has joined #openstack | 19:40 | |
*** Adrian is now known as Guest40192 | 19:41 | |
*** benoitc has quit IRC | 19:45 | |
holoway | creiht: hrm, so we call add_storage_account | 19:46 |
holoway | creiht: is there anything that needs to change there as well? | 19:46 |
redbo | You can call that multiple times with the same account hash and it'll be fine. Maybe better! | 19:46 |
creiht | what redbo said | 19:47 |
holoway | redbo: gotcha, so if it exists we should extract the account hash | 19:47 |
holoway | redbo: and pass it to the call | 19:47 |
holoway | redbo: otherwise, let it generate a new one like it does now | 19:47 |
creiht | all add_storage_account does is go out and make sure the account is in the swift cluster | 19:47 |
creiht | holoway: correct | 19:47 |
redbo | I think that'd be best, yeah. | 19:48 |
creiht | We want to do that because a user may have cleared the environment (which doesn't clear the db) which in this case if they add their user again, it will make sure that the account gets propigated into swift | 19:49 |
holoway | is this valid python? http://gist.github.com/485016 | 19:51 |
creiht | holy crap what is that? | 19:51 |
creiht | :) | 19:51 |
_0x44 | holoway: Nope. | 19:51 |
jonesy_ | heh | 19:51 |
jdmaturen | SyntaxError: invalid syntax | 19:52 |
redbo | looks... perly. | 19:52 |
_0x44 | I just commented on your gist | 19:52 |
jdmaturen | s/!/not/ | 19:52 |
jdmaturen | [ on the comment] | 19:52 |
holoway | gotcha | 19:52 |
_0x44 | Ah, good call jdmaturen | 19:53 |
redbo | you could do "account_hash = account_hash or self.add_storage_account()" | 19:53 |
* creiht hates it when redbo does that :) | 19:54 | |
_0x44 | I wish that python had a shortcutting ||= | 19:54 |
holoway | _0x44: yeah, I was just about to ask about ||= | 19:54 |
holoway | :) | 19:54 |
RandalSchwartz | there's a reason there's lots of stuff in Perl :) | 19:54 |
notmyname | account_hash = account_hash if account_hash else self.add_storage_account() | 19:54 |
notmyname | just to be thorough :-) | 19:55 |
jonesy_ | python does have a ternary operator, holoway, but it uses 'if' and 'else' instead of '?' and ':' | 19:55 |
holoway | jonesy_: crazy! | 19:55 |
jonesy_ | it would've been clumsy there anyway, but just for future reference. | 19:55 |
* holoway nods | 19:55 | |
redbo | and it was added in 2.5. I don't know what we're targeting anymore. Probably 2.5. | 19:56 |
creiht | It would be nice to get to 2.5 compatibility | 19:56 |
_0x44 | I think all the docs say 2.6 | 19:56 |
creiht | but at the moment it only works in 2.6 | 19:56 |
creiht | 2.4 wouldn't be reasonable | 19:56 |
redbo | it requires 2.6 right now, but I think mostly we'd have to do a bunch of "from __future__ import with_statement"s to do 2.5. | 19:56 |
jonesy_ | is there some key thing that requires 2.6, or a bunch of syntactic things sprinkled about? | 19:56 |
notmyname | the with statements for timeouts | 19:57 |
jonesy_ | oh. So make context managers not be context managers. That's kind of a bummer. | 19:57 |
_0x44 | What about the exception syntax, that changed between 2.5 and 2.6 didn't it? | 19:57 |
creiht | jonesy_: they are available in 2.5 with an import | 19:58 |
jonesy_ | _0x44: actually, I think the 2.6 syntax is valid in 2.5, but 2.5 supported the older comma syntax. I could be forgetting. | 19:58 |
creiht | I think 2.6 made it so that you didn't have to nest try/except in a try/finally if you wanted both | 19:58 |
creiht | so we may have some changes there | 19:59 |
redbo | I think that was 2.5 | 19:59 |
creiht | yeah I think you are right now | 19:59 |
jonesy_ | me too | 19:59 |
creiht | well | 19:59 |
gholt | He might be right later too | 19:59 |
*** xfrogman5 has quit IRC | 19:59 | |
*** urandom has quit IRC | 20:00 | |
*** _cerberus_ has quit IRC | 20:00 | |
creiht | heh | 20:00 |
_0x44 | redbo's right per http://www.python.org/download/releases/2.5/highlights/ | 20:00 |
creiht | so yeah, shouldn't be difficult to make it 2.5 compatible | 20:02 |
jaypipes | creiht: do we use the multiprocessing module? if so, that's 2.6-only | 20:03 |
*** tobym has quit IRC | 20:03 | |
creiht | jaypipes: I think wre do for some log processing stuff, but you can also pull that in externally | 20:03 |
creiht | (the log processing stuff isn't in the swift codebase yet) | 20:04 |
creiht | It will be once we have it smoothed out a bit | 20:04 |
jaypipes | creiht: nova uses multiprocessing module, too... | 20:04 |
redbo | you can install it for 2.5 | 20:04 |
jaypipes | creiht: in a number of places... | 20:04 |
redbo | it'll just be an external dependency | 20:05 |
jaypipes | redbo: true.. | 20:05 |
anm_ | i have a user behind NAT, and their phone has worked for three months, and now, they dont receive any inbound calls, outbound is fine. looking t the registration, i get this: | 20:05 |
anm_ | http://pastebin.freeswitch.org/13499 | 20:05 |
creiht | anm_: that requires a username and password | 20:05 |
redbo | I think that's a mischan ;) | 20:05 |
*** _cerberus_ has joined #openstack | 20:05 | |
creiht | wrong channel? | 20:05 |
creiht | :) | 20:05 |
jaypipes | hehe | 20:06 |
jaypipes | creiht: user/pass is on the dialog box :) | 20:06 |
jaypipes | creiht: but yeah, wrong channel | 20:06 |
creiht | heh | 20:06 |
anm_ | creiht: snap, wrong channel! | 20:06 |
anm_ | freeswitch | 20:06 |
vish1 | just proposed another refactor for merge | 20:06 |
vish1 | :) | 20:06 |
jaypipes | vish1: hola. | 20:06 |
redbo | anm_: have you tried rebooting it? | 20:06 |
*** urandom has joined #openstack | 20:06 | |
vish1 | jaypipes: hi | 20:06 |
*** tobym has joined #openstack | 20:06 | |
redbo | anm_: is it plugged in? | 20:07 |
jaypipes | vish1: how are ya? I'm also getting a number of refactoring things done around testing... | 20:07 |
anm_ | redbo: i think i did, i sent the reboot command from the FS_cli console, but not sure the command made it to it | 20:07 |
anm_ | ;) | 20:07 |
anm_ | i'll switch to #freeswitch now | 20:07 |
anm_ | the life of a datacenter owner! :P | 20:07 |
*** tobym has quit IRC | 20:08 | |
jaypipes | vish1: is anyone working on generalizing the datastore using a driver/adapter model? If so, I wouldn't mind doing that... | 20:08 |
vish1 | yes i think justin was making steps in that direction | 20:08 |
*** jaker74 has quit IRC | 20:09 | |
vish1 | https://code.launchpad.net/~justin-fathomdb/nova/abstract-data-stores | 20:09 |
jaypipes | vish1: awesome, thx. | 20:10 |
*** xfrogman5 has joined #openstack | 20:11 | |
*** binaryWarrior has quit IRC | 20:12 | |
*** binaryWarrior has joined #openstack | 20:12 | |
*** silassewell has quit IRC | 20:12 | |
*** pvo has joined #openstack | 20:14 | |
*** binaryWarrior has quit IRC | 20:15 | |
vish1 | jaypipes: in all seriousness, do you think an abstract base class is necessary for the interface? | 20:16 |
vish1 | jaypipes: I like having the flexibility to change the interface in only one place, especially since there are plans to completely redo auth soon | 20:17 |
creiht | ABC's aren't available in 2.5 :) | 20:18 |
*** abecc has joined #openstack | 20:18 | |
jonesy_ | :) | 20:18 |
jonesy_ | I was about to type that | 20:18 |
vish1 | well it doesn't have to be a true ABC | 20:18 |
creiht | :) | 20:18 |
vish1 | i can just put the methods in with raise NotImplemented() | 20:18 |
creiht | indeed | 20:18 |
jaypipes | vish1: an abstract base class is designed so the interface only needs to be changed in one place, and that stabilizes the interface. Implementing classes should adhere to the ABC's interface | 20:18 |
*** Guest40192 has quit IRC | 20:19 | |
jaypipes | vish1: so, yeah, I still think an ABC is the right choice... | 20:19 |
jaypipes | vish1: but if it's more bother than it's worth right now, no worries ;) | 20:19 |
vish1 | k | 20:19 |
jonesy_ | vish1: were you thinking that just extending a base class might be an option? | 20:19 |
vish1 | i am considering turning fakeldapdriver into a subclass of ldapdriver | 20:20 |
*** mynnx has left #openstack | 20:20 | |
* jonesy_ thinks that little subproject looks cool, like 100 other subprojects within openstack he's stumbled across. | 20:20 | |
vish1 | with a different __enter__ | 20:20 |
vish1 | rather than checking a flag in the constructor | 20:20 |
jaypipes | vish1: why? why not just a FakeAuthDriver? Do we want to follow LDAP's semantics/API or devise a common auth API and have the LdapDriver do LDAP specific stuff privately? | 20:21 |
vish1 | well we will have other AuthDrivers like RedisDriver or DataStoreDriver | 20:22 |
vish1 | but i think the fake is still useful for testing ldap without installing it | 20:22 |
jaypipes | vish1: right, but no need to have a FakeRedisDriver... | 20:22 |
jonesy_ | vish1: so? | 20:22 |
vish1 | (since we still have to support ldap) | 20:22 |
jaypipes | vish1: all I'm saying is that I think there should be a single generic interface...and we only need a single FakeAuthInterface if that is the case.. | 20:23 |
jaypipes | sorry, FakeAuthDriver.. | 20:23 |
vish1 | i understand what you are saying | 20:23 |
* jonesy_ looks for code for non-fake AuthDriver | 20:23 | |
vish1 | but fakeldap serves another purpose | 20:23 |
jaypipes | jonesy: :) | 20:23 |
jaypipes | vish1: explain? | 20:24 |
vish1 | which is write ldap code for the ldapdriver and be able to test it without installing ldap | 20:24 |
*** joshuamckenty has joined #openstack | 20:24 | |
vish1 | installing ldap on a mac is kind of a pain | 20:24 |
vish1 | :) | 20:24 |
jonesy | ohhh. vish1 -- you want to mock out an ldap server? | 20:24 |
jonesy | for testing? | 20:24 |
joshuamckenty | there's fakeldap | 20:24 |
joshuamckenty | vish wrote that | 20:24 |
joshuamckenty | or do we need something more feature-ful | 20:25 |
jaypipes | vish1: but there is no need to fake an LDAP on mac...just fake the Auth interface.. | 20:25 |
joshuamckenty | yeah, the xen-server patch from Ewan has a nicer approach to multiple implementations of the same interface | 20:25 |
jaypipes | joshuamckenty, vish1: all I'm saying is that we should focus as much as possible on the interface, not an implemention of an interface (which is what LDAP is...) | 20:25 |
jonesy | vish1: is fakeldap the same mocked ldap server as the one written as part of the Twisted project for their testing? | 20:26 |
joshuamckenty | well, technically LDAP is an implementation of the x500 interface, but I don't think we want to standardize at that layer | 20:26 |
vish1 | sure...that is the point of driver | 20:26 |
vish1 | so driver is the interface | 20:26 |
vish1 | ldap is an implementation of interface | 20:26 |
jaypipes | right. | 20:26 |
vish1 | fakeldap is an implementation of interface that happens to use a fake for ldap instead of real ldap | 20:27 |
vish1 | hence the idea of a subclass | 20:27 |
adrian_otto | jaypipes: didn't we have this same problem wit Drizzle... trouble getting good unit tests for the LDAP auth? | 20:27 |
jaypipes | adrian_otto: no, nothing to do with unittests yet :) | 20:27 |
vish1 | class hierarchy as follows AbstractDriver > LdapDriver > FakeLdapDriver | 20:27 |
vish1 | FakeLdapDriver just overrides the import in LdapDriver to use fake | 20:28 |
jaypipes | vish1: why not just AbstractDriver -> FakeDriver and AbstractDriver -> LdapDriver? | 20:28 |
*** silassewell has joined #openstack | 20:28 | |
vish1 | jonesy: no i didn't know that existed. Ours is an ultra simple one that backends to redis | 20:29 |
creiht | jaypipes: I imagine they want to be able to test the code in LdapDriver, without having to actually connect to ldap | 20:29 |
jaypipes | creiht: well then that's not testing the LDAP driver is it? ;) | 20:29 |
vish1 | because I can't test ldap code with FakeDriver | 20:29 |
vish1 | i think we should have FakeDriver as well that doesn't use ldap at all | 20:29 |
jaypipes | creiht, vish1: I got you... | 20:29 |
jaypipes | understood | 20:29 |
*** justinsheehy has quit IRC | 20:30 | |
*** brejoc has quit IRC | 20:30 | |
*** abecc has quit IRC | 20:30 | |
creiht | Or you guys could just pull it out into wsgi middleware, and it doesn't matter what the auth interface looks like :) | 20:30 |
*** tomjones has quit IRC | 20:30 | |
vish1 | so i'll change fakeldap to inherit from ldap and make a todo to put in an ABC | 20:31 |
jaypipes | vish1: sounds good :) | 20:31 |
* creiht doesn't even know if ya'lls services use wsgi | 20:31 | |
redbo | I think you should use zope.interfaces. How else are you going to auto discover available auth drivers? | 20:32 |
glange | redbo: true | 20:32 |
* creiht smacks redbo | 20:32 | |
creiht | don't tempt them :) | 20:32 |
*** spackest has joined #openstack | 20:33 | |
*** bretpiatt has joined #openstack | 20:33 | |
*** vish1 has quit IRC | 20:36 | |
* joshuamckenty smacks redbo as well | 20:36 | |
*** novas0x2a has joined #openstack | 20:44 | |
spackest | so, I am trying to go through http://swift.openstack.org/development_saio.html and on fdisk /dev/sdb I get Unable to open /dev/sdb | 20:44 |
spackest | any ideas | 20:44 |
creiht | spackest: what platform? | 20:44 |
spackest | ubuntu 10.4 | 20:44 |
creiht | and are you doing it as root? | 20:44 |
spackest | yes | 20:44 |
*** clayg has quit IRC | 20:45 | |
spackest | bare metal | 20:45 |
creiht | do you have another device on /dev/sdb (or could it be named something else) | 20:45 |
jonesy | spackest: is there some bit in the error after "Unable to open /dev/sdb"? | 20:45 |
*** benoitc has joined #openstack | 20:45 | |
*** clayg_ has joined #openstack | 20:45 | |
spackest | I don't see anything, the next line is my prompt | 20:46 |
creiht | spackest: how many drives do you have on this server? | 20:46 |
spackest | good question :) | 20:46 |
creiht | hah | 20:46 |
jonesy | lol | 20:46 |
creiht | well it sounds like you don't have a drive connected to /dev/sdb | 20:46 |
spackest | I think you're right | 20:46 |
spackest | darn | 20:47 |
jonesy | inside the box is a big, wide, ribbon cable.... | 20:47 |
jonesy | ;-) | 20:47 |
creiht | or hopefully a small sata cable :) | 20:47 |
spackest | we're only an hour drive away | 20:47 |
spackest | can we do swift on the drive running the os? | 20:48 |
creiht | sure | 20:48 |
creiht | spackest: what filesystem is the drive running? | 20:48 |
creiht | I wouldn't recommend it for production use that way | 20:48 |
spackest | looks like ext4 | 20:48 |
spackest | yeah, we're just trying to get our feet wet :) | 20:48 |
creiht | I think you have to set a mount option to support extended attributes | 20:49 |
creiht | but after you do that, it should be fine | 20:49 |
spackest | might try virtual box | 20:49 |
creiht | It is a lot nicer to play around on a vm | 20:50 |
spackest | this guy? user_xattr | 20:50 |
jonesy | spackest: I believe that's it. I was thinking 'xattr', but I'm not at a linux box to check. | 20:51 |
creiht | spackest: after a quick google, that looks right | 20:51 |
redbo | We need to specify those are the VMWare instructions more clearly. I've always used loopback devices on my OS drive for my VMs. | 20:52 |
creiht | yeah you can do that too | 20:52 |
spackest | trying this user_xattr,acl,extents | 20:52 |
holoway | spackest: yep, that's it for ext3 | 20:52 |
holoway | spackest: you might want to just use a loopback device | 20:52 |
holoway | spackest: if you don't have ane xtra disk | 20:53 |
* creiht always forgets about the loopback device | 20:53 | |
holoway | creiht: I should update the SAIO manual instructions with loopback device stuff | 20:53 |
*** justinsheehy has joined #openstack | 20:53 | |
gholt | redbo: We should probably just change the destructions to do loopback and let others tune things other ways if they want. | 20:53 |
gholt | holoway++ | 20:53 |
creiht | holoway: that would be cool, since redbo never seems to get around to it :) | 20:53 |
glange | destructions? | 20:54 |
redbo | here's what I do: http://pastebin.com/B5dZqvu5 | 20:54 |
*** benoitc has quit IRC | 20:54 | |
*** benoitc has joined #openstack | 20:54 | |
holoway | gholt: yeah, it would be trivial to blow it away and rebuild it | 20:54 |
holoway | redbo: you might need to stick a noauto in there | 20:55 |
holoway | redbo: so you don't try and mount it on reboot | 20:55 |
redbo | I'm fine with it mounting on reboots! | 20:55 |
holoway | redbo: does loop handle the losetup step for you? | 20:55 |
redbo | yes | 20:55 |
holoway | redbo: I believe I've been doing it old skool! | 20:55 |
redbo | it'll use the first available loop device | 20:55 |
holoway | fabulous | 20:56 |
holoway | I'll change my stuff to look like yours then | 20:56 |
jonesy_ | darn. vish1 is gone. Here's the link to the twisted ldap server. Might be useful to him. http://webmail.inoi.fi/open/trac/ldaptor/browser/trunk/ldaptor/protocols/ldap/ldapserver.py | 20:57 |
*** jonesy has quit IRC | 20:57 | |
*** jonesy_ is now known as jonesy | 20:57 | |
*** tobym has joined #openstack | 20:58 | |
*** spackest1 has joined #openstack | 20:58 | |
*** tobym has quit IRC | 21:00 | |
*** xfrogman5 has quit IRC | 21:01 | |
*** bretpiatt has left #openstack | 21:02 | |
*** bretpiatt has joined #openstack | 21:02 | |
*** spackest has quit IRC | 21:02 | |
*** jakedahn_ has joined #openstack | 21:03 | |
*** justinsheehy_ has joined #openstack | 21:04 | |
*** justinsheehy has quit IRC | 21:05 | |
*** justinsheehy_ is now known as justinsheehy | 21:05 | |
*** rajijoom has quit IRC | 21:06 | |
*** jakedahn has quit IRC | 21:06 | |
*** jakedahn_ is now known as jakedahn | 21:06 | |
holoway | is python like perl, in that the first argument being 'self' is implied if you do 'self.foo'? | 21:10 |
RandalSchwartz | that's not true in Perl, so I'm already confused. :) | 21:11 |
holoway | RandalSchwartz: what I meant was $foo->bar('baz') in perl causes $foo to get passed to the bar method as the first argument | 21:11 |
*** clayg_ is now known as clayg | 21:11 | |
RandalSchwartz | that's true | 21:11 |
RandalSchwartz | I just didn't parse the analogy | 21:11 |
holoway | and it looks like in Python something similar is happening with 'self' as the first argument to some of these methods | 21:12 |
holoway | so I'm unclear about whether I have to pass self explicitly, or like perl, it's being passed in implicitly | 21:12 |
_0x44 | holoway: You have to pass self explicitly in Python | 21:12 |
*** vish1 has joined #openstack | 21:14 | |
holoway | _0x44: thanks! | 21:16 |
_0x44 | You're welcome :) | 21:16 |
chmouel | holoway: do python -m this to understand the concept of python | 21:17 |
*** justinsheehy has quit IRC | 21:18 | |
holoway | I think I'm down to a failure because I don't understand how the mock http stuff is working:http://gist.github.com/485158 | 21:20 |
*** tobym has joined #openstack | 21:22 | |
notmyname | holoway: I think someone told you wrong. for methods on an object, you don't explicitly pass self. but you do have to define self in the argument list | 21:22 |
notmyname | err..parameter list of the method | 21:22 |
holoway | notmyname: that makes more sense | 21:23 |
notmyname | so "self.foo(self, arg1)" should be "self.foo(arg1)" and the definition is "def foo(self, a)" | 21:23 |
holoway | awesome | 21:23 |
_0x44 | holoway: Oh sorry, I misunderstood your question. | 21:23 |
*** avsm has joined #openstack | 21:23 | |
holoway | no worries | 21:23 |
holoway | I think I was not very clear | 21:24 |
*** avsm has quit IRC | 21:24 | |
spackest1 | redbo: thanks, worked like a champ | 21:25 |
*** justinsheehy has joined #openstack | 21:26 | |
spackest1 | in swift, what user/group do you folks use for <your-user-name>:<your-group-name> | 21:27 |
spackest1 | might want to put a note in the saio, just kind of in there | 21:27 |
*** pvo has quit IRC | 21:27 | |
spackest1 | for just playing around can we use root? | 21:28 |
holoway | spackest1: whatever your own username/groupname is | 21:28 |
holoway | spackest1: for the SAIO, it's better to run it as yourself | 21:28 |
spackest1 | this box currently just has root, so make a swift user / group? | 21:28 |
notmyname | spackest1: I use "john:john". But that might not work for you | 21:28 |
*** pvo has joined #openstack | 21:29 | |
spackest1 | just thinking an saio note might be nice | 21:29 |
spackest1 | I have done guides where I do something like export MY_USER=john, then use $MY_USER in the other commands | 21:29 |
notmyname | I think the idea is to use whatever you used when you set up the distro (step one of SAIO) | 21:29 |
spackest1 | a little less cut and paste | 21:29 |
notmyname | ya, that would be nice | 21:30 |
spackest1 | I can buy that, guess since we're doing bare metal it is a little different | 21:30 |
notmyname | I always get about halfway through it and think that | 21:30 |
gholt | There's also the idea that you'd be setting things up to do development; and generally you want to be who you are so you don't get confused when you push changes. :) | 21:32 |
*** silassewell has quit IRC | 21:33 | |
rdw | what is SAIO? | 21:34 |
notmyname | swift all in one | 21:34 |
notmyname | rdw: http://swift.openstack.org/development_saio.html | 21:34 |
*** pvo has quit IRC | 21:34 | |
creiht | which are instructions to run swift on one vm for development | 21:34 |
rdw | ah cool | 21:34 |
*** bretpiatt has quit IRC | 21:39 | |
*** xtoddx has quit IRC | 21:43 | |
*** Predominant has quit IRC | 21:43 | |
holoway | creiht: okay, that bug is now set to fix committed and a merge is proposed | 21:44 |
holoway | creiht: which seems like the two right things to do, yes? | 21:44 |
*** vish1 has quit IRC | 21:44 | |
creiht | holoway: I think it should be in progress | 21:45 |
creiht | And when merged, changed to fix committed | 21:45 |
holoway | creiht: done | 21:45 |
creiht | (after the approval) | 21:45 |
holoway | creiht: gotcha | 21:45 |
spackest1 | might be worth putting something in about how your launchpad id is the username part of your email address | 21:46 |
holoway | spackest1: it isn't always | 21:46 |
spackest1 | and maybe a little launchpad help | 21:46 |
holoway | (mine isn't, for example) | 21:46 |
spackest1 | a little confusing :) | 21:46 |
spackest1 | I am new to launchpad | 21:46 |
spackest1 | like having to do a key | 21:46 |
creiht | holoway: yeah I'm still a bit new to this as well | 21:46 |
creiht | holoway: http://wiki.openstack.org/LifeWithBzrAndLaunchpad | 21:47 |
creiht | is useful | 21:47 |
holoway | ya | 21:47 |
spackest1 | but yes, I understand this is a swift guide, not a launchpad guide :) | 21:47 |
holoway | that's how I got this far | 21:47 |
holoway | creiht: yay, patches | 21:47 |
holoway | okay, now back to the rest of my day job | 21:47 |
spackest1 | maybe a link to that page in saio | 21:47 |
*** lbieber has joined #openstack | 21:48 | |
holoway | oh no | 21:48 |
holoway | I left a debug statemnet in | 21:48 |
holoway | lame me | 21:48 |
creiht | hah | 21:48 |
creiht | holoway: at least when you push to your branch on lp, the merge request gets auto updated | 21:48 |
spackest1 | I do this bzr branch lp:swift trunk and get bzr: ERROR: exceptions.NotImplementedError: should resend request to http://feeds.edge.launchpad.net/bazaar/, but this isn't implemented | 21:49 |
*** jdmaturen_ has joined #openstack | 21:49 | |
creiht | spackest1: what version of bzr do you have? | 21:49 |
spackest1 | 2.1.1 | 21:49 |
*** jdmaturen_ has quit IRC | 21:49 | |
spackest1 | just what installed from apt-get install bzr | 21:49 |
creiht | hrm | 21:49 |
*** jdmaturen_ has joined #openstack | 21:50 | |
spackest1 | on ubuntu 10.4 | 21:50 |
creiht | same here | 21:50 |
spackest1 | I am behind a proxy which has been biting me a bit | 21:50 |
spackest1 | there a bzr proxy setting somewhere? | 21:50 |
*** jdmaturen_ has quit IRC | 21:50 | |
creiht | just tried in a clean area and had no problems | 21:50 |
creiht | yeah I wonder if the proxy is causing an issue | 21:50 |
*** jdmaturen has quit IRC | 21:51 | |
*** jdmaturen has joined #openstack | 21:51 | |
johnblue | ... speaking of Ubuntu ... http://www.collegehumor.com/video:1886349 | 21:51 |
johnblue | spackest1: can't ya set an enviroment variable for root that will pass proxy traffic? | 21:53 |
johnblue | ie: setenv http_proxy http://my.awesomeproxy.com:8080 | 21:54 |
creiht | spackest1: it does indeed look like a proxy issue | 21:55 |
spackest1 | http_proxy is already set, as is https_proxy | 21:55 |
creiht | but it might also be part of a bug | 21:55 |
creiht | which is fixed in 2.2 | 21:55 |
spackest1 | creiht: yeah, that's what I am seeing | 21:55 |
*** dabo has quit IRC | 21:55 | |
johnblue | roger. | 21:55 |
creiht | that kinds of stinks :/ | 21:56 |
spackest1 | so, something like this? apt-get install bzr=2.2.1 | 21:57 |
spackest1 | (but with the right version :) ) | 21:57 |
creiht | I think 2.2 is still in beta | 21:57 |
spackest1 | ugh | 21:58 |
*** vish1 has joined #openstack | 21:58 | |
creiht | spackest1: https://launchpad.net/~bzr-beta-ppa/+archive/ppa | 21:58 |
creiht | or maybe not | 21:58 |
creiht | https://launchpad.net/~bzr-nightly-ppa/+archive/ppa | 22:00 |
*** vish1 has quit IRC | 22:00 | |
creiht | gets you nightly builds of 2.2 | 22:00 |
*** phildoc is now known as philaway | 22:00 | |
spackest1 | think I will just try this http://launchpad.net/swift/1.0/1.0.1/+download/swift-1.0.1.tar.gz | 22:00 |
creiht | hehe | 22:00 |
spackest1 | I think I have to do this /etc/swift/auth-server.conf as root, but it is in the As you section? | 22:04 |
creiht | spackest1: root creates the /etc/swift folder and then sets the owner:group to your user | 22:04 |
*** lbieber has quit IRC | 22:05 | |
spackest1 | sorry, right you are | 22:05 |
*** johnblue has quit IRC | 22:08 | |
*** novas0x2a has quit IRC | 22:11 | |
*** londo has quit IRC | 22:11 | |
*** novas0x2a has joined #openstack | 22:11 | |
*** abecc has joined #openstack | 22:12 | |
*** joshuamckenty has quit IRC | 22:15 | |
*** Predominant has joined #openstack | 22:17 | |
spackest1 | hmm, I grabbed http://launchpad.net/swift/1.0/1.0.1/+download/swift-1.0.1.tar.gz and there isn't a unittests file | 22:19 |
spackest1 | there is a test/unit directory | 22:19 |
creiht | mtaylor: -^ | 22:20 |
creiht | spackest1: you can run the unit tests by running: | 22:21 |
creiht | nosetests test/unit | 22:21 |
creiht | from the base dir | 22:21 |
spackest1 | thanks, only got two failures :) | 22:22 |
spackest1 | http://pastebin.com/Mquna4uU | 22:23 |
*** lbieber has joined #openstack | 22:24 | |
creiht | spackest1: hrm | 22:26 |
creiht | spackest1: did you export your python path? | 22:26 |
creiht | or do python setup.py develop yet? | 22:26 |
*** _jc___ has joined #openstack | 22:27 | |
*** jonesy_ has joined #openstack | 22:28 | |
spackest1 | is this good for my path? /usr/local/bin:/usr/bin:/bin:/usr/local/games:/usr/games:/home/swift/bin | 22:31 |
creiht | spackest1: that's different | 22:31 |
spackest1 | I ran the develop | 22:31 |
creiht | that should take care of things | 22:31 |
spackest1 | the /home/swift/bin is the addition | 22:31 |
creiht | k | 22:31 |
spackest1 | let me try again | 22:31 |
spackest1 | yeah, I get the two failures | 22:32 |
creiht | ok then that must be a different issue | 22:32 |
spackest1 | how about the functests? | 22:32 |
spackest1 | things seem ok | 22:32 |
creiht | do you have something in the test/unit/__init__.py file? | 22:32 |
creiht | or is it blank | 22:32 |
spackest1 | there's stuff | 22:33 |
creiht | k | 22:33 |
creiht | let me try on my vm | 22:34 |
redbo | he might have a package installed named "test" or something | 22:34 |
creiht | yeah the local path should override that though right? | 22:34 |
spackest1 | I think? | 22:35 |
*** lbieber has quit IRC | 22:35 | |
redbo | ummmm... well I'd probably just launch a python interpreter and "import test" and "import test.unit" and see what it says. | 22:35 |
uvirtbot | New bug: #608469 in swift ".* script is not in the 1.0.1 download" [Medium,New] https://launchpad.net/bugs/608469 | 22:35 |
*** rdw has left #openstack | 22:35 | |
*** rdw has joined #openstack | 22:36 | |
rdw | the test package is included with many Python installs | 22:36 |
creiht | hrm | 22:36 |
rdw | it contains the python regression tests | 22:36 |
redbo | yeah, ours is outside right now | 22:36 |
*** vish1 has joined #openstack | 22:36 | |
*** cw has joined #openstack | 22:39 | |
*** anticw has joined #openstack | 22:40 | |
creiht | there must be something wrong with the download | 22:41 |
spackest1 | I think it all worked, though I am not completely sure | 22:41 |
creiht | I get errors if I run unit tests on the download, but not so from code | 22:41 |
spackest1 | I got a url and auth back | 22:41 |
creiht | cool | 22:41 |
spackest1 | so swift speaks s3? | 22:42 |
* creiht sighs... there is a lot of stuff missing from the download :/ | 22:42 | |
spackest1 | sounds a little too good to be true :) | 22:42 |
creiht | spackest1: our api differes a little from s3 | 22:42 |
spackest1 | glad we could help ;) | 22:42 |
* creiht smacks mtaylor | 22:42 | |
spackest1 | have this vision of using swift with eucalyptus as walrus | 22:43 |
spackest1 | or with s3fox | 22:43 |
creiht | ahh... the test dir is missing the __init__.py | 22:44 |
_jc___ | tornado web server has a primitive s3 server | 22:44 |
*** miclorb_ has joined #openstack | 22:44 | |
_jc___ | I like the simplicity of the cloudfiles api | 22:45 |
creiht | primative being a key word :) | 22:45 |
*** bretpiatt has joined #openstack | 22:45 | |
redbo | The simplicity is nice. There's stuff we'd like to do that like the XML format isn't really extensible enough for. | 22:46 |
_0x44 | It's scary that the extensible markup language isn't extensible enough. :| | 22:47 |
redbo | haha.. well, I was just short-sighted and didn't realize we'd want to include more information in it later. I wish the list of files was a level deeper. | 22:48 |
*** vish1 has quit IRC | 22:48 | |
spackest1 | so, we have this http://code.google.com/p/lasic/wiki/UsersGuide and I am pondering making a nice install for swift | 22:50 |
spackest1 | things very stable? | 22:50 |
spackest1 | or what is the release schedule like | 22:50 |
spackest1 | it would deploy at amazon, given your credentials | 22:50 |
_0x44 | Here's the expected release schedule: http://wiki.openstack.org/Release | 22:51 |
mshadle | will openstack make my laptop faster? | 22:51 |
spackest1 | or mow my lawn? | 22:52 |
_0x44 | Yes. Additionally, it will give you not one, but two unicorns. | 22:52 |
mshadle | i hope they keep their horns aligned the same way, that could get dangerrrrrrrheerous | 22:52 |
creiht | spackest1: I consider swift to be very stable | 22:52 |
_jc___ | are those release dates for both swift and nova? | 22:52 |
spackest1 | so, I would pull down a tarball with the swift code, just wondering the best way to do that | 22:52 |
*** toddmorey has joined #openstack | 22:52 | |
spackest1 | since the launchpad stuff would be hard to automate | 22:53 |
creiht | spackest1: once we get the kinks of the tarball build fixed, that should be good | 22:53 |
redbo | I don't consider the installation to be very stable :) | 22:53 |
_0x44 | _jc___: Those are the release dates for the next release of both swift and nova, but swift is in production now at RS | 22:53 |
creiht | hehe | 22:53 |
redbo | there's a lot of packaging stuff I think is still settling down | 22:54 |
_jc___ | so RS uses this code base? not some branch or something older? | 22:54 |
redbo | it was forked a few weeks ago | 22:54 |
notmyname | swift is what we use in production for cloud files | 22:54 |
creiht | _jc___: There are some very minor differences | 22:54 |
creiht | Mainly for things like auth | 22:54 |
_jc___ | I assume RS has their own auth thing | 22:54 |
_0x44 | And unicorns | 22:54 |
creiht | handling billing, etc. | 22:54 |
_jc___ | I see | 22:55 |
creiht | There is more code coming as well, but it is a bit auxillary to swift running | 22:55 |
spackest1 | is the cloud files size public? | 22:55 |
mshadle | RS has unicorns? | 22:55 |
creiht | spackest1: don't think so | 22:55 |
mshadle | they'd probably get more job apps if they listed it on the paper | 22:55 |
redbo | We're working on getting our production environments onto this code base, though. | 22:55 |
spackest1 | like 100 gig? :) | 22:55 |
spackest1 | 200? | 22:55 |
_0x44 | mshadle: RS doesn't have unicorns. :( | 22:55 |
creiht | spackest1: you really think that is all we have in production? :) | 22:55 |
spackest1 | 100 PB? | 22:56 |
redbo | _0x44: you didn't get a unicorn? | 22:56 |
spackest1 | well, I saw some 300,000 server number, so I am betting you have a non-trivial amount stored :) | 22:56 |
_0x44 | redbo: I was grandfathered in, so I didn't attend rookie-o and didn't get a unicorn. | 22:56 |
creiht | man... when I came back, I had to do rookie-o all over again :/ | 22:57 |
dubs | yea Will and I had to as well. everyone who walked in the room already knew us :) | 22:59 |
spackest1 | any known significant differences between ubuntu 9.04 and 10.04? | 22:59 |
creiht | spackest1: 10.04 has python 2.6 | 22:59 |
_0x44 | When RS acquired us, they asked if we wanted to. I don't think anyone from pre-RS Slicehost did. | 22:59 |
mshadle | grub2, upstart | 22:59 |
redbo | creiht: you needed a refresher sexual harassment video viewing anyway | 23:00 |
_jc___ | I trid using cyberduck against swift on a deployment I have here, but it seems like the GUI is hardcoded to go to cloudfiles.com, are there are similar user interface tools I can browse my containers and upload files, etc that work with swift? | 23:00 |
*** silassewell has joined #openstack | 23:00 | |
spackest1 | hmm, I get this from 9.04 Python 2.6.2 | 23:00 |
creiht | spackest1: oh cool | 23:01 |
notmyname | _jc___: I've used Cloud Mobile on the iPhone (it lets you set the auth server) | 23:01 |
_0x44 | redbo: We need to send Major back to rookie-o then | 23:01 |
spackest1 | ay idea when the tarball will have everything? | 23:01 |
creiht | _jc___: for command line, swift comes with the st tool | 23:01 |
spackest1 | any | 23:01 |
creiht | spackest1: I've let the guy who handles that know, but he is on vacation :/ | 23:01 |
_jc___ | thanks | 23:01 |
* creiht smacks mtaylor again for good measure | 23:01 | |
notmyname | _jc___: any tool that lets you set the auth server should work | 23:02 |
spackest1 | just can't run all the tests | 23:02 |
spackest1 | but I could add them later | 23:02 |
creiht | spackest1: if you touch test/__init__.py, your unit tests should all run | 23:03 |
*** binaryWarrior has joined #openstack | 23:03 | |
dubs | creiht: are public containers slotted to go in the next release? | 23:05 |
dubs | (or redbo or notmyname) | 23:05 |
redbo | dubs: he just packed up. I don't think anything's slotted yet, but if you mean the "5 months from now" release, I'd say there's a good likelihood. | 23:06 |
dubs | i mean the "3 months from now" release :) | 23:06 |
dubs | http://wiki.openstack.org/Release | 23:06 |
_0x44 | The 2 months from now feature-freeze | 23:06 |
redbo | oh :) I read somewhere the first release would be in 6 months. | 23:08 |
dubs | surprise! | 23:08 |
*** toddmorey has left #openstack | 23:09 | |
_0x44 | Destro announced that the first release was going to be in October in the Releases/Projectwhoozawotzit talk at the summit. | 23:09 |
redbo | dubs: do you have a requirement for that? | 23:09 |
redbo | We'll probably start planning it whenever wreese gets back. | 23:10 |
_0x44 | redbo: If swift supports public containers, it means iback v2 will be easier for this next release | 23:10 |
dubs | well, we are drafting our roadmap for the image service and registry. if swift supports public containers then we can work on the assumption that, at least initially, images are all publicly available | 23:10 |
dubs | there's already a lot scheduled for our initial version, so dealing with authenticated object stores would be nice to leave for v2. swift supporting public containers would be a big deal for us. | 23:12 |
dubs | i spoke to creiht and wreese about it briefly last week, and both seemed to indicate it was one of the next things on the list | 23:12 |
*** joshuamckenty has joined #openstack | 23:13 | |
dubs | redbo: not asking you to commit to it here on the spot, I was just wondering if it had already been decided | 23:14 |
redbo | dubs: No, we don't have anything on paper yet. Will's been too busy. | 23:14 |
dubs | redbo: thanks, I'll follow up with him | 23:15 |
spackest1 | creiht: yeah, all the tests passed after the touch, thanks | 23:16 |
*** joshuamckenty has quit IRC | 23:18 | |
*** adrian_otto has quit IRC | 23:19 | |
*** sophiap has quit IRC | 23:22 | |
*** abecc has quit IRC | 23:22 | |
*** silassewell has quit IRC | 23:24 | |
*** bretpiatt has quit IRC | 23:28 | |
*** rbergeron has joined #openstack | 23:31 | |
*** sophiap has joined #openstack | 23:31 | |
*** pvo has joined #openstack | 23:32 | |
spackest1 | fwiw, 9.04 doesn't have python-eventlet | 23:33 |
_jc___ | maybe its better to use easy_install for that, then its distro independent? | 23:35 |
*** sirp2 has quit IRC | 23:35 | |
*** poli has quit IRC | 23:38 | |
*** dmd17 has joined #openstack | 23:40 | |
*** londo has joined #openstack | 23:42 | |
*** mordred_ has joined #openstack | 23:43 | |
*** ChanServ sets mode: +v mordred_ | 23:43 | |
*** mtaylor has quit IRC | 23:44 | |
*** mordred_ is now known as mtaylor | 23:44 | |
mtaylor | creiht: should be fixed in my latest merge | 23:44 |
mtaylor | creiht: you want I should release tarball again? | 23:45 |
*** joshuamckenty has joined #openstack | 23:49 | |
*** jtdowney has quit IRC | 23:49 | |
*** tobym has quit IRC | 23:50 | |
*** jaker has joined #openstack | 23:52 | |
*** pdelgallego has quit IRC | 23:56 | |
*** spackest1 has quit IRC | 23:59 | |
*** mordred_ has joined #openstack | 23:59 | |
*** ChanServ sets mode: +v mordred_ | 23:59 | |
*** pvo has quit IRC | 00:00 | |
*** pvo has joined #openstack | 00:01 | |
*** mtaylor has quit IRC | 00:01 | |
*** mordred_ has quit IRC | 00:07 | |
*** tobym has joined #openstack | 00:09 | |
*** pvo has quit IRC | 00:13 | |
*** mtaylor has joined #openstack | 00:15 | |
*** ChanServ sets mode: +v mtaylor | 00:15 | |
*** mtaylor has quit IRC | 00:15 | |
*** mtaylor has joined #openstack | 00:15 | |
*** ChanServ sets mode: +v mtaylor | 00:15 | |
*** _jc___ has quit IRC | 00:20 | |
*** johnpur has quit IRC | 00:24 | |
*** xtoddx has joined #openstack | 00:24 | |
*** vish1 has joined #openstack | 00:29 | |
*** littleidea has joined #openstack | 00:32 | |
uvirtbot | New bug: #608496 in nova "README URLs are out of date" [Undecided,New] https://launchpad.net/bugs/608496 | 00:41 |
*** abecc has joined #openstack | 00:41 | |
*** littleidea has quit IRC | 00:44 | |
*** joshuamckenty has quit IRC | 00:46 | |
*** silassewell has joined #openstack | 00:48 | |
notmyname | mtaylor: yes please rerelease the tarball | 00:50 |
mtaylor | notmyname: k | 00:57 |
*** sophiap has quit IRC | 01:04 | |
*** sophiap has joined #openstack | 01:05 | |
*** Predominant has quit IRC | 01:12 | |
*** lbieber has joined #openstack | 01:16 | |
*** vish1 has quit IRC | 01:25 | |
uvirtbot | New bug: #608505 in nova "Error in unit test NetworkTestCase test_subnet_edge" [Undecided,New] https://launchpad.net/bugs/608505 | 01:31 |
*** Predominant has joined #openstack | 01:35 | |
*** adjohn has joined #openstack | 01:36 | |
*** abecc has quit IRC | 01:37 | |
*** pvo has joined #openstack | 01:37 | |
*** pvo has joined #openstack | 01:37 | |
*** pvo has quit IRC | 01:38 | |
*** lew has joined #openstack | 01:38 | |
*** lbieber has quit IRC | 01:43 | |
*** novas0x2a has quit IRC | 01:45 | |
*** lbieber has joined #openstack | 01:48 | |
*** lew has quit IRC | 01:50 | |
*** pvo has joined #openstack | 01:55 | |
*** pvo has joined #openstack | 01:55 | |
*** jhalogen has quit IRC | 01:57 | |
*** lbieber has quit IRC | 02:00 | |
*** Predominant has quit IRC | 02:03 | |
*** sirp2 has joined #openstack | 02:03 | |
*** novas0x2a has joined #openstack | 02:07 | |
*** silassewell has quit IRC | 02:11 | |
*** jdmaturen has quit IRC | 02:12 | |
exlt | (I brought this up before..) the wiki.openstack.org header link to www.openstack.org makes it rather difficult to search the wiki ;-) | 02:19 |
creiht | hah | 02:20 |
exlt | is there a high res copy of the logo used on launchpad that I can grab from somewhere - the 64x64 one is a little small | 02:28 |
creiht | exlt: http://c0179631.cdn.cloudfiles.rackspacecloud.com/openstack-logo_viewimage.jpg | 02:29 |
exlt | creiht: perfect - thanks! | 02:31 |
*** pcrews has quit IRC | 02:32 | |
exlt | that makes the gravatar image a lot better for the github mirror :-) | 02:34 |
exlt | http://github.com/openstack | 02:34 |
*** abecc has joined #openstack | 02:34 | |
creiht | cool | 02:34 |
pvo | is it automatically pulling from launchpad? | 02:34 |
*** gustavomzw has quit IRC | 02:34 | |
creiht | heh... forgot tha tI had created a github login :) | 02:35 |
exlt | pvo yes - cron | 02:35 |
exlt | every 30 minutes | 02:35 |
chmouel | sweet | 02:35 |
*** jdmaturen has joined #openstack | 02:37 | |
*** tjyag has joined #openstack | 02:40 | |
*** miclorb_ has quit IRC | 02:43 | |
*** tupshin has quit IRC | 02:53 | |
*** Predominant has joined #openstack | 03:07 | |
*** jdmaturen has quit IRC | 03:12 | |
*** tjyag has quit IRC | 03:15 | |
*** Predominant has quit IRC | 03:26 | |
*** jonesy has quit IRC | 03:26 | |
*** termie has quit IRC | 03:26 | |
*** alekibango has quit IRC | 03:26 | |
*** cparedes has quit IRC | 03:26 | |
*** soren has quit IRC | 03:26 | |
*** jonesy_ is now known as jonesy | 03:26 | |
*** joey_ has joined #openstack | 03:26 | |
*** alekibango has joined #openstack | 03:28 | |
*** jonesy is now known as 52AABRVLU | 03:31 | |
*** Predominant has joined #openstack | 03:31 | |
*** jonesy has joined #openstack | 03:31 | |
*** termie has joined #openstack | 03:31 | |
*** cparedes has joined #openstack | 03:31 | |
*** soren has joined #openstack | 03:31 | |
*** niven.freenode.net sets mode: +v soren | 03:31 | |
*** Predominant has quit IRC | 03:32 | |
*** comstud has left #openstack | 03:34 | |
*** comstud has joined #openstack | 03:34 | |
mshadle | holy crap, comstud | 03:34 |
comstud | yo | 03:34 |
mshadle | that's a name for the ages | 03:35 |
comstud | :) | 03:35 |
comstud | do i know you? | 03:35 |
mshadle | last time your name got brought up was when i met eday | 03:35 |
comstud | oh, you know eric | 03:36 |
mshadle | jah | 03:36 |
mshadle | and have been efneter for like, ever | 03:36 |
comstud | ahh :) | 03:36 |
mshadle | even ran csircd for a couple days | 03:36 |
comstud | lol.. a couple of days... great! | 03:36 |
mshadle | i wasnt leet enough to be an ircop | 03:36 |
comstud | too bad I could never 'finish' it | 03:36 |
mshadle | hey, being able to reload the server without restarting was great | 03:37 |
mshadle | an idea before its time | 03:37 |
comstud | yeah... it's very poorly implemented, but it worked | 03:37 |
comstud | was a last minute thing | 03:37 |
mshadle | -everything- nowadays should be designed in that fashion | 03:37 |
mshadle | an angelesque master | 03:37 |
comstud | yep | 03:37 |
comstud | pretty much anything I do is modular like that | 03:37 |
comstud | i'm working for Rackspace now | 03:38 |
mshadle | would be great when microsoft figures it out | 03:38 |
comstud | yeah really | 03:38 |
mshadle | rackspace is snatching up everyone! | 03:38 |
comstud | yeah :) | 03:38 |
mshadle | part of openstack should be a new cloud-friendly ircd | 03:38 |
mshadle | less netsplits. use gearman for relay! | 03:39 |
comstud | i would like to implement a new one | 03:39 |
comstud | well | 03:39 |
mshadle | heh | 03:39 |
comstud | eric and I are working on: scalestack.org | 03:39 |
mshadle | yeah | 03:39 |
mshadle | i wanted to bug him about how everything intersects | 03:39 |
comstud | (the name similarity to openstack is i think coincidental :) | 03:39 |
comstud | i will be creating an irc module for it at some point | 03:39 |
mshadle | he's been wanting to call it scalestack before brian put pandora on it | 03:39 |
comstud | i have a feeling :) | 03:39 |
mshadle | haha nice | 03:40 |
comstud | it's always weird running into ppl that know me :) | 03:41 |
comstud | where do you work/what you do? | 03:41 |
jk0 | yeah, asl? | 03:42 |
jk0 | :P | 03:42 |
comstud | jk0 :) | 03:42 |
jk0 | :D | 03:42 |
*** miclorb has joined #openstack | 03:43 | |
*** terje has joined #openstack | 03:47 | |
*** dabo has joined #openstack | 03:47 | |
mshadle | comstud: you still talk to tay? :) | 03:49 |
comstud | rarely | 03:49 |
comstud | random comments exchanged on facebook sometimes. | 03:49 |
*** miclorb has quit IRC | 03:50 | |
mshadle | she liked me. | 03:50 |
comstud | she liked everyone | 03:50 |
mshadle | h3h | 03:50 |
mtaylor | creiht: internet ... terrible... uploading ... ass ... tomorrow ... USA ... better ... will do then | 03:54 |
mshadle | mtaylor: what beach at you at? | 03:55 |
mtaylor | mshadle: cozumel | 03:55 |
*** jsgotangco has joined #openstack | 03:56 | |
mshadle | probably faster than the wifi at oscon on monday/tuesday | 03:56 |
mtaylor | mshadle: did some excellent snorkeling today - but now my back is, well, quite burned | 03:56 |
*** nettok has joined #openstack | 03:56 | |
mshadle | i could walk the 1's and 0's on paper quicker | 03:56 |
mtaylor | mmm | 03:56 |
*** miclorb has joined #openstack | 03:56 | |
terje | ahoy, ahoy... I'm interested in fiddling around with swift and have a couple of questions | 03:57 |
terje | does it show up as a file system on the compute nodes? | 03:58 |
*** poli has joined #openstack | 03:58 | |
pvo | terje: no, you use it like S3 | 03:58 |
chmouel | terje: it's not a POSIX FS it's designed to be used via API | 03:58 |
terje | sorry, I mean the vm container systems | 03:58 |
terje | I get how VM's will access it | 03:58 |
pvo | they will use it just like they would any other api | 03:59 |
terje | or, perhaps I don't. | 03:59 |
terje | ok | 03:59 |
pvo | the vm would run and your code would run inside the vm operating system | 03:59 |
terje | ok, that's for my customer's VM's | 03:59 |
* pvo thinks maybe we need a good diagram detailing how this all works together | 03:59 | |
pvo | I realize it can be kind of vague | 03:59 |
terje | well, I've run eucalyptus | 04:00 |
terje | so I do understand s3 and ebs | 04:00 |
terje | I was hoping for some kind of distributed file system | 04:00 |
pvo | no, that isn't what swift is | 04:00 |
terje | gotcha. | 04:00 |
*** dmd17 has quit IRC | 04:00 | |
terje | so, how do you do block storage then? | 04:01 |
pvo | there are a ton of those out there though. Lots to choose from | 04:01 |
terje | is that swift also? | 04:01 |
*** adjohn has quit IRC | 04:01 | |
comstud | swift doesn't provide block storage | 04:01 |
pvo | nova uses aoe right now, iirc | 04:01 |
terje | ah | 04:01 |
comstud | nod | 04:02 |
terje | alrighty | 04:02 |
comstud | iscsi support in a branch | 04:02 |
terje | k | 04:02 |
terje | so, unrelated question then.. | 04:03 |
comstud | go for it | 04:03 |
terje | will nova manage infrastructure systems? | 04:03 |
terje | like, imaging servers, etc. | 04:03 |
terje | say, I want to add a compute node | 04:04 |
comstud | i think the answer is yes.. but depends on what you mean | 04:04 |
_0x44 | terje: The image proxy/pusher sub-projects aren't done yet. | 04:04 |
terje | ok | 04:04 |
*** adjohn has joined #openstack | 04:04 | |
*** dmd17 has joined #openstack | 04:04 | |
pvo | it should hit the blueprints soon | 04:04 |
terje | so, to get started, I take my OS of choice and drop them on a set of systems? | 04:04 |
pvo | anything kvm supports, sure | 04:05 |
terje | cool | 04:05 |
pvo | kvm today, xen server soonish | 04:05 |
terje | CentOS it is. | 04:05 |
terje | :) | 04:05 |
comstud | yep | 04:05 |
terje | ok, thanks. | 04:05 |
mshadle | ubuntu! | 04:07 |
terje | or that | 04:07 |
terje | ah.. so does swift use rsync on the back end for file replication? | 04:13 |
pandemicsyn | yep | 04:13 |
terje | that's very clever, is it. | 04:13 |
terje | s/is/isn't/ | 04:13 |
pvo | why reinvent the wheel? | 04:14 |
terje | why xfs I wonder? | 04:14 |
pandemicsyn | terje: should check out http://swift.openstack.org/overview_replication.html particularly the Object Replication bit | 04:15 |
redbo | doesn't have to be xfs, it just performs the best | 04:15 |
terje | I like this project already.. | 04:16 |
terje | :) | 04:16 |
terje | i will, thanks. | 04:16 |
pvo | terje: thats good to hear. Tell your friends! | 04:19 |
*** joey_ has quit IRC | 04:20 | |
pandemicsyn | I just setup a openstack group on linkedin any idea who our community/marketing person is so I can give them ownership of the group ? | 04:21 |
jk0 | Rick Clark | 04:23 |
jk0 | http://www.linkedin.com/profile?viewProfile=&key=18199677 | 04:24 |
*** poli has quit IRC | 04:24 | |
*** jonesy has quit IRC | 04:24 | |
*** termie has quit IRC | 04:24 | |
*** cparedes has quit IRC | 04:24 | |
*** soren has quit IRC | 04:24 | |
pvo | jk0: not sure Rick would want to run the linked in marketing stuff. | 04:24 |
jk0 | he'll know who to give it to, if anything | 04:25 |
pvo | I suggested someone in recruiting. I hear they like that stuff | 04:25 |
*** poli has joined #openstack | 04:29 | |
*** jonesy has joined #openstack | 04:29 | |
*** termie has joined #openstack | 04:29 | |
*** cparedes has joined #openstack | 04:29 | |
*** soren has joined #openstack | 04:29 | |
*** niven.freenode.net sets mode: +v soren | 04:29 | |
*** dabo has quit IRC | 04:31 | |
*** jakedahn has quit IRC | 04:32 | |
*** jakedahn_ has joined #openstack | 04:32 | |
terje | ever consider hadoop to do the work of swift? | 04:34 |
terje | hdfs, even | 04:34 |
*** tupshin has joined #openstack | 04:34 | |
chmouel | terje: what for ? hadoop need a backend like swift or s3 to work not the other way around | 04:34 |
terje | for replication | 04:34 |
terje | just a thought.. | 04:35 |
chmouel | replication is actually quick and not data intensive enough to need a mapreduce kind of data sets | 04:35 |
terje | ok, so replication isn't a big deal. | 04:37 |
terje | I was just suggesting using hadoop's file system, not really map reduce | 04:38 |
*** joey_ has joined #openstack | 04:38 | |
terje | I was just curious, I would like to find something akin to gfs | 04:38 |
*** joey_ has quit IRC | 04:39 | |
chmouel | ah I did not know about HDFS thought it was still using S3 as backend | 04:39 |
terje | sure, s3 is kind of independent of how you store and replicate the data. | 04:40 |
redbo | hdfs is nice as storage for hadoop, but it has some availability and scaling problems. | 04:40 |
terje | ok, that's what I was wondering about. | 04:40 |
redbo | mostly single point of failure in the namespace server, and I think the entire filesystem has to fit in RAM on the namespace server. | 04:41 |
redbo | I mean the metadata of course, not the file data :) | 04:42 |
chmouel | there is the CEPH distributed file system that I think is getting integrated in the linux-kernel if I am not wrong if you look for something GFS alike | 04:42 |
chmouel | cf: http://lwn.net/Articles/258516/ | 04:42 |
terje | so the metadata has to be replicated too | 04:43 |
terje | right | 04:43 |
chmouel | it's one of the reason why GFS(1) have so much locking problem sometime :-\ | 04:44 |
terje | I see | 04:44 |
*** miclorb has quit IRC | 04:52 | |
*** littleidea has joined #openstack | 04:57 | |
*** joshuamckenty has joined #openstack | 05:17 | |
*** xtoddx has quit IRC | 05:20 | |
mshadle | i cannot figure out where a confirmation that it is okay to use this: //foo.com/bar.jpg and the browser will auto fill in http or https, appropriately (good for cross-domain requests that may be mixed mode otherwise) | 05:23 |
joshuamckenty | Evening everyone, did I miss the excitement? | 05:26 |
pvo | yea, pretty dead in here right now | 05:27 |
pvo | mshadle: not sure | 05:27 |
mshadle | i can't tell if it's a nice little browser thing or a universal | 05:27 |
*** littleidea has quit IRC | 05:27 | |
redbo | mshadle: looks like that's called a "network-path reference" | 05:27 |
redbo | in http://www.ietf.org/rfc/rfc2396.txt | 05:28 |
mshadle | sweet | 05:29 |
mshadle | "network-path reference" | 05:29 |
mshadle | er | 05:29 |
*** pvo has left #openstack | 05:31 | |
*** miclorb_ has joined #openstack | 05:35 | |
mshadle | yeehaw | 05:38 |
*** sureshgv has joined #openstack | 05:39 | |
*** littleidea has joined #openstack | 05:40 | |
sureshgv | Hi all | 05:40 |
sureshgv | This is sureshgv from India | 05:41 |
mshadle | http://michaelshadle.com/2010/07/21/little-known-uri-shorthand-the-network-path-reference/ :) | 05:41 |
sureshgv | would like to involve in development of opestack | 05:41 |
sureshgv | mshadle: hey | 05:41 |
sureshgv | i thought u guys are sleep :) | 05:41 |
sureshgv | sleeping* | 05:42 |
sureshgv | logger url | 05:43 |
sureshgv | hmm | 05:43 |
_0x44 | sureshgv: Not all of us are. | 05:43 |
_0x44 | sureshgv: Some of us are in the middle of a deploy. :P | 05:43 |
sureshgv | ohh | 05:43 |
sureshgv | nice! | 05:43 |
sureshgv | whats the deploy? | 05:44 |
sureshgv | can you give me the details.. | 05:45 |
sureshgv | please | 05:45 |
sureshgv | any one around? | 05:50 |
antonym | i need sleep | 05:50 |
sureshgv | am just curious about what you guys are hacking on.. | 05:51 |
sureshgv | antonym: please let me know | 05:51 |
_0x44 | sureshgv: Ah, we're not hacking on openstack stuff right now, just old stuff | 05:51 |
sureshgv | old stuff means? | 05:51 |
mshadle | i'm gonna port openstack to php | 05:52 |
antonym | sureshgv: we're working on an internal rackspace deploy | 05:52 |
*** littleidea has quit IRC | 05:52 | |
sureshgv | ohh..okie | 05:52 |
sureshgv | mshadle: cool | 05:52 |
sureshgv | antonym: thnx | 05:52 |
redbo | mshadle: about time | 05:52 |
antonym | no prob | 05:52 |
jk0 | mshadle: I was thinking ColdFusion might make for a better port | 05:53 |
mshadle | i don't know python and hate the idea of whitespace dependent stuff and i know procedural php awesome. | 05:53 |
*** ghchinoy has left #openstack | 05:53 | |
*** ghchinoy has joined #openstack | 05:53 | |
mshadle | so.. i don't want to adjust for openstack i want openstack to adjust for me | 05:53 |
sureshgv | he he.. | 05:53 |
redbo | Well, php is really the best choice anyway. And why isn't everything stored in mysql? | 05:54 |
jk0 | redbo: how about some sort of PHP/ColdFusion combo? | 05:55 |
redbo | I like how you think | 05:55 |
jk0 | that way we can run it on Windows | 05:55 |
mshadle | PHP could be the model and coldfusion could be the controller | 05:55 |
redbo | ok, someone write up a blueprint | 05:56 |
jk0 | we have ourselves a project | 05:56 |
* jk0 nominates sureshgv | 05:56 | |
sureshgv | oh.. yeah | 05:56 |
sureshgv | let me know who can i help u guys | 05:57 |
mshadle | can we run it through parrot and make it into php | 05:57 |
mshadle | that would save a couple steps | 05:57 |
sureshgv | jk0: what should i do | 06:00 |
jk0 | sureshgv: we were just joking around about the PHP/ColdFusion version (or at least I was) :) | 06:01 |
sureshgv | :) okie | 06:03 |
*** pvo has joined #openstack | 06:03 | |
sureshgv | mean while will go through the wiki of openstack .. | 06:03 |
sureshgv | and nova development | 06:03 |
sureshgv | have a question.. | 06:05 |
sureshgv | nova and swift both developed in python right? | 06:06 |
sureshgv | jk0: ^^ | 06:06 |
jk0 | that is correct | 06:07 |
sureshgv | okie | 06:07 |
sureshgv | jk0: there is some meetup on jul 29th for nova development etc | 06:08 |
sureshgv | is there any special irc for it? | 06:08 |
sureshgv | suppose if i want to know what going on there | 06:09 |
jk0 | I haven't heard anything, but I'm sure they'll post something on the twitter account: http://twitter.com/openstack | 06:09 |
sureshgv | ohh..okie | 06:10 |
sureshgv | i just get the new from meetup | 06:10 |
sureshgv | will follow twitter from now onwards | 06:10 |
jk0 | they're pretty good about posting updates there | 06:11 |
sureshgv | nice .. | 06:11 |
*** rajijoom has joined #openstack | 06:12 | |
*** pvo has quit IRC | 06:14 | |
*** sirp2 has quit IRC | 06:14 | |
*** jhalogen has joined #openstack | 06:14 | |
*** jdmaturen has joined #openstack | 06:21 | |
*** spackest has joined #openstack | 06:24 | |
spackest | so, I think I have a couple scripts that will install saio | 06:25 |
spackest | still don't have a good feel for how to really use it though it looks like I have it working | 06:25 |
spackest | somehow I thought I could point s3 tools at my saio, but it sounds like that's not true? | 06:26 |
*** jdmaturen has quit IRC | 06:29 | |
spackest | or something I can do in a browser? | 06:31 |
*** pvo has joined #openstack | 06:31 | |
*** pvo has joined #openstack | 06:31 | |
joshuamckenty | sureshgv: the meetup is organized by cloudkick | 06:33 |
joshuamckenty | it's at the cloudkick offices in the South Bay | 06:33 |
joshuamckenty | and posted on (I believe) eventbrite | 06:33 |
joshuamckenty | it's also listed on the wiki somewhere... | 06:34 |
*** dmd17 has quit IRC | 06:35 | |
comstud | actually in the city.. not south bay | 06:38 |
comstud | meetup.com/openstack | 06:38 |
*** pvo has quit IRC | 06:42 | |
*** Predominant has joined #openstack | 06:44 | |
joshuamckenty | ah, my bad | 06:47 |
*** joshuamckenty has quit IRC | 06:47 | |
*** joshuamckenty has joined #openstack | 07:03 | |
*** pdelgallego has joined #openstack | 07:04 | |
*** arcane has quit IRC | 07:05 | |
*** Predominant has quit IRC | 07:06 | |
*** Predominant has joined #openstack | 07:06 | |
joshuamckenty | So I noticed we've got a bot running, any chance we know where the logs end up? | 07:09 |
joshuamckenty | I'd love to be able to read through scrollback and figure out what was going on while I was on airplanes | 07:09 |
*** jakedahn has joined #openstack | 07:12 | |
*** nettok has quit IRC | 07:12 | |
*** jsgotangco has quit IRC | 07:15 | |
*** jakedahn_ has quit IRC | 07:15 | |
joshuamckenty | ah, of course. There's carousing right now | 07:16 |
*** joshuamckenty is now known as joshuamckenty|aw | 07:16 | |
*** joshuamckenty|aw is now known as jmckenty|away | 07:16 | |
sureshgv | jmckenty|away: oh..okie | 07:25 |
sureshgv | thank you so much for the info | 07:25 |
*** z0 has quit IRC | 07:27 | |
*** jakedahn has quit IRC | 07:46 | |
*** abecc has quit IRC | 07:50 | |
*** ewan_ has joined #openstack | 07:58 | |
*** ttx has joined #openstack | 08:00 | |
*** calavera has joined #openstack | 08:04 | |
*** ralph has joined #openstack | 08:04 | |
ralph | which hypervisors will openstack support? | 08:05 |
*** miclorb_ has quit IRC | 08:18 | |
*** miclorb has joined #openstack | 08:18 | |
*** urandom has left #openstack | 08:20 | |
*** miclorb has quit IRC | 08:23 | |
*** giany has joined #openstack | 08:37 | |
*** adjohn_ has joined #openstack | 08:39 | |
adjohn_ | ralph: kvm now, xen later. | 08:39 |
giany | openstack works only on ubuntu? | 08:39 |
*** adjohn has quit IRC | 08:40 | |
*** adjohn_ is now known as adjohn | 08:40 | |
*** Predominant has quit IRC | 08:44 | |
*** Predominant has joined #openstack | 08:44 | |
*** metoikos has joined #openstack | 08:58 | |
*** z0 has joined #openstack | 09:07 | |
*** dmd17 has joined #openstack | 09:11 | |
*** dendro-afk is now known as dendrobates | 09:13 | |
dendrobates | anyone notice we got a wikipedia page: http://en.wikipedia.org/wiki/OpenStack | 09:14 |
*** sophiap has quit IRC | 09:16 | |
mattt | that didn't take long :) | 09:19 |
*** spackest has quit IRC | 09:22 | |
*** jhalogen has quit IRC | 09:29 | |
*** Predominant has quit IRC | 09:31 | |
*** Predominant has joined #openstack | 09:31 | |
*** dmd17 has quit IRC | 09:32 | |
*** bahadir has joined #openstack | 09:50 | |
*** avsm has joined #openstack | 10:16 | |
*** matclayton has joined #openstack | 10:24 | |
*** ttx has quit IRC | 10:37 | |
*** sophiap has joined #openstack | 10:41 | |
*** 52AABRVLU has quit IRC | 10:42 | |
*** sophiap has quit IRC | 10:43 | |
*** miclorb has joined #openstack | 11:13 | |
*** dabo has joined #openstack | 11:16 | |
*** jonesy_ has joined #openstack | 11:25 | |
jaypipes | yawn... | 11:37 |
*** miclorb has quit IRC | 11:38 | |
*** tobym has quit IRC | 11:40 | |
jonesy | jaypipes: you work for RS now?! | 11:44 |
* jonesy met jaypipes at OSCON... '08? | 11:45 | |
jonesy | anyway, congrats on the move. | 11:46 |
*** justinsheehy has quit IRC | 11:47 | |
jaypipes | jonesy: ya, RS now :) since March. | 12:02 |
jaypipes | jonesy: and thx! | 12:02 |
*** ttx has joined #openstack | 12:09 | |
*** philaway is now known as phildoc | 12:11 | |
jonesy | so, is there a patch submission howto for launchpad, or is it just a matter of posting a diff in a bug ticket, or mirroring a branch and pointing to a changeset.... ? | 12:23 |
* jonesy is new to launchpad, and to a lesser extent, bzr. | 12:23 | |
* jonesy has a doc patch. | 12:24 | |
mtaylor | jonesy: the general process is to push a branch to launchpad with the fix | 12:25 |
jonesy | ok, thanks. | 12:25 |
mtaylor | jonesy: and then to request a merge on that branch | 12:25 |
jonesy | right-o. | 12:26 |
jonesy | :) | 12:26 |
mtaylor | jonesy: I've got a link for you real quick... | 12:26 |
mtaylor | http://wiki.openstack.org/LifeWithBzrAndLaunchpad | 12:26 |
*** ChanServ sets mode: +o mtaylor | 12:26 | |
*** mtaylor changes topic to "Find out more and get involved here: http://www.openstack.org | Nova Docs: nova.openstack.org | Swift Docs: swift.openstack.org | Wiki: wiki.openstack.org | http://wiki.openstack.org/LifeWithBzrAndLaunchpad" | 12:26 | |
jonesy | oh man. I was reading that page, but didn't get to the 'patch queue manager' bit :-/ my bad. | 12:26 |
mtaylor | realized that should probably be prominently displayed for now | 12:26 |
mtaylor | heh. | 12:26 |
mtaylor | I should probably make a greatest-hits version | 12:27 |
jonesy | :) | 12:27 |
mtaylor | jaypipes: what are you doing awake? | 12:27 |
jaypipes | mtaylor: :) | 12:28 |
jaypipes | mtaylor: workin. | 12:28 |
mtaylor | jaypipes: heh | 12:28 |
*** justinsheehy has joined #openstack | 12:29 | |
mtaylor | jaypipes: if you get 3 secs, could you give me a quick merge review - I need to cut a new tarball for creiht but need a fix in first | 12:29 |
mtaylor | jaypipes: https://code.edge.launchpad.net/~mordred/swift/bug608469/+merge/30641 | 12:29 |
jaypipes | mtaylor: yep | 12:30 |
mtaylor | thanks bug | 12:30 |
mtaylor | bufd | 12:30 |
mtaylor | goddammit | 12:30 |
jaypipes | hehe | 12:31 |
*** binaryWarrior has quit IRC | 12:32 | |
*** tobym has joined #openstack | 12:36 | |
*** z0 has quit IRC | 12:36 | |
*** sophiap has joined #openstack | 12:38 | |
jonesy | you know, from an end-user perspective (someone looking for info or code for a project they're interested in), I've never liked launchpad, but I just did a merge proposal, and it was one of the easiest experiences I've had outside of maybe github. | 12:45 |
jonesy | r0ck! | 12:45 |
mtaylor | jonesy: YAY! | 12:52 |
clayg | good morning | 13:00 |
*** j2darcy has joined #openstack | 13:02 | |
*** h1tman has quit IRC | 13:02 | |
uvirtbot | New bug: #608725 in swift "swift should be internationalized/translated" [Undecided,New] https://launchpad.net/bugs/608725 | 13:05 |
*** z0 has joined #openstack | 13:07 | |
*** adjohn has quit IRC | 13:15 | |
*** binaryWarrior has joined #openstack | 13:20 | |
jonesy_ | mtaylor: I whipped this up this morning and am wondering if I should change the 'Swift Cluster' to 'Example Ring'. http://www.protocolostomy.com/swift-diag1.png | 13:21 |
mtaylor | ooh. nice job with uvirtbot whoever added that | 13:22 |
mtaylor | jonesy_: looking | 13:22 |
jonesy_ | it can stand alone, but I also want it to be a piece of a larger swift deployment diagram (which I haven't seen, is there one?) | 13:22 |
mtaylor | jonesy_: I'm not sure re: large diagram ... but yeah, I'm a fan of genercising graphics like that | 13:23 |
jonesy_ | it can help tie things together in more complex scenarios for folks who don't have their hands in this type of thing a lot. Guys like me bring stuff like that to people who wear ties :) | 13:24 |
anm_ | hey, I'm following http://wiki.openstack.org/NovaInstallFestInstructions, and it falls apart at: wget -q -O - http://173.203.107.207/ubuntu-lucid.tar | tar xSv | 13:28 |
anm_ | euca-bundle-image -i vmlinuz-2.6.32-23-server --kernel true | 13:28 |
anm_ | because euca-bundle-image says it requires a cert must be specified. | 13:29 |
anm_ | private key must be specified. | 13:29 |
anm_ | user must be specified. | 13:29 |
anm_ | ec2cert must be specified. | 13:29 |
*** sophiap has quit IRC | 13:30 | |
*** xfrogman5 has joined #openstack | 13:33 | |
anm_ | so im on the HackDay-Slices page trying to sort it out | 13:33 |
*** taf2 has left #openstack | 13:42 | |
* creiht stretches | 13:47 | |
creiht | morning | 13:47 |
g0rdy | howdy | 13:48 |
creiht | jonesy_: I'm currently working on deployment docs for swift | 13:48 |
creiht | and you aren't too far off with that graph | 13:49 |
jonesy_ | creiht: sweet. For my own use, can you answer the question I posed earlier? Should the label in the pic be changed? | 13:49 |
jonesy_ | yeah, but I think I'm a little off. :) | 13:49 |
creiht | hehe | 13:49 |
creiht | I think swift cluster is right | 13:49 |
creiht | does D stand for data, and P for proxy? | 13:50 |
jonesy | wow, I was way off ;-P | 13:50 |
mtaylor | creiht: 1.0.2 tarball released - also released to pypi | 13:50 |
jonesy | D is device, P is partition | 13:50 |
creiht | mtaylor: woot | 13:50 |
jonesy | it's a lower-level diagram I guess, creiht | 13:50 |
creiht | jonesy_: ahh | 13:50 |
jonesy | I'm building a larger diagram bottom-up | 13:51 |
jonesy | ....because it's easier to do that in omnigraffle (and other diag tools) | 13:51 |
jonesy | ...I can just import the smaller bits into the larger diag. | 13:51 |
creiht | hrm | 13:51 |
*** shadeslayer has joined #openstack | 13:52 | |
creiht | ok so I thought you were describing a cluster at first, but if you are describing the ring with that diagram, then you would rename the swift cluster to the ring | 13:52 |
jonesy | so, what I'm expressing there is the notion of separate zones inside one ring, which guarantees a partition only appears once in a given zone, and is replicated three times. | 13:52 |
creiht | jonesy: yeah that is starting to make sense now | 13:53 |
* creiht is still waking up | 13:53 | |
jonesy | creiht: well, yeah, but what I'm asking for is confirmation that the diagram is valid :) | 13:53 |
jonesy | ok, thanks creiht! | 13:53 |
creiht | so is there a reason there are several devices named D1? | 13:53 |
jonesy | so then I can make copies of that to represent other rings in a larger deployment diag. | 13:53 |
jonesy | creiht: yeah, I messed up and didn't change the D's when I copy/pasted | 13:54 |
creiht | k | 13:54 |
creiht | I think otherwsie it looks valid, though I would also mix partitions from the top row with the bottom row so people wouldn't think that partitions are mapped in some sort of logical order | 13:55 |
jonesy | good idea. Will do. | 13:55 |
creiht | It is a different way of looking at it | 13:55 |
creiht | I would also add a legend so it makes it obvious what the parts are | 13:56 |
*** shadeslayer has left #openstack | 13:57 | |
creiht | Usually when we diagram the ring, we have a ring of partitions, and then show one of those partitions getting mapped to 3 devices, each in a different zone | 13:57 |
jonesy | creiht: if it's not the intended way of looking at it, then I'll do more work on the docs after I do an actual setup, 'cos that diagram is what's in my head as a result of reading the docs. | 13:57 |
creiht | Will has a diagram in his presentation for OSCON that was pretty good | 13:57 |
creiht | jonesy: I think it is more of just a different way of looking at it... nothing wrong with that | 13:57 |
g0rdy | creiht: Is Will's diagram available anywhere ? | 13:57 |
*** pcrews has joined #openstack | 13:58 | |
* creiht goes to check again | 13:58 | |
* creiht doesn't see it on the oscon site yet | 13:58 | |
jonesy | creiht: an actual ring in the diagram would be good, but my plan is to put the bits that use the ring in the same diagram, up to the proxy server, and I only have so much space :) | 13:59 |
jonesy | if Will's diagram includes all that, I'd love to get it. | 13:59 |
creiht | Next time I hear from him, I will ping him about the presentation | 14:00 |
g0rdy | Excellent, thanks much - I'm with Washington University's Genome Center in St Louis, and we are very intrigued with this | 14:01 |
creiht | g0rdy: great... feel free to ask any questions | 14:01 |
creiht | cool... swift is now in pypi | 14:01 |
creiht | http://pypi.python.org/pypi/swift/1.0.2 | 14:01 |
jaypipes | justinsb: ping me if you want to chat about the BasicModel stuff...I've got lots of thoughts ;) | 14:02 |
creiht | jonesy: looking at your merge proposal for the doc changes now | 14:06 |
jonesy | cool. Hope it's useful. | 14:07 |
jaypipes | dendrobates: ping | 14:16 |
uvirtbot | New bug: #606340 in nova "test_too_many_addresses is not idempotent" [Low,Fix released] https://launchpad.net/bugs/606340 | 14:31 |
uvirtbot | New bug: #608772 in nova "Failure in NetworkTestCase for range_allocation and " [Undecided,New] https://launchpad.net/bugs/608772 | 14:31 |
*** bahadir_ has joined #openstack | 14:37 | |
jonesy | wow. Just.... wow. http://mail.python.org/pipermail/python-dev/2010-July/102189.html | 14:39 |
* creiht looks to see if it is 4/1 | 14:40 | |
*** bahadir has quit IRC | 14:40 | |
*** silassewell has joined #openstack | 14:41 | |
gholt | I would laugh about it, but I'm too sad. | 14:41 |
*** pvo has joined #openstack | 14:42 | |
cory_ | I'd be sad about it, but it's too funny | 14:43 |
*** xfrogman5 has quit IRC | 14:45 | |
*** xfrogman5 has joined #openstack | 14:46 | |
*** bahadir has joined #openstack | 14:48 | |
*** avsm has quit IRC | 14:48 | |
*** avsm has joined #openstack | 14:48 | |
*** bahadir has quit IRC | 14:49 | |
*** bahadir_ has quit IRC | 14:50 | |
j2darcy | So, what is testYourmomContainer all about anyway? | 14:55 |
creiht | lol | 14:56 |
creiht | redbo: would you like to comment? :) | 14:56 |
creiht | j2darcy: is that in the functests? | 14:57 |
creiht | yeah | 14:58 |
alekibango | that name sounds terrible. having mom in a container is evil. | 14:58 |
creiht | so um... that was an easter egg that was removed a while ago | 14:58 |
jonesy | or not :) | 14:58 |
creiht | I guess they forgot to remove the test for it | 14:58 |
jonesy | that's awesome. | 14:59 |
j2darcy | I was amused. | 15:00 |
creiht | hrm | 15:01 |
*** xfrogman5 has quit IRC | 15:02 | |
creiht | will be removed shortly :) | 15:04 |
anm_ | i wish i could put my mom in a container sometimes | 15:05 |
anm_ | Im trying to walk through http://wiki.openstack.org/NovaInstallFestInstructions, and i got all the way to the steps where you bundle an image, and it fails: | 15:06 |
anm_ | euca-bundle-image -i vmlinuz-2.6.32-23-server --kernel true | 15:06 |
anm_ | requires more parameters than this | 15:06 |
anm_ | I'll be happy to update if I knew what to put | 15:06 |
anm_ | I think it mneeds some environment variables that are not set | 15:07 |
anm_ | based on the error in euca-bundle-image: | 15:08 |
anm_ | Euca2ools will use the environment variables EC2_URL, EC2_ACCESS_KEY, EC2_SECRET_KEY, EC2_CERT, EC2_PRIVATE_KEY, S3_URL, EUCALYPTUS_CERT by default. | 15:08 |
*** JordanRinke has quit IRC | 15:08 | |
anm_ | does anyone have a hint? | 15:11 |
mtaylor | creiht: I also set up a nightly builds ppa for swift | 15:15 |
creiht | mtaylor: cool | 15:15 |
creiht | mtaylor: so is there a reason that we need the swift.egg-info dir in the repo? | 15:15 |
mtaylor | creiht: https://edge.launchpad.net/~swift-core/+archive/trunk/+packages | 15:16 |
mtaylor | creiht: yes and no. (and to be clear, I'm not particularly pleased with it being there either) | 15:16 |
mtaylor | creiht: here's the thing... | 15:16 |
*** tobym has quit IRC | 15:17 | |
*** Predominant has quit IRC | 15:17 | |
mtaylor | creiht: there's a couple of ways to manage the whole debian dir as an upstream who maintains it. none of them are fully pleasant. | 15:18 |
creiht | It's debian, what did you expect? :) | 15:18 |
mtaylor | creiht: the one I've found to be the best when upstream is using bzr is to use bzr-builddeb and the import-dsc/merge-upstream workflow - but this puts the contents of the released tarball into the branch... | 15:18 |
mtaylor | creiht: "normally" ... one would have this be a branch from trunk that does not get merged back in | 15:19 |
mtaylor | so you have trunk, and you have a packaging branch, and it's not ugly that the packaging branch has a few more files checked in | 15:19 |
mtaylor | but then you normally remove the debian dir from trunk | 15:19 |
mtaylor | I _like_ having the debian dir in trunk, because then normal devs can update stuff in the packaging - but I may be weird | 15:20 |
chmouel | i think debian advise to keep debian dir not in upstream but in an another branch | 15:21 |
mtaylor | yes. | 15:21 |
mtaylor | we could certainly adopt that model... (it's how I do all of my other projects, honestly) but I think since you already had an internal debian dir, it would be worth poking at options that might keep that workflow around | 15:22 |
creiht | I don't think I'm adverse to having the debian dir in a separate branch | 15:22 |
jmckenty|away | anm_: I haven't checked the instructions recently, but sourcing the novarc file should set the necessary environment variables | 15:23 |
anm_ | ok, let me check that out - | 15:23 |
mtaylor | ok. how about I put up a couple of branches - a trunk minus files that aren't needed and a trunk+packaging and let you look at them? | 15:23 |
jmckenty|away | mtaylor: I like having the debian dir in trunk, but I'm not attached to it | 15:23 |
creiht | mtaylor: that would be great, thanks! | 15:23 |
*** RandalSchwartz has left #openstack | 15:23 | |
chmouel | mtaylor: for all other public repo for rackspace we have used that model based on exlt recommendation | 15:24 |
anm_ | jmckenty|away: looks like thats whats needed - thanks | 15:24 |
* anm_ updates wiki | 15:24 | |
jmckenty|away | np | 15:24 |
mtaylor | cool. I'll make some branches for folks to look at | 15:24 |
mtaylor | exlt: you have non-merged debian dir changes right now, yes? | 15:24 |
*** calavera has quit IRC | 15:25 | |
anm_ | whan i click in the serach box on the OpenStack wiki, it takes me to the OpenStack.org home page :| | 15:25 |
mtaylor | fail | 15:26 |
*** ewan_ has quit IRC | 15:26 | |
gholt | anm_: You're now part of exlt's club (he's been mentioning that daily for a bit as well). | 15:28 |
gholt | I'm not sure who "owns" that page to ping them... | 15:29 |
anm_ | gholt: sorry, I just started with OpenStack Yesterday! | 15:29 |
gholt | No, np, weird things amuse me, and that particular bug is one of those things. :) | 15:29 |
chmouel | can always write a greasemonkey to 'fix' it | 15:31 |
chmouel | greasemonkey script.. | 15:31 |
*** samppah_ has joined #openstack | 15:33 | |
*** spackest has joined #openstack | 15:37 | |
*** sirp1 has joined #openstack | 15:39 | |
*** metoikos has quit IRC | 15:39 | |
jonesy | hm. I have a client that uses isilon storage, which is NAS (they access it using NFS). For some reason I thought that swift assumed direct attached storage, but I must've gotten that idea from the code or something 'cos I can't find mention of it in the docs. Clue? | 15:41 |
j2darcy | Not sure why it would care, or even how it would tell, though it is kind of redundant running a distributed object store on top of clustered NAS. | 15:42 |
*** JordanRinke has joined #openstack | 15:42 | |
jonesy | j2darcy: then why doesn't rackspace just use NFS? ;-P | 15:42 |
exlt | mtaylor: I started working on init script changes, but got a little wrapped up in LDAP+ssh-keys | 15:43 |
creiht | jonesy: swift is designed to use commodity hardware | 15:43 |
mtaylor | exlt: ok - well, I'll have a new branch for you to look at (probably after my flight) ... will have to diff/patch/re-apply to it | 15:43 |
creiht | For example are storage nodes are fairly basic 4u servers with 24 drives | 15:43 |
exlt | wrt init scripts, would we want to stick with using swift-init instead of start-stop-daemon? | 15:44 |
jonesy | I think there's still benefit over and above using NFS. They have issues related to managing large numbers of files in a million directories for example. | 15:44 |
j2darcy | jonesy: I'll best cost is one reason (open-source clustered NFS is in a sorry state) and also needed a slightly different feature set. But they're right here to answer more authoritatively. | 15:44 |
creiht | That said, there isn't anything that prevents you from using DAS with it | 15:44 |
*** pvo_ has joined #openstack | 15:44 | |
*** pvo_ has joined #openstack | 15:44 | |
*** pvo has quit IRC | 15:44 | |
*** pvo_ is now known as pvo | 15:44 | |
mtaylor | exlt: would we _not_ want to use start-stop-daemon? | 15:44 |
creiht | An actually interesting use case is the possibility of using a 1U or blade server connected to dense DAS storage units | 15:46 |
*** mtaylor has quit IRC | 15:46 | |
exlt | mtaylor: I started looking at swift-init, and it has all the needed pid and actions, etc - that was really my question - what is the best way to do it? start-stop-daemon is the debian way and I think keeping pid locations, etc. the same as swift-init would be a must, but leave primary use of swift-init to source installs | 15:47 |
exlt | and, he left.. | 15:47 |
gholt | Hehe | 15:47 |
creiht | Swift would not work very well on NFS (not only for for performance reasons) but also due to how we use the filesystem for things like locking | 15:48 |
*** Predominant has joined #openstack | 15:48 | |
exlt | screen ftw.. conversational history on irc is one of the killer features for me - without it, it's like missing messages in a long email thread | 15:49 |
spackest | so, wrote some scripts and a blog post about getting saio up and running | 15:49 |
spackest | it is here http://blog.spack.net/?p=162 | 15:49 |
spackest | would be happy to contribute the scripts | 15:49 |
exlt | anyway.. creiht, do you have any thoughts on init? | 15:50 |
spackest | there are three http://blog.spack.net/saio.sh http://blog.spack.net/saio-guest-1.sh http://blog.spack.net/saio-guest-2.sh | 15:50 |
j2darcy | creiht: What about the locking particularly? | 15:51 |
creiht | j2darcy: for example sqlite uses posix file locking to do it's locking, and that can be very problematic on NFS | 15:51 |
notmyname | sqlite 3.7 released with WAL http://www.sqlite.org/releaselog/3_7_0.html | 15:51 |
creiht | woot! | 15:52 |
alekibango | woot really | 15:52 |
j2darcy | The stuff I'm working on will probably run on top of Swift or GlusterFS separately. Don't see any particular need to stack them, but it would be good to know of any reasons I couldn't.\ | 15:52 |
j2darcy | I'll have to check the status of sqlite3 on gluster. I'm pretty sure it's supposed to work. | 15:53 |
chmouel | is GlusterFS support xattr? | 15:53 |
creiht | spackest: hah... that's awesome... installing swift on Amazon EC2 :) | 15:53 |
j2darcy | chmouel: Yep. Depends heavily on them, in fact. | 15:53 |
spackest | creiht: thanks, we were trying bare metal, but it is pretty nice to be able to just start from scratch | 15:53 |
chmouel | that script could work on a anything really good job spacktest | 15:53 |
*** ianl has joined #openstack | 15:53 | |
spackest | thanks | 15:54 |
spackest | you're welcome to link to the post on here http://swift.openstack.org/development_saio.html if you like | 15:55 |
creiht | spackest: I'll take a look at the scripts shortly | 15:56 |
*** littleidea has joined #openstack | 15:56 | |
*** j2darcy has quit IRC | 15:56 | |
alekibango | what does this line mean? why 18 3 1 ?? >> swift-ring-builder object.builder create 18 3 1 | 15:57 |
creiht | spackest: by the way, holoway is also contributing a chef script that sets up SAIO | 15:58 |
alekibango | i think more about debian FAI setup to configure the cluster. that would help | 15:58 |
*** root__ has joined #openstack | 15:58 | |
*** root__ has quit IRC | 16:00 | |
creiht | exlt: back to init scripts, sorry | 16:01 |
*** ttx has quit IRC | 16:01 | |
gholt | alekibango: 18 == number of ring partitions (2**18), 3 == replicas to make of each object, 1 == hours to wait before moving a partition a second time. | 16:02 |
chmouel | j2darcy: i started to read about GlusterFS but I am wondering why would you want to run swift on top of it ? | 16:02 |
creiht | swift-init is useful when you haven't installed it via packaging | 16:02 |
creiht | for example when you are doing development | 16:02 |
chmouel | creiht: or other OS than debian based ones | 16:03 |
creiht | I think we have some stuff in the debian packaging for creating init scripts | 16:03 |
spackest | chef script would be nice, too | 16:04 |
creiht | but not sure how comprehensive it is | 16:04 |
creiht | the ops guys got used to using swift-init :) | 16:04 |
creiht | so I'm all for having init scripts, but would also like to keep swift-init | 16:04 |
spackest | so, is there a swift client that will allow me to mess around with swift a bit, without having to write code? | 16:04 |
*** matclayton has left #openstack | 16:04 | |
creiht | spackest: there is the st commandline tool that comes with swift | 16:05 |
creiht | which will allow you to upload and download files | 16:05 |
creiht | get stats, etc | 16:05 |
creiht | just run st, and it should give you all the options available | 16:05 |
silassewell | eww, swift on glusterfs sounds terrible | 16:06 |
creiht | any cloudfiles tools that allow you to set the auth url should work as well | 16:06 |
gholt | You know, I have tried fireuploader in a while; wonder if he lets you change the auth url. | 16:06 |
gholt | s/have/haven't/ | 16:06 |
creiht | that would be nice | 16:06 |
spackest | I bet I could have my script set ST_AUTH, ST_USER, and ST_KEY | 16:08 |
jonesy | ah. The code leads once again to enlightenment. :) | 16:08 |
*** pvo_ has joined #openstack | 16:08 | |
*** avsm has quit IRC | 16:09 | |
creiht | jonesy: care to share the enlightenment? :) | 16:09 |
jonesy | in the arch overview, I'd propose that the proxy server is the first thing to be covered, working toward the ring, following the path of a request handled by the proxy server. | 16:10 |
* jonesy has too much time on his hands while long-running tests grind on. | 16:10 | |
creiht | jonesy: ahh | 16:11 |
jonesy | understanding the ring requires, I think, prior knowledge of all of the other stuff on the page, which is the context for the concept of the ring, and its purpose. | 16:11 |
creiht | yeah I can see where you are coming from | 16:11 |
spackest | fwiw, from an outsider, that page is a little tough to grok ;) | 16:12 |
*** pvo has quit IRC | 16:12 | |
*** pvo_ is now known as pvo | 16:12 | |
exlt | creiht: cool - thanks. I'll go down the path of using start-stop-daemon in the debian init scripts, and I'll mirror pid file placements, etc - it should "just work" regardless of which the user actually calls | 16:12 |
jonesy | well, lemme see what I can do. | 16:12 |
creiht | exlt: sounds good... thanks | 16:12 |
justinsb | jaypipes: Just catching up on IRC & bugs... looking at bug 608772, is that against trunk or from abstract-data-store? I started checking exit codes in abstract-data-store, because I ran into some problems in dhcpbridge. Maybe I broke things if you're on my branch? | 16:12 |
uvirtbot | Launchpad bug 608772 in nova "Failure in NetworkTestCase for range_allocation and " [Undecided,New] https://launchpad.net/bugs/608772 | 16:12 |
exlt | creiht: dunno if I'll actually get to that, today, though.. | 16:13 |
creiht | spackest: yeah... we are getting there, if you think of things that will make the docs more clear, let me know | 16:13 |
jaypipes | justinsb: trunk. totally unrelated to your stuff | 16:13 |
creiht | exlt: oh yeah man, no rush :) | 16:13 |
*** Guest345 has joined #openstack | 16:13 | |
spackest | so based on http://pastebin.com/yqmpeBkF what would I use for ST_AUTH, ST_USER and ST_KEY | 16:14 |
creiht | ST_AUTH=http://127.0.0.1:11000/v1.0 | 16:14 |
creiht | ST_USER=test:tester | 16:14 |
justinsb | jaypipes: That's a relief! If you push it up though, I'm happy to try to run it on mine to see if it 'works for me' | 16:14 |
creiht | ST_KEY=tkb0618b37-a750-4aee-93e5-abb5d3a222d5 | 16:14 |
justinsb | jaypipes: As I've been playing with it anyway! | 16:14 |
creiht | note that the key can change | 16:14 |
gholt | key=password | 16:15 |
creiht | oh | 16:15 |
creiht | heh | 16:15 |
creiht | ignore what I just said :) | 16:15 |
jaypipes | justinsb: push what? 608772 is a bug I'm trying to tackle right now to no avail :( vish is trying to assist, but no luck yet :( | 16:15 |
creiht | ST_KEY=testing | 16:15 |
creiht | gholt: thanks | 16:15 |
creiht | :) | 16:15 |
justinsb | jaypipes: Ah, so totally clean trunk? | 16:15 |
jaypipes | justinsb: yeah :( been happening ever since I started working on nova. | 16:16 |
*** dabo has left #openstack | 16:16 | |
jaypipes | justinsb: I'm sure it's some sort of sys.path issue, but I'm a little frustrated right now and prolly with work on something else for a bit ;) | 16:16 |
justinsb | jaypipes: OK, well totally out of my depth here, but are your PYTHONPATH and/or /usr/local/lib/python2.6/dist-packages/nova.pth pointing to the right branch? | 16:21 |
justinsb | jaypipes: Or we could talk about the data model ... much more fun! | 16:21 |
jaypipes | justinsb: yeah, I think I'd rather talk about the data model for right now! :) | 16:22 |
jaypipes | justinsb: did you read my code review on your merge prop? | 16:22 |
justinsb | jaypipes: Let me read it again so it's fresh | 16:23 |
jmckenty|away | jaypipes: Vish and I created the dhcp_flagfile thing to fix this specific issue | 16:24 |
jmckenty|away | justinsb: and jaypipes : just sent comments on the data model stuff | 16:25 |
jaypipes | jmckenty|away: yep, just read em :) so...re: ORMs | 16:25 |
*** gustavomzw has joined #openstack | 16:26 | |
jaypipes | jmckenty|away: the "canonical" ORM for Python is probably SQLAlchemy. The issue we'll have with it is, well, it's SQL-based :) | 16:26 |
jmckenty|away | right, that's why I was asking about OMs :) | 16:26 |
jmckenty|away | jaypipes: <module 'nova.compute.network' from '/usr/lib/pymodules/python2.6/nova/compute/network.py'> is your clue | 16:26 |
jmckenty|away | that's loading the network module that was installed from a package or easy_install | 16:27 |
jmckenty|away | not the one from the trunk that you're trying to work in | 16:27 |
jaypipes | jmckenty|away: The BasicModel that's in there right now is OK and with a few adaptations (like making an ABC for the datastore driver), I think we'd be 90% there. | 16:27 |
jaypipes | jmckenty|away: so...how to fix? suggestions? | 16:27 |
jmckenty|away | jaypipes: which is the reason for the flagfile | 16:27 |
jmckenty|away | do you have a nova-dhcp.conf flagfile? | 16:28 |
jaypipes | jmckenty|away: not that I know of... | 16:28 |
spackest | hmm, so can I create a container with st? | 16:28 |
jaypipes | jmckenty|away: is it in the the branch somewhere? | 16:28 |
jmckenty|away | k, two secs | 16:28 |
jaypipes | jmckenty|away: thx mate :) | 16:28 |
justinsb | For the DataModel, I think there are 3 cases that are co-mingled currently: 1) Creating a new object 2) Loading an object by PK and 3) Building an object from datastore results. | 16:29 |
justinsb | I think there should be 3 separate calls, for clarity & safety & efficiency. | 16:29 |
*** adrian_otto has joined #openstack | 16:29 | |
jaypipes | justinsb: agreed, but with 3) Building multiple objects from datastore results | 16:29 |
justinsb | In particular, for a SQL back-end, we want to avoid the 1+N reads problem | 16:30 |
jaypipes | right | 16:30 |
jmckenty|away | justinsb: can you clarify how 2 is different from 3? | 16:30 |
jaypipes | jmckenty|away: 1 vs. more than 1 object... | 16:30 |
jmckenty|away | e.g., does 3 always call 2 eventually? | 16:30 |
jmckenty|away | ah, gotcha | 16:30 |
jmckenty|away | k | 16:30 |
jaypipes | yep | 16:30 |
jaypipes | jmckenty|away: I tried to show some example code because I think it's easier to show in code than describe some of this stuff :) | 16:31 |
jmckenty|away | While I like the idea of both a) being able to specify a filter with multiple params, and b) getting a good generator pattern, | 16:31 |
jmckenty|away | I'm not sold on the syntax of having the base class support .next | 16:31 |
jmckenty|away | I don't object, I just want to see a couple of other cases stubbed out | 16:32 |
jaypipes | jmckenty|away: k. you prefer the all() methods? | 16:32 |
jaypipes | jmckenty|away: classmethods, that is.. | 16:32 |
jmckenty|away | the only advantage of those | 16:32 |
jmckenty|away | (method or property) | 16:32 |
spackest | I think this could be a little more clear Usage: st [options] upload [options] container file_or_directory [file_or_directory] [...] | 16:32 |
jmckenty|away | is that the external callers don't need to know the name of the 'field' that you want to search on | 16:32 |
jmckenty|away | it's conceptual, and exposed in the class def | 16:33 |
jmckenty|away | e.g., 'all', or 'by_project' | 16:33 |
jmckenty|away | but that's a pretty weak defense | 16:33 |
jaypipes | jmckenty|away: yes, I understand that argument. | 16:33 |
jmckenty|away | for something that admittedly doesn't extend nicely | 16:33 |
justinsb | The other advantage is indexing... | 16:33 |
justinsb | Particularly for Redis, there's a world of difference between a scan of all objects and something we've indexed manually | 16:34 |
justinsb | I mean it's still bad in SQL, but it's catastrophic in Redis | 16:34 |
jaypipes | justinsb: indexing is an implementation detail that can be hidden in the datastore driver... | 16:34 |
spackest | I keep getting ClientException: Object PUT failed: http://127.0.0.1:8080/v1/6267e345-652d-487e-8d4c-ee7c83f544c8/directory/my_dir 503 Service Unavailable and I am at a bit of a loss | 16:34 |
*** jdmaturen has joined #openstack | 16:34 | |
justinsb | jaypipes: I totally agree... we should hide indexing | 16:34 |
*** jdmaturen has quit IRC | 16:34 | |
spackest | an example or two for st upload would be nice | 16:34 |
creiht | spackest: first thing is to check to make sure the services are running | 16:34 |
justinsb | jaypipes: (Not least because I love reimplementing relational DBs on top of key-value stores) :-) | 16:34 |
*** jdmaturen has joined #openstack | 16:34 | |
jaypipes | justinsb: :) | 16:35 |
jmckenty|away | I *do* love the symmetry between the 'filters' and the context object that's used elsewhere | 16:35 |
justinsb | jaypipes: But the advantage of having specific by_project methods is that you can see more easily which secondary indexes you need | 16:35 |
jmckenty|away | esp. since (often), the context 'will' be the filter | 16:35 |
spackest | st stat works and it looks like I made two containers, container and directory :) | 16:35 |
jaypipes | jmckenty|away: come to think of it, there's really no reason both interfaces couldn't exist... | 16:35 |
creiht | spackest: and then check /var/log/syslog (or messages if it isn't in syslog) to see what errors are there | 16:35 |
justinsb | Perhaps the | 16:35 |
justinsb | Perhaps the 'filters' should be private | 16:35 |
jmckenty|away | jaypipes: http://bazaar.launchpad.net/~hudson-openstack/nova/trunk/annotate/head:/debian/nova-dhcp.conf | 16:35 |
jaypipes | jmckenty|away: meaning, the filtering constructor could stay around, and stuff like by_project() would be a simple wrapper around it.. | 16:36 |
justinsb | And called by the by_project/by_X methods | 16:36 |
jmckenty|away | I have to run for a meeting, alas | 16:36 |
jaypipes | justinsb: sure, though there's really no "privacy" in Python ;) | 16:36 |
spackest | guess I just mean basic syntax for adding a container and then adding something to that container | 16:36 |
jmckenty|away | I like that pattern, though | 16:36 |
jaypipes | jmckenty|away: cya | 16:36 |
jmckenty|away | so BasicModel gets filter | 16:36 |
justinsb | jaypipes: True, but the underscore thing serves as a big warning! | 16:36 |
jaypipes | and subclasses can have whatever they like :) | 16:36 |
jmckenty|away | and the object-specific properties are thin, index-aware wrappers | 16:36 |
jmckenty|away | I like | 16:37 |
jaypipes | jmckenty|away: precisely | 16:37 |
creiht | spackest: I don't think there is a create container command in st, but if you upload an object to a container that doesn't exist with st, it will create the container first | 16:37 |
jaypipes | justinsb: ack on that :) | 16:37 |
jmckenty|away | jaypipes: that's link will give you which flags are required for dhcpleasor to work | 16:37 |
jaypipes | justinsb: I come from C++, where there actually *is* privacy ;) | 16:37 |
jaypipes | jmckenty|away: thx. cheers! | 16:37 |
jmckenty|away | talk later | 16:37 |
*** jmckenty|away has quit IRC | 16:37 | |
justinsb | jaypipes: I come from Java/C++, and am really missing static typing right now | 16:37 |
jaypipes | justinsb: heh, yeah :) | 16:37 |
creiht | stay a while, and you will get used to it :) | 16:38 |
jaypipes | justinsb: though I do love Python. such an elegant language. | 16:38 |
jaypipes | justinsb: compared to C++, which can be, well, a difficult red-headed stepchild at times. | 16:38 |
justinsb | jaypipes: Well, I want to stay away from the language wars, but the self. and cls. thing is bugging me right now | 16:38 |
jaypipes | justinsb: hehe, and that's better than "this"? ;P | 16:39 |
*** Thweetch has joined #openstack | 16:39 | |
justinsb | jaypipes: Anyway, I think we've got consensus on the model classes | 16:39 |
jaypipes | justinsb: there's actually nothing special about the word "self" in Python...it's just a convention. You could replace self with "me" if you really felt like pissing everyone off ;) | 16:39 |
justinsb | jaypipes: That they can use your super-magic filters, which we'll keep 'private', wrapped by public methods named by_project etc | 16:40 |
jaypipes | justinsb: yeah, I agree on the data model stuff. How do you want to split up the work? I'm all ears. | 16:40 |
justinsb | jaypipes: Do you want to do the magic kwargs stuff? | 16:40 |
jaypipes | sure thing :) | 16:40 |
justinsb | jaypipes: I might attack the 1+n problem | 16:40 |
jaypipes | justinsb: perfect. | 16:40 |
*** _cerberus_ has quit IRC | 16:40 | |
*** _cerberus_ has joined #openstack | 16:40 | |
justinsb | jaypipes: I think we should also have a benchmark suite for the datastore | 16:41 |
creiht | spackest: did you get it working? | 16:41 |
jaypipes | justinsb: I'll just push to LP and ping you when I'm done...off to get another coffee... | 16:41 |
jaypipes | justinsb: yep, that is planned. | 16:41 |
justinsb | jaypipes: Just to make sure it can scale to Rackspace's needs | 16:41 |
justinsb | jaypipes: Might slay some sacred cows as well ;-) | 16:41 |
jaypipes | justinsb: yep, I'm creating something similar to drizzle-automation for openstack that should stress the systems pretty well. | 16:41 |
jaypipes | hehe :) | 16:41 |
justinsb | jaypipes: Yes, I think plugging in Drizzle / MySQL should be next | 16:42 |
spackest | yip, had a bug, will try a fresh deploy. missed a <your-user-name> or the like or two | 16:42 |
justinsb | jaypipes: But probably best to figure out all this with SQLite first! | 16:42 |
creiht | spackest: ahh good | 16:42 |
*** tomjones has joined #openstack | 16:43 | |
justinsb | jaypipes: See you after your coffee! That sounds like a good idea right now... | 16:43 |
spackest | hoping to add some actual interaction, so I (or other users) can see a file going into swift | 16:43 |
justinsb | jaypipes: I'm happy to give you commit access to the abstract-data-store branch if you let me know how I can do that (and if that's the right thing to do)! | 16:45 |
* jonesy_ returns from lunch and scrolls | 16:48 | |
jonesy_ | that was the weakest language war I've ever witnessed. | 16:48 |
jonesy | thanks for updating the review status, creiht -- I was probably supposed to do that. | 16:50 |
*** pvo has quit IRC | 16:57 | |
*** JordanRinke has left #openstack | 16:58 | |
*** JordanRinke has joined #openstack | 16:59 | |
spackest | yeah! it works :) | 17:01 |
spackest | I can put things in and get them out again | 17:01 |
*** jeromatron has joined #openstack | 17:06 | |
*** stewart has joined #openstack | 17:09 | |
*** christiekoehler has joined #openstack | 17:09 | |
*** Disputin has joined #openstack | 17:10 | |
*** Disputin has quit IRC | 17:10 | |
*** pvo has joined #openstack | 17:11 | |
*** pvo has joined #openstack | 17:11 | |
*** stewart has quit IRC | 17:11 | |
*** jhalogen has joined #openstack | 17:12 | |
*** pvo has quit IRC | 17:12 | |
*** tobym has joined #openstack | 17:15 | |
*** christiekoehler has quit IRC | 17:15 | |
chmouel | i have run the script btw: on a fresh vm and aside of a missing cd /home/${MY_USER} before wgetting the other script it seems to work well good job | 17:15 |
creiht | jonesy: nah... I forgot to do that | 17:15 |
*** tobym has quit IRC | 17:15 | |
*** tobym_ has joined #openstack | 17:17 | |
*** dendrobates is now known as dendro-afk | 17:22 | |
jaypipes | justinsb: trying to fix my paths issue right now...after that, on to the datastore code :) | 17:23 |
justinsb | jaypipes: Cool. I do have a question on the best Pythonic way to do something... | 17:25 |
jaypipes | justinsb: shoot. | 17:25 |
justinsb | jaypipes: I'm trying to avoid the 1+N problem, so want to pass in the ID and the state map into the constructor | 17:25 |
justinsb | jaypipes: What's the best way to do that with the *args, *kwargs pattern? | 17:26 |
justinsb | jaypipes: Pass the state as an optional named parameter? | 17:26 |
chmouel | any idea why i am getting that when running the func test / http://pastebin.com/9PRBiTK8 | 17:27 |
jaypipes | justinsb: pass *args and **kwargs in the constructor, and ensure args[0] is the identifier (first unnamed param) | 17:27 |
chmouel | i don't know where that X11 access come from | 17:27 |
creiht | chmouel: are you running X11 on the box you are testing on? | 17:27 |
chmouel | creiht: noip.. i have ubuntu server installed nothing X11 there | 17:28 |
justinsb | jaypipes: And pass my 'state' hash as the second unnamed parameter? | 17:28 |
justinsb | jaypipes: def _build_object_from_data_row(self, kls, row): | 17:29 |
justinsb | state = self._build_state_from_row(row) | 17:29 |
justinsb | return kls(row['id']) #, state) | 17:29 |
jaypipes | justinsb: I'd pass it as the first named param. | 17:29 |
jaypipes | justinsb: kls(row['id'], state=state) | 17:29 |
creiht | chmouel: haven't seen that before | 17:29 |
justinsb | jaypipes: Cool, thanks! Then I guess the filtering code will need to check for the magic 'state' parameter and not treat it as a filter... | 17:30 |
justinsb | jaypipes: Might call it __state to avoid name collisions! | 17:30 |
*** abecc has joined #openstack | 17:30 | |
jonesy | justinsb: I don't have all the background, but if that results in like 50 magical attributes, I wouldn't call it the most Pythonic way. | 17:30 |
justinsb | jonesy: Should just be one magical attribute... the one that contains the map of all the object's properties. But suggestions welcome! | 17:31 |
jaypipes | justinsb: the filter code will only check for non-sequence params... | 17:31 |
jonesy | yeah, and if it's just one that you're returning from the method, no big deal | 17:31 |
justinsb | jaypipes: Forgive my Python noob-ish ness. I'll read up on this stuff! | 17:32 |
*** abecc has quit IRC | 17:32 | |
jaypipes | justinsb: I planned on getting rid of the state attribute entirely. it's pointless. we should just use self.__getattr__ and self.__dict__[key] | 17:33 |
jaypipes | justinsb: no prob :) | 17:33 |
jaypipes | justinsb: the state dict is not really necessary since BasicModel inherits from object and therefore has an attribute dict already... | 17:34 |
jaypipes | justinsb: in other words, the BasicModel really should just be plan old data, with a few wrappers/constructors for convenience. | 17:34 |
jaypipes | s/plan/plain/ | 17:34 |
justinsb | jaypipes: Ah... cool. I'm guessing maybe the problem is that sometimes there are variables that they don't want to save to the datastore maybe? But who knows...! | 17:35 |
jaypipes | justinsb: there might be...but just prefix em with _... | 17:36 |
justinsb | jaypipes: Sounds like you want POPOs (Plain-Old-Python-Objects)? ;-) | 17:36 |
jaypipes | justinsb: as of right now, though, there aren't any "hidden" variables. | 17:36 |
jaypipes | justinsb: :) | 17:36 |
chmouel | creiht: yeah weird, if i connect with ssh -x and remove libx11-6 and connect with ssh -x it seems to don't do that anymore | 17:37 |
creiht | heh | 17:37 |
justinsb | jaypipes: So long term the state can be passed in using your new named-arg constructor pattern | 17:37 |
justinsb | jaypipes: But in the short term we'll just get something that works! | 17:38 |
jaypipes | justinsb: yep. though we'll need to review Ewan's code and get that merged in first...he changes a number of things that will conflict with our work. | 17:38 |
chmouel | can we add that by the way http://pastebin.com/bBd7AFmL to allow to pass -v to the unit tests or other args | 17:39 |
jonesy | chmouel: have you tried just using ssh -Y ? | 17:40 |
jonesy | instead of -x | 17:40 |
creiht | chmouel: that is a good idea, why don't you send a merge proposal? :) | 17:42 |
chmouel | creiht: damn i'll need to learn launchpad for a two char diff :) | 17:45 |
creiht | :) | 17:45 |
chmouel | jonesy: seems to be it when I have the xauth installed (and accessing without -Y) its trying to access some X11 related stuff (or at least to libX11 from the strace) cf: chmouel@lutece:~$ ssh 172.16.42.129 | 17:47 |
jonesy | chmouel: use the wiki page. Got me through pretty painlessly. http://wiki.openstack.org/LifeWithBzrAndLaunchpad | 17:47 |
chmouel | cf: i mean http://pastebin.com/HBPk9Vug | 17:47 |
chmouel | jonesy: yeah i am just being lazy but tks | 17:47 |
jonesy | chmouel: iirc, the error you're getting, and the fact that you don't get it using -Y means that whatever you're running requires root access to an X11 resource, and you're not root (or, whoever you need to be to access the resource, usually a window in my experience) | 17:49 |
jonesy | also going on memory, the error comes from your local machine, not the remote one. | 17:51 |
chmouel | sounds about right but I wonder what in python unit test would need access to X11 | 17:51 |
jonesy | I think -Y was added to openssh to encourage people not to use 'xhost +' | 17:51 |
jonesy | yeah, that I don't know. | 17:51 |
jonesy | hope python isn't linked to idle or something on that box ;-P | 17:52 |
jonesy | that would be really weird. | 17:52 |
*** abecc has joined #openstack | 17:52 | |
chmouel | it's a straight install from ubuntu server | 17:53 |
*** justinsb has quit IRC | 17:54 | |
*** justinsb has joined #openstack | 17:56 | |
jaypipes | yeah \o/. friggin paths. resolved finally. thank goodness for virtualenv. | 18:00 |
* jaypipes goes off to write a blog entry on developing nova in virtualenv... | 18:01 | |
* jonesy wishes for time for developing nova with or without virtualenv | 18:01 | |
*** littleidea has quit IRC | 18:02 | |
*** christiekoehler has joined #openstack | 18:02 | |
creiht | I just got confirmation that Will's talk about swift at OSCON was filmed, I'll update everyone when it is available online | 18:02 |
jonesy | r0ck! | 18:03 |
*** littleidea has joined #openstack | 18:03 | |
*** JordanRinke has quit IRC | 18:07 | |
jonesy | how does this page not have a feed? http://www.youtube.com/view_play_list?p=12696FB0B040FA53&sort_field=added | 18:07 |
creiht | heh | 18:08 |
*** jeromatron has left #openstack | 18:09 | |
*** christiekoehler has quit IRC | 18:12 | |
jk0 | C | 18:14 |
jk0 | wrong window | 18:14 |
*** christiekoehler has joined #openstack | 18:32 | |
*** pvo has joined #openstack | 18:33 | |
*** pvo has joined #openstack | 18:33 | |
*** tobym_ has quit IRC | 18:36 | |
*** tobym has joined #openstack | 18:37 | |
*** gholms has joined #openstack | 18:37 | |
*** LoRez has joined #openstack | 18:37 | |
*** joshuamckenty has joined #openstack | 18:40 | |
gholms | Would anyone be willing to stop by #fedora-meeting in about 2h15m to help answer questions people from the Fedora cloud group may have about Openstack? | 18:42 |
*** ikk has joined #openstack | 18:43 | |
pvo | gholms: I'll see if I can scare up folks. What time are you talking? 4PM EST? | 18:44 |
joshuamckenty | ooh, that would be cool | 18:44 |
pvo | ^^ he would be one of those folks | 18:45 |
uvirtbot | pvo: Error: "^" is not a valid command. | 18:45 |
joshuamckenty | If I can track down reliable power and network, I'll come | 18:45 |
gholms | 5 PM EDT, or whatever that is in $localtime. | 18:45 |
pvo | gholms: sounds good.. I think that would be 2pm west coast... I'm time zone confused right now | 18:46 |
*** poli has quit IRC | 18:47 | |
joshuamckenty | yeah, pretty sure that's 2PM | 18:48 |
joshuamckenty | k, I have to find a venue with power, down to 10 minutes | 18:49 |
*** JordanRinke has joined #openstack | 18:49 | |
gholms | Good luck ;) | 18:49 |
*** pvo has quit IRC | 18:50 | |
creiht | gholms: Do you need represenatives for both compute and storage? | 18:52 |
*** pvo has joined #openstack | 18:53 | |
gholms | creiht: It wouldn't hurt. From the sound of things everyone on the mailing list is unfamiliar with how things work on openstack. | 18:55 |
creiht | gholms: I can't be there, but I will try to round up someone else from our group to be there | 18:56 |
gholms | The goal is basically to get official Fedora images on some of the more popular cloud platforms and/or make it easy to create one for one's platform of choice. | 18:56 |
gholms | creiht: Thanks! | 18:56 |
*** joshuamckenty has quit IRC | 18:57 | |
*** sketchup has joined #openstack | 18:58 | |
*** abecc has quit IRC | 19:00 | |
jaypipes | FYI: http://www.joinfu.com/2010/07/developing-nova-on-linux-getting-started/ | 19:03 |
*** wreese has joined #openstack | 19:04 | |
creiht | jaypipes: last image on the blog doesn't load for me (the one about passing tests) | 19:05 |
creiht | It does if I see the article from your front page though | 19:06 |
jaypipes | creiht: thx for the heads up...fixing now. | 19:06 |
jaypipes | creiht: should be fixed now. thx again :) | 19:08 |
creiht | yup | 19:09 |
*** pvo has quit IRC | 19:09 | |
justinsb | jaypipes: Awesome... what works for me is not to install anything and to use export PYTHONPATH=`pwd` instead... but your blog post made me realize I've been forgetting to do that... :-( | 19:10 |
jaypipes | justinsb: that's essentially what virtualenv does for you, plus some extra goodness... | 19:11 |
justinsb | jaypipes: Good tip, thanks! Can it remind me when I've forgotten to run it? :-) | 19:12 |
jaypipes | justinsb: hehe, no, but failed test cases will do that just fine ;P | 19:13 |
*** rdw has quit IRC | 19:13 | |
creiht | The added advantage of virtualenv is that it isolates all the packages that you are using, not just the one project that you are developing on | 19:13 |
*** christiekoehler has quit IRC | 19:13 | |
jaypipes | yup | 19:13 |
jaypipes | creiht: which is critical when you're working in numerous branches...especially when those branches have different dependencies ;) | 19:14 |
* creiht is glad swift has very few external dependencies :) | 19:14 | |
* jaypipes smacks swift with hundreds of dependencies, just for fun. | 19:15 | |
creiht | heh | 19:15 |
jaypipes | :) | 19:15 |
creiht | man... documentation can be monotonous :/ | 19:15 |
*** jonesy_ has quit IRC | 19:16 | |
* jonesy applauds jaypipes for properly spelling 'dependencies' in his blog post (and for the blog post as a whole) :) | 19:18 | |
glange | creiht: spice it up by "lieing" | 19:20 |
creiht | heh | 19:20 |
jaypipes | jonesy: hehe | 19:23 |
*** tomjones has quit IRC | 19:28 | |
*** novas0x2a has quit IRC | 19:31 | |
*** RandalSchwartz has joined #openstack | 19:34 | |
RandalSchwartz | FLOSS Weekly 128 is out... recorded live at OSCON... twit.tv/floss128 | 19:34 |
*** bd_ has left #openstack | 19:34 | |
jaypipes | RandalSchwartz: awesome :) thx! | 19:35 |
*** rajijoom has quit IRC | 19:36 | |
spackest | ok, I updated the post http://blog.spack.net/?p=162 and the scripts to include some st support and examples | 19:36 |
*** joshuamckenty has joined #openstack | 19:41 | |
*** Predominant has quit IRC | 19:43 | |
*** ewan_ has joined #openstack | 19:46 | |
*** poli has joined #openstack | 19:47 | |
*** sophiap has joined #openstack | 19:48 | |
*** christiekoehler has joined #openstack | 19:51 | |
*** jonesy_ has joined #openstack | 19:55 | |
*** theron has joined #openstack | 19:55 | |
*** jhalogen has quit IRC | 19:57 | |
theron | whoa. | 19:58 |
creiht | howdy | 19:58 |
*** tobym has quit IRC | 19:58 | |
*** ewan_ has quit IRC | 20:15 | |
*** rdw has joined #openstack | 20:17 | |
*** tobym has joined #openstack | 20:18 | |
*** pvo has joined #openstack | 20:18 | |
*** pvo has joined #openstack | 20:18 | |
*** _jcsmith has joined #openstack | 20:20 | |
notmyname | did everyone go to a meeting. I think this is the quietest it's been in here for days :-) | 20:27 |
JordanRinke | HELLO! | 20:28 |
pvo | heyooooooo | 20:28 |
creiht | shhh... I'm finally able to get some work done :) | 20:29 |
*** justinsheehy has quit IRC | 20:29 | |
* _0x44 stomps around clanging pots together. | 20:29 | |
notmyname | creiht: I thought this was your full time job now! | 20:29 |
creiht | hah | 20:29 |
*** Predominant has joined #openstack | 20:30 | |
gholms | Well, there is the Fedora Cloud SIG meeting in 30 minutes... ;) | 20:30 |
pvo | looking for dendrobates now. He was just here.... | 20:31 |
* jaypipes yells at creiht | 20:31 | |
creiht | :P | 20:32 |
jaypipes | pvo: heya, tell rick I need him to create a new milestone...can't do it since I'm not an admin...was going to create a low-hanging-fruit milestone for small or easy/medium-difficulty tasks we can toss up for interested contributors | 20:32 |
jaypipes | pvo: of course, nothing urgent, just a thought... | 20:33 |
*** miclorb has joined #openstack | 20:34 | |
notmyname | gholms: I don't know anything about fedora, but I do know (a bit) about swift. will you be talking about packaging or more about what the projects do? | 20:34 |
spackest | btw, I got the swift-1.0.2 download and it looks much the same, missing things as per https://bugs.launchpad.net/swift/+bug/608469 | 20:34 |
uvirtbot | Launchpad bug 608469 in swift ".* script is not in the 1.0.1 download" [Medium,Fix released] | 20:34 |
gholms | notmyname: Probably both. There are people looking to package openstack bits and people looking to build images for a number of cloud platforms. | 20:35 |
*** christiekoehler has quit IRC | 20:35 | |
creiht | spackest: gah! | 20:36 |
pvo | jaypipes: yea, thats right. We do need to do that. | 20:37 |
spackest | I might be missing it, but this doesn't get me anything find . -name unittests | 20:37 |
*** christiekoehler has joined #openstack | 20:37 | |
creiht | spackest: it is called .unittests | 20:38 |
creiht | and I see it in my download | 20:38 |
creiht | gotta run... but will be back later | 20:38 |
*** joshuamckenty has quit IRC | 20:39 | |
*** joshuamckenty has joined #openstack | 20:39 | |
alekibango | gholt: thansk for answer you gave me 4 hours ago | 20:40 |
*** pvo has quit IRC | 20:43 | |
RandalSchwartz | Ruh roh. the editors left the closing off the FLOSS Weekly show. :( | 20:44 |
* RandalSchwartz fires off a hasty email | 20:45 | |
*** pvo has joined #openstack | 20:45 | |
spackest | creiht: darn, my bad, yes I see them | 20:47 |
spackest | better than a nasty email | 20:47 |
*** Predominant has quit IRC | 20:57 | |
*** mcgurrin has joined #openstack | 20:58 | |
joshuamckenty | it's time? | 20:58 |
gholms | Reminder: Fedora Cloud SIG meeting in 2 minutes in #fedora-meeting | 20:58 |
*** dendro-afk is now known as dendrobates | 20:58 | |
holoway | joshuamckenty: you going to be in SF next week? | 20:58 |
pvo | yea, think so. | 20:58 |
*** Predominant has joined #openstack | 20:58 | |
joshuamckenty | holoway: no, Zurich next week | 20:58 |
mcgurrin | I have a question about the mailing lists and a few other qwuestions on OS compatibility | 20:58 |
holoway | joshuamckenty: nice! | 20:58 |
joshuamckenty | work, not holiday | 20:58 |
joshuamckenty | but still sort of nice | 20:59 |
mcgurrin | I can't seem t find the mailings lists on launchpad | 20:59 |
RandalSchwartz | I'll be in SF tomorrow | 20:59 |
notmyname | mcgurrin: I'll defer to a launchpad guru about the mailing lists, but what questions do you have about OS compatibility? | 20:59 |
mcgurrin | will at least the object storage run on a mac? | 21:00 |
mcgurrin | I think I might have a use for it but it is for a mac centric pace | 21:00 |
_jcsmith | probably, just don't do the xfs part of the install? | 21:00 |
*** dendrobates is now known as dendro-afk | 21:00 | |
notmyname | the requirements are python and a filesystem that support xattrs (HFS+ does, I think) | 21:00 |
*** dendro-afk is now known as dendrobates | 21:00 | |
mcgurrin | and can the disk used on any node be limited? | 21:01 |
mcgurrin | I am thinking of putting this on a set of workstations to provide reliable storage for large media files | 21:01 |
mcgurrin | but we need to keep disk free for other things | 21:02 |
adrian_otto | _jcsmith: Use python 2.6. I ran into trouble using an older version | 21:02 |
notmyname | mcgurrin: there is no way to limit nodes right now | 21:02 |
mcgurrin | dang, can I chose where on the filesystem it stores files, maybe a partition? | 21:03 |
*** miclorb has quit IRC | 21:03 | |
gholt | Yes, no problem there. It uses /srv/node/* by default. | 21:03 |
mcgurrin | I would also wonder if the components are availible to run the compute on a mac? | 21:04 |
mcgurrin | I work at a school where we do a lot of media editing so the main thing would be the storage but compute might be nice | 21:04 |
notmyname | mcgurrin: one solution may be to write an auth server that tracks usage and limit accordingly | 21:05 |
JordanRinke | So, there is free cotton candy at one of the Rackspace buildings right now and it is... AWESOME - just wanted to share my sugar high. Carry on. | 21:06 |
mcgurrin | I have need to limit per computer node, not per user | 21:06 |
*** joshuamckenty is now known as jmckenty|away | 21:06 | |
_jcsmith | one hack maybe would be to run more object servers on servers you want more data to be put on, since the partitioner would see them as seperate instances | 21:07 |
mcgurrin | these computers would be primarily workstations so I need to keep spae free for local users | 21:07 |
mcgurrin | that would be a neat solution | 21:07 |
mcgurrin | space* | 21:07 |
gholt | Well, if you just want a "soft" limit, the weights of the devices you add to the ring would do that. | 21:07 |
mcgurrin | thats the issue, I need hard limits, I think that is solved though, just partition the drives so that it can't use to much | 21:08 |
*** jsgotangco has joined #openstack | 21:08 | |
gholt | Another option is to use quota. Mac supports that I believe | 21:08 |
_jcsmith | maybe put a quota on the swift user? | 21:08 |
mcgurrin | I think that might be possible but a pain, I don't think there is an easy way to do quotas on anything other than the server OS | 21:09 |
holoway | creiht: do you have some sample tests for the sqlite database? | 21:09 |
holoway | creiht: sucked that my unit test didn't actually test the database part :) | 21:09 |
creiht | mcgurrin: yeah i think a separate partition would solve your problem | 21:09 |
redbo | We pull in fallocate(2) with ctypes, that would have to fail gracefully for it to work on mac. That's the only linux-ism I know of. Maybe posix_fadvise. | 21:09 |
creiht | holoway: not at the moment but i can help later | 21:10 |
holoway | creiht: no worries | 21:10 |
mcgurrin | I think partitioning is the best option though because I need to protect it from re images as well | 21:10 |
holoway | I'm going to fix it and test manually | 21:10 |
holoway | we can get around 2 it :) | 21:10 |
mcgurrin | well, thanks, I might come back later but I have to get other stuff done. | 21:12 |
*** mcgurrin has quit IRC | 21:12 | |
*** miclorb_ has joined #openstack | 21:16 | |
*** justinsheehy has joined #openstack | 21:17 | |
*** Predominant has quit IRC | 21:21 | |
*** christiekoehler has quit IRC | 21:23 | |
*** gholms has left #openstack | 21:23 | |
*** christiekoehler has joined #openstack | 21:27 | |
*** theron has quit IRC | 21:28 | |
*** spackest has quit IRC | 21:28 | |
*** justinsheehy has quit IRC | 21:33 | |
*** dendrobates is now known as dendro-afk | 21:35 | |
*** pvo has quit IRC | 21:35 | |
*** DMD has joined #openstack | 21:37 | |
*** DMD has left #openstack | 21:37 | |
*** DMD2 has joined #openstack | 21:37 | |
*** spackest has joined #openstack | 21:38 | |
*** theron has joined #openstack | 21:39 | |
*** pdelgallego has quit IRC | 21:48 | |
*** John|RDP is now known as eth00 | 21:51 | |
*** ratasxy has joined #openstack | 21:52 | |
ratasxy | hello | 21:52 |
ratasxy | i have a problem | 21:53 |
ratasxy | with swift | 21:53 |
ratasxy | ratasxy@ratasxy-desktop:~/cloudstorage/swift$ python setup.py build_sphinx | 21:53 |
ratasxy | Traceback (most recent call last): | 21:53 |
ratasxy | File "setup.py", line 17, in <module> | 21:53 |
ratasxy | from setuptools import setup, find_packages | 21:53 |
ratasxy | ImportError: No module named setuptools | 21:53 |
ratasxy | ratasxy@ratasxy-desktop:~/cloudstorage/swift$ | 21:53 |
anm_ | ratasxy: you need to instakl setuptools | 21:53 |
_jcsmith | apt-get install python-setuptools | 21:53 |
ratasxy | thamks anm_ _jcsmith | 21:54 |
*** Thweetch has quit IRC | 21:54 | |
anm_ | ratasxy: you'll probably find more missing dependencies as soon as you get past that one | 21:54 |
*** errr has joined #openstack | 21:54 | |
_jcsmith | check out the packages on here | 21:55 |
_jcsmith | http://swift.openstack.org/development_saio.html | 21:55 |
ratasxy | thanks | 21:55 |
*** theron has quit IRC | 21:57 | |
*** DMD2 has quit IRC | 22:08 | |
*** Predominant has joined #openstack | 22:10 | |
*** abecc has joined #openstack | 22:11 | |
*** miclorb_ has quit IRC | 22:11 | |
*** DMD2 has joined #openstack | 22:14 | |
*** abecc has quit IRC | 22:15 | |
*** christiekoehler has quit IRC | 22:18 | |
anm_ | if i could just get nova running, i'd breathe a big sigh and know that everythings gonna be ok ;) | 22:18 |
*** DMD2 has quit IRC | 22:18 | |
anm_ | ive cut and pasted my way through every wiki howto and hackday slice, no joy | 22:18 |
*** patcito has joined #openstack | 22:21 | |
ratasxy | hi pandemicsyn | 22:22 |
ratasxy | hi patcito | 22:23 |
patcito | hey :p | 22:23 |
anm_ | can anyone help with this error, returnrd from euca-register: TypeError: character mapping must return integer, None or unicode | 22:25 |
antonym | anm_: we ran into that during our installfest and i *think* this was the fix, http://bazaar.launchpad.net/~soren/nova/str-object-for-boto/revision/140 | 22:28 |
anm_ | i'm following teh installfest directions, so maybe so, thanks | 22:29 |
antonym | cool | 22:29 |
*** abecc has joined #openstack | 22:30 | |
*** spackest has quit IRC | 22:32 | |
*** jsgotangco has quit IRC | 22:37 | |
*** tobym has quit IRC | 22:38 | |
*** joshuamckenty has joined #openstack | 22:39 | |
*** holoway has quit IRC | 22:39 | |
*** joshuamckenty has quit IRC | 22:39 | |
*** DMD2 has joined #openstack | 22:40 | |
*** holoway has joined #openstack | 22:41 | |
*** littleidea has quit IRC | 22:42 | |
*** DMD2 has quit IRC | 22:43 | |
*** pcrews has quit IRC | 22:45 | |
*** metoikos has joined #openstack | 22:45 | |
anm_ | antonym: that was it | 22:45 |
anm_ | is that patch in the trunk? | 22:45 |
anm_ | sorry, all I know is svn terminology | 22:46 |
*** jsmith has joined #openstack | 22:47 | |
antonym | anm_: cool, yeah, i believe it's been added | 22:47 |
anm_ | ok | 22:47 |
*** abecc has quit IRC | 22:48 | |
anm_ | so, can I manage nova instances using something like ElasticFox? | 22:50 |
*** sophiap has quit IRC | 22:52 | |
*** littleidea has joined #openstack | 22:53 | |
holoway | creiht: I've got an updated version of that patch, but I'm not in a position to test it | 22:53 |
anm_ | antonym: well, i thought that was it : euca-register mybucket/vmlinuz-2.6.32-23-server.manifest.xml | 22:53 |
anm_ | TypeError: character mapping must return integer, None or unicode | 22:53 |
holoway | creiht: you want me to push it up anyway, in case one of ya'll gets a chance? | 22:53 |
anm_ | i was alomost to the bottom of the installfest howto | 22:53 |
*** justinsheehy has joined #openstack | 22:57 | |
*** littleidea has quit IRC | 23:01 | |
*** ratasxy has quit IRC | 23:02 | |
_jcsmith | whats the preferred way to distribute load for the proxies in swift? run a bunch of them and front end them with haproxy/zeus/hw load balancer, or is the auth server returning different storage urls and spreading load? | 23:07 |
*** jsgotangco has joined #openstack | 23:11 | |
anm_ | argh, if i had any hair, i would pull it out | 23:14 |
anm_ | euca-run-instances $machine --kernel $kernel --ramdisk $ramdisk | 23:14 |
anm_ | ValueError: No JSON object could be decoded | 23:14 |
anm_ | does openstack have its own pastebin? | 23:15 |
anm_ | or use etherpad maybe? | 23:15 |
anm_ | nova-api is spilling its guts when trying to run an instance | 23:16 |
*** mtaylor has joined #openstack | 23:18 | |
*** ChanServ sets mode: +v mtaylor | 23:18 | |
anm_ | heres the errors I'm getting: | 23:20 |
anm_ | http://etherpad.openstack.org/NPNjekyFvv | 23:20 |
*** metoikos has quit IRC | 23:21 | |
*** JordanRinke has quit IRC | 23:22 | |
*** miclorb has joined #openstack | 23:27 | |
*** littleidea has joined #openstack | 23:39 | |
mtaylor | crappit. what's exlt's real-life name? | 23:40 |
alekibango | Michael Shuler? | 23:41 |
alekibango | just write /whois exlt | 23:41 |
mtaylor | alekibango: hehe. oh crappit - I didn't see him up there in the voiced list | 23:42 |
mtaylor | alekibango: I was trying to be lazy and use the mouse | 23:42 |
* mtaylor smacks self | 23:42 | |
mtaylor | thanks! | 23:42 |
*** dmd17 has joined #openstack | 23:42 | |
alekibango | np | 23:42 |
*** littleidea has quit IRC | 23:43 | |
anm_ | ValueError: No JSON object could be decoded | 23:44 |
*** jsmith is now known as jsmith-dinner | 23:45 | |
*** miclorb has quit IRC | 23:45 | |
*** jsmith-dinner is now known as jsmith-away | 23:45 | |
*** anm_ is now known as anm_away | 23:45 | |
*** miclorb_ has joined #openstack | 23:45 | |
*** christiekoehler has joined #openstack | 23:48 | |
*** bretpiatt has joined #openstack | 23:51 | |
*** poli has quit IRC | 23:51 | |
*** tobym has joined #openstack | 23:51 | |
*** biosshadow has joined #openstack | 23:56 | |
*** biosshadow has left #openstack | 23:56 | |
*** notmyname has quit IRC | 00:02 | |
*** notmyname has joined #openstack | 00:02 | |
*** ChanServ sets mode: +v notmyname | 00:02 | |
*** tjyang has joined #openstack | 00:03 | |
*** rando has joined #openstack | 00:10 | |
*** mtaylor has quit IRC | 00:12 | |
*** bretpiatt has quit IRC | 00:13 | |
*** Predominant has quit IRC | 00:13 | |
*** silassewell has quit IRC | 00:14 | |
*** tonywolf has quit IRC | 00:15 | |
*** binaryWarrior has quit IRC | 00:16 | |
*** silassewell has joined #openstack | 00:16 | |
*** Predominant has joined #openstack | 00:18 | |
*** lbieber has joined #openstack | 00:18 | |
*** christiekoehler has quit IRC | 00:19 | |
*** christiekoehler has joined #openstack | 00:22 | |
*** silassewell has joined #openstack | 00:22 | |
*** rando has quit IRC | 00:23 | |
*** vish1 has joined #openstack | 00:26 | |
vish1 | what did i miss? | 00:26 |
*** gustavomzw has quit IRC | 00:30 | |
*** vish1 has quit IRC | 00:32 | |
patcito | viewlogic, the bus | 00:33 |
*** sophiap has joined #openstack | 00:34 | |
viewlogic | wha? | 00:35 |
viewlogic | oh, i see, autocomplete fail ;) | 00:36 |
*** mtaylor has joined #openstack | 00:37 | |
*** ChanServ sets mode: +v mtaylor | 00:37 | |
*** bretpiatt has joined #openstack | 00:39 | |
*** vish1 has joined #openstack | 00:43 | |
bretpiatt | vish1, NTT's block storage is out if you haven't seen it yet http://www.osrg.net | 00:47 |
*** mordred_ has joined #openstack | 00:49 | |
*** ChanServ sets mode: +v mordred_ | 00:49 | |
*** vish1 has quit IRC | 00:49 | |
*** mtaylor has quit IRC | 00:52 | |
*** dendro-afk is now known as dendrobates | 00:53 | |
*** rvalentino has joined #openstack | 00:55 | |
*** mordred_ has quit IRC | 01:01 | |
*** Predominant has quit IRC | 01:01 | |
*** christiekoehler has quit IRC | 01:02 | |
*** _jcsmith has left #openstack | 01:02 | |
*** bretpiatt has quit IRC | 01:05 | |
*** Predominant has joined #openstack | 01:06 | |
*** binaryWarrior has joined #openstack | 01:06 | |
*** jhalogen has joined #openstack | 01:08 | |
*** vish1 has joined #openstack | 01:15 | |
*** Predominant has quit IRC | 01:16 | |
*** pvo has joined #openstack | 01:19 | |
*** Predominant has joined #openstack | 01:19 | |
*** bretpiatt has joined #openstack | 01:24 | |
*** lbieber has quit IRC | 01:32 | |
*** littleidea has joined #openstack | 01:33 | |
*** sirp1 has quit IRC | 01:38 | |
*** dmd17 has quit IRC | 01:43 | |
*** Predominant has quit IRC | 01:54 | |
*** bretpiatt has quit IRC | 01:55 | |
creiht | holoway: Go ahead and push it and I will take a look | 01:58 |
*** dendrobates is now known as dendro-afk | 01:59 | |
*** jsgotangco has quit IRC | 02:03 | |
*** rvalentino has quit IRC | 02:09 | |
*** Predominant has joined #openstack | 02:18 | |
*** rvalentino has joined #openstack | 02:24 | |
*** rdw has quit IRC | 02:28 | |
alekibango | http://vmblog.com/archive/2010/07/22/opennebula-position-on-openstack-announcement.aspx <- seen this? | 02:34 |
*** jeromatron has joined #openstack | 02:36 | |
*** jdmaturen has quit IRC | 02:39 | |
*** jdmaturen has joined #openstack | 02:41 | |
uvirtbot | New bug: #608988 in nova "checksum images before booting" [Undecided,New] https://launchpad.net/bugs/608988 | 02:41 |
*** jdmaturen has quit IRC | 02:41 | |
*** jdmaturen has joined #openstack | 02:41 | |
*** randybias has joined #openstack | 02:47 | |
*** seshupasam has joined #openstack | 02:48 | |
*** seshupasam has left #openstack | 02:49 | |
*** littleidea has quit IRC | 02:52 | |
*** Predominant has quit IRC | 02:56 | |
*** jhalogen has quit IRC | 03:01 | |
*** Predominant has joined #openstack | 03:02 | |
*** vish1_ has joined #openstack | 03:03 | |
*** vish1 has quit IRC | 03:03 | |
*** vish1_ is now known as vish1 | 03:03 | |
*** jhalogen has joined #openstack | 03:03 | |
creiht | http://www.youtube.com/watch?v=kS3a0lDeJAU | 03:10 |
creiht | Is Lew's talk about openstack at OSCON | 03:10 |
*** christiekoehler has joined #openstack | 03:11 | |
PiotrSikora | creiht: thanks! | 03:12 |
creiht | Haven't seen anything about Wills talk yet | 03:12 |
PiotrSikora | only 8mins long? | 03:12 |
creiht | I think most of their keynote type talks are pretty short | 03:13 |
*** tobym has quit IRC | 03:14 | |
*** randybias is now known as randybias|away | 03:14 | |
*** Predominant has quit IRC | 03:17 | |
*** sirp1 has joined #openstack | 03:18 | |
*** jdmaturen has quit IRC | 03:20 | |
PiotrSikora | creiht: ah... i thought it was supposed to be whole presentation | 03:25 |
PiotrSikora | or maybe there will be that as well? | 03:25 |
creiht | PiotrSikora: There are more presentations, but the videos have not been posted yet | 03:26 |
PiotrSikora | ok, thx | 03:26 |
*** binaryWarrior has quit IRC | 03:29 | |
*** rdw has joined #openstack | 03:31 | |
*** jsgotangco has joined #openstack | 03:31 | |
*** tobym has joined #openstack | 03:31 | |
*** jdmaturen has joined #openstack | 03:40 | |
*** nettok has joined #openstack | 03:47 | |
*** adrian_otto has quit IRC | 03:49 | |
tobym | is OpenNebula related to NASA's Nebula project? | 03:58 |
sureshgv | hi all.. | 04:02 |
sureshgv | morning! | 04:02 |
silassewell | Could anyone explain real quick what swift-auth-recreate-accounts is used for? | 04:03 |
silassewell | Other than recreating accounts... | 04:03 |
pandemicsyn | silassewell: thats basically it | 04:04 |
pandemicsyn | if you wipe an enviornment for example | 04:04 |
pandemicsyn | it | 04:05 |
pandemicsyn | wow keyboard doesn't wanna cooperate today sorry | 04:05 |
pandemicsyn | if you wipe an enviornment for example you can just swift-auth-recreate-accounts and it will run out and recreate the auth accounts on the storage nodes | 04:05 |
silassewell | Perfect, that makes sense, thanks. | 04:05 |
pandemicsyn | comes in handy during hardware/config testing | 04:06 |
silassewell | I'm packaging this stuff for Fedora and they require at least a stub man page, trying to fill out something semi-useful without just bs'ing from the bin name | 04:07 |
pandemicsyn | awesome (we're an ubuntu/debian shop) but im still a Rhel/fedora guy deep down | 04:09 |
silassewell | haha, cool | 04:10 |
silassewell | standardizing on optparse (or whatever) and creating a man page generator from that would be pretty cool | 04:10 |
* redbo shakes his head in shame. | 04:10 | |
silassewell | seems like ever bin uses a different option parser package | 04:11 |
redbo | ha... didn't even notice. | 04:12 |
*** Predominant has joined #openstack | 04:15 | |
redbo | apparently I'm stuck on getopt and everyone else has moved on to optparse | 04:16 |
silassewell | its easier once you pick it up, plus it gives you free help and basic type checking | 04:17 |
silassewell | might be a fun little project to force myself to learn bzr | 04:17 |
silassewell | fun being defined as something I can do while drinking | 04:18 |
wreese | redbo: you suck ;) | 04:19 |
wreese | PiotrSikora: they did not record my presentation. I'm not cool enough :P | 04:20 |
redbo | I wish people would quit saying that. | 04:20 |
wreese | I will post my slides soon | 04:20 |
*** lbieber has joined #openstack | 04:20 | |
wreese | but they are probably not that interesting without some audio | 04:21 |
pandemicsyn | thats easily fixed ;) | 04:22 |
vish1 | tobym: unrelated | 04:24 |
tobym | vish1: thought so, thanks. | 04:24 |
redbo | Just wait until creiht's 70s Carl Sagan style videos about swift come out. | 04:25 |
chmouel | redbo: and in 2.7 there is again a new option parsing library http://docs.python.org/dev/library/argparse.html | 04:28 |
wreese | that's exactly what I thought of for an wasy fix :P | 04:28 |
wreese | s/wasy/easy/ | 04:28 |
*** dmd17 has joined #openstack | 04:28 | |
wreese | tobym: this you? http://tobym.posterous.com/openstack-nebula-and-eucalyptus-whats-the-dea | 04:29 |
tobym | wreese: yes...been trying to make sense of the state of things after openstack was announced | 04:30 |
tobym | wreese: any feedback? | 04:30 |
wreese | you were at the top of the list for "openstack" on a google blog search sorted by date :P | 04:30 |
tobym | wow nice...I just posted that 20 minutes ago | 04:30 |
wreese | nice post | 04:31 |
tobym | thanks | 04:32 |
*** novas0x2a has joined #openstack | 04:35 | |
*** sirp1 has quit IRC | 04:36 | |
PiotrSikora | wreese: shame on thme :/ | 04:37 |
PiotrSikora | wreese: please let me know when you'll post slides ;) | 04:37 |
wreese | np | 04:38 |
vish1 | are all of you rs guys still in portland? | 04:45 |
Predominant | I am | 04:46 |
_0x44 | vish1: pvo leaves tomorrow | 04:47 |
vish1 | i'd love to see more of the servers team participate on launchpad.... | 04:47 |
vish1 | :) | 04:47 |
*** littleidea has joined #openstack | 04:47 | |
vish1 | but i guess you guys are all busy with oscon | 04:47 |
* vish1 is trying to get his patches reviewed more quickly | 04:47 | |
_0x44 | We haven't participated more this week on LP/nova because we had a release for our current code on Wednesday | 04:48 |
*** jeromatron has quit IRC | 04:48 | |
vish1 | ah gotcha | 04:48 |
_0x44 | It was originally planned to have gone out on Monday, but someone realized that might be problematic. I wonder why? ;) | 04:49 |
vish1 | by the time you look at nova again it is going to be completely unrecognizable | 04:49 |
vish1 | (assuming i get the rest of my refactoring patches through) | 04:49 |
*** abecc has joined #openstack | 04:49 | |
_0x44 | Awesome, I won't look at it until those refactoring patches are through. ;) | 04:50 |
*** rajijoom has joined #openstack | 04:51 | |
*** Predominant has quit IRC | 04:57 | |
*** Predom[osx] has joined #openstack | 04:57 | |
*** Predom[osx] is now known as Predominant | 04:57 | |
*** lbieber has quit IRC | 05:06 | |
*** alexh__ has joined #openstack | 05:08 | |
*** alexh_ has quit IRC | 05:11 | |
*** poli has joined #openstack | 05:15 | |
*** vish1 has left #openstack | 05:21 | |
*** pdelgallego has joined #openstack | 05:22 | |
*** rajijoom has quit IRC | 05:25 | |
*** littleidea has quit IRC | 05:25 | |
*** adjohn has joined #openstack | 05:36 | |
*** miclorb_ has quit IRC | 05:40 | |
*** abecc has quit IRC | 05:40 | |
*** kashyapc has joined #openstack | 05:50 | |
*** randybias|away is now known as randybias | 05:51 | |
*** dendro-afk is now known as dendrobates | 06:01 | |
*** jdmaturen has quit IRC | 06:02 | |
*** pdelgallego has quit IRC | 06:02 | |
*** pvo has quit IRC | 06:15 | |
*** justinsheehy has quit IRC | 06:17 | |
*** justinsheehy has joined #openstack | 06:21 | |
*** nettok has quit IRC | 06:22 | |
*** Predominant has quit IRC | 06:35 | |
*** Predominant has joined #openstack | 06:35 | |
*** pdelgallego has joined #openstack | 06:44 | |
*** randybias has quit IRC | 06:46 | |
*** rvalentino has quit IRC | 06:46 | |
*** rajijoom has joined #openstack | 07:29 | |
*** rajijoom has quit IRC | 07:34 | |
*** jhalogen has joined #openstack | 07:42 | |
*** rajijoom has joined #openstack | 07:50 | |
*** stewart has joined #openstack | 08:03 | |
*** ttx has joined #openstack | 08:03 | |
*** ewan_ has joined #openstack | 08:06 | |
*** cparedes has quit IRC | 08:13 | |
*** cparedes has joined #openstack | 08:13 | |
*** cparedes has joined #openstack | 08:13 | |
*** giany has quit IRC | 08:14 | |
*** giany has joined #openstack | 08:14 | |
*** soren has quit IRC | 08:15 | |
*** soren has joined #openstack | 08:16 | |
*** ChanServ sets mode: +v soren | 08:16 | |
*** poli has quit IRC | 08:24 | |
*** poli has joined #openstack | 08:25 | |
*** sander has quit IRC | 08:29 | |
*** silassewell has quit IRC | 08:40 | |
*** patcito has quit IRC | 09:07 | |
*** mattt has quit IRC | 09:08 | |
*** mattt has joined #openstack | 09:08 | |
*** arthurc has joined #openstack | 09:10 | |
*** jhalogen has quit IRC | 09:18 | |
*** rajijoom has quit IRC | 09:20 | |
*** adjohn has quit IRC | 09:27 | |
*** dmd17 has quit IRC | 09:27 | |
*** dmd17 has joined #openstack | 09:27 | |
*** adjohn has joined #openstack | 09:27 | |
*** kashyapc has quit IRC | 09:30 | |
*** jtdowney has joined #openstack | 09:32 | |
*** kashyapc has joined #openstack | 09:37 | |
*** bahadir has joined #openstack | 09:45 | |
*** dmd17 has quit IRC | 09:48 | |
*** adjohn has quit IRC | 09:48 | |
*** rajijoom has joined #openstack | 09:50 | |
*** gustavomzw has joined #openstack | 10:15 | |
*** kashyapc has quit IRC | 10:21 | |
*** kashyapc has joined #openstack | 10:22 | |
*** jsgotangco has quit IRC | 10:28 | |
*** jonesy_ has quit IRC | 10:40 | |
*** jtdowney has quit IRC | 10:49 | |
*** ttx has quit IRC | 10:59 | |
*** dabo has joined #openstack | 11:02 | |
*** dabo has quit IRC | 11:02 | |
*** sander__ has joined #openstack | 11:13 | |
*** Guest345 is now known as joke | 11:15 | |
*** joke is now known as matteo74 | 11:15 | |
*** jonesy_ has joined #openstack | 11:24 | |
*** ewan_ has quit IRC | 11:26 | |
*** ewan_ has joined #openstack | 11:29 | |
*** rajijoom has quit IRC | 11:41 | |
*** justinsheehy has quit IRC | 11:44 | |
*** ttx has joined #openstack | 11:53 | |
*** avi has quit IRC | 11:56 | |
*** avi has joined #openstack | 11:57 | |
*** bahadir has quit IRC | 12:13 | |
*** ratasxy has joined #openstack | 12:14 | |
ratasxy | hello | 12:14 |
ratasxy | how compile swift? | 12:15 |
*** RandalSchwartz has left #openstack | 12:16 | |
jonesy | ratasxy: try this --> http://swift.openstack.org/development_saio.html | 12:17 |
ratasxy | jonesy, thanks | 12:17 |
jonesy | np | 12:17 |
*** jsmith-away is now known as jsmith | 12:30 | |
*** justinsheehy has joined #openstack | 12:33 | |
*** jdarcy has quit IRC | 12:40 | |
*** jdarcy has joined #openstack | 12:41 | |
*** alekibango has quit IRC | 12:44 | |
*** jsgotangco has joined #openstack | 12:54 | |
*** ewan_ has quit IRC | 12:59 | |
*** lbieber has joined #openstack | 13:12 | |
*** kashyapc has quit IRC | 13:13 | |
*** mcgurrin has joined #openstack | 13:18 | |
mcgurrin | hello | 13:18 |
mcgurrin | can anyone help me join your mailng list, I see on the website that they are on launchpad but I can't find them there, it may be my inexperiance with launchpad though, I normally use sourceforge | 13:20 |
mcgurrin | launchpad seems to say there isn't one wen I looked | 13:20 |
mcgurrin | when* | 13:20 |
*** ratasxy has quit IRC | 13:25 | |
_0x44 | mcgurrin: The mailing lists are linked from: https://launchpad.net/~nova and https://launchpad.net/~swift | 13:25 |
*** binaryWarrior has joined #openstack | 13:25 | |
*** ratasxy has joined #openstack | 13:25 | |
jsgotangco | good morning! | 13:25 |
mcgurrin | thats what I thought but: | 13:26 |
notmyname | howdy! | 13:26 |
mcgurrin | This team does not use Launchpad to host a mailing list. | 13:26 |
mcgurrin | comes up under mailing list | 13:26 |
mcgurrin | hello | 13:26 |
_0x44 | mcgurrin: OH, right... you have to be a member of those teams to be able to subscribe to those lists. | 13:26 |
_0x44 | mcgurrin: I think this one should be public: https://launchpad.net/~openstack-discuss | 13:27 |
mcgurrin | There’s no page with this address in Launchpad. | 13:27 |
mcgurrin | I see the member thing for nova | 13:27 |
_0x44 | Also, it doesn't look like swift has a mailing list yet. | 13:27 |
*** lbieber has quit IRC | 13:27 | |
mcgurrin | thats what it looked like | 13:28 |
_0x44 | Sorry, I thought they had. | 13:28 |
mcgurrin | I just wanted to make sure I was looking where I should | 13:28 |
mcgurrin | it's fine | 13:28 |
mcgurrin | any other ideas on the public list? | 13:29 |
mcgurrin | I could use to have one for my questions and I might be able to help out a little as well, I'm more of an email guy than IRC usually | 13:29 |
mcgurrin | thanks for what you have been able to do | 13:31 |
mcgurrin | It helps to know I am not missing something, the website/launchpad differences had me thinking I might have | 13:31 |
_0x44 | You're welcome :) | 13:31 |
_0x44 | I think monty or will will have to create the swift list, but they're both away one on vacation the other at oscon | 13:32 |
mcgurrin | cool | 13:32 |
notmyname | I think it might be creiht rather than wreese | 13:33 |
_0x44 | That makes sense. | 13:33 |
notmyname | (but I'm just saying that to make sure they see their names highlighted when they get on) | 13:33 |
_0x44 | I didn't look for creiht or wreese because I saw mtaylor wasn't on... | 13:34 |
anm_away | can anyone give me some assistance? Im at the end of the installfest instructions for getting nova running, and when I finally try to launce an instance using euca-run0instances, i get: | 13:38 |
anm_away | ValueError: No JSON object could be decoded | 13:38 |
anm_away | everything was fine up until trying to launce the instance | 13:38 |
anm_away | s/launce/launch/ | 13:38 |
*** anm_away is now known as anm_ | 13:39 | |
*** ratasxy has quit IRC | 13:39 | |
anm_ | so, i've built the machine, but the gas tank is empty | 13:41 |
anm_ | this ride takes 3 tickets, and I only have two | 13:41 |
anm_ | etc. etc. | 13:41 |
*** ewan_ has joined #openstack | 13:42 | |
*** jdarcy has quit IRC | 13:52 | |
*** silassewell has joined #openstack | 13:53 | |
*** allsystemsarego has joined #openstack | 14:01 | |
exlt | mtay<TAB> isn't here.. but, got your branch review request - soon as I can, I'll look ;) | 14:02 |
mcgurrin | do you guys know if any other software will support openstack like cyberduck, (I am really thinking about software that supports cloud files currently. | 14:05 |
notmyname | mcgurrin: swift is API compatible with Cloud Files | 14:06 |
notmyname | so any program that supports Cloud FIles (like cyberduck) will support swift | 14:07 |
notmyname | there is a trick, though | 14:07 |
mcgurrin | I figured that but if any URL's are hardcoded into something like cyberduck it won't work | 14:07 |
notmyname | most of those programs have the rackpsace auth server hardcoded | 14:07 |
notmyname | right | 14:07 |
mcgurrin | That's the sort of thing Iwas figuring | 14:07 |
mcgurrin | you mean faking | 14:07 |
mcgurrin | the DNS or hosts file? | 14:07 |
notmyname | there is only one app I know of that will work out of the box. "cloud mobile" on the iphone allows you to set the auth server | 14:08 |
notmyname | other tools (like wordpress plugins) could be made to work very easily by editing the PHP code | 14:08 |
*** mattf has joined #openstack | 14:09 | |
notmyname | you might be able to get away with some faking via dns or hosts | 14:09 |
notmyname | the auth server included in swift is compatible with the rackspace auth server, so it should work | 14:09 |
mcgurrin | I think I could even do IP based rewriting through our firewall | 14:10 |
notmyname | you will need to rewrite auth.api.rackspacecloud.com and auth.mosso.com (I think that 2nd one is right. it's an older url that some tools still have hard coded) | 14:11 |
exlt | duplicity will also honor the env var CLOUDFILES_AUTHURL=<https://yourauthsrvr> | 14:11 |
mcgurrin | that would work, do you know of any backup programs that use cloud files and support keeping old versions of files? | 14:12 |
mcgurrin | his might be a cool backup solution for servers to extra workstation space | 14:12 |
exlt | I use duplicity monthly full backups with nightly incrementals and 90-day retention - that gives me 3 months of rolling back, if I want | 14:13 |
notmyname | mcgurrin: that second one should be api.mosso.com | 14:13 |
mcgurrin | does duplicity use cloud files I assume then? | 14:14 |
*** poli has quit IRC | 14:14 | |
exlt | mcgurrin: yep | 14:14 |
mcgurrin | thanks notmyname | 14:14 |
mcgurrin | we were looking at a fairly expensive remote backup solution, this might be a cool one to try for free/cheap | 14:14 |
exlt | duplicity uses a wide range of back end services, ftp, scp, local/remote fs mounts, s3, CF.. | 14:14 |
mcgurrin | cool, we might even be able to use it for both then | 14:15 |
*** tonywolf has joined #openstack | 14:15 | |
mcgurrin | thanks you guys, I'm slowly hashing out uses with your help | 14:16 |
notmyname | any time. i just think it's awesome that someone wants to use some code I helped write :-) | 14:16 |
exlt | mcgurrin: I use it for dual backups to cloud files and s3 on important data - http://github.com/mshuler/duplication | 14:16 |
exlt | that script need some love - it's just the backup run - it needs listing, verify, etc.. | 14:17 |
mcgurrin | we already backup to a second hard drive in case of system failure, we are already talking second and maybe third full backup ere | 14:17 |
mcgurrin | here* | 14:17 |
notmyname | if you had a small swift cluster, you would have 3 copies of the data protected against many system and drive failures (dependent a little on how you deployed the hardware) | 14:18 |
mcgurrin | we are looking at a relatively cheap dedi with extremely cheap SAN storage to backup to and now this | 14:18 |
mcgurrin | $1/month per 100GB of storage | 14:19 |
*** mtaylor has joined #openstack | 14:27 | |
*** ChanServ sets mode: +v mtaylor | 14:27 | |
*** pvo has joined #openstack | 14:30 | |
*** johnpur has joined #openstack | 14:31 | |
*** ChanServ sets mode: +v johnpur | 14:31 | |
*** pshouse has joined #openstack | 14:43 | |
*** pshouse has left #openstack | 14:45 | |
*** jsmith is now known as jsmith-away | 14:45 | |
*** patrickshouse has joined #openstack | 14:45 | |
*** ratasxy has joined #openstack | 14:48 | |
*** jsgotangco has quit IRC | 14:48 | |
*** sirp1 has joined #openstack | 14:51 | |
jonesy | so, in launchpad, if I do a merge proposal, and then I get feedback, and make some changes, do I then have to resubmit the proposal? | 14:54 |
jonesy | I committed my changes to my branch, and changed the status back to Needs Review, but I don't see a change on the lp page at all. | 14:55 |
jonesy | ....which is kind of unintuitive. | 14:55 |
jonesy | damn. Nevermind. It's covered on the wiki. Again. :-P | 14:56 |
*** littleidea has joined #openstack | 14:58 | |
anm_ | can I use EC2 tools with to manage instances in nova? euca tools are not working for me | 15:10 |
*** christiekoehler has quit IRC | 15:12 | |
dendrobates | anm_: the ec2 tools license says they can only be used with ec2 | 15:12 |
anm_ | ok, well crap | 15:13 |
*** lbieber has joined #openstack | 15:13 | |
anm_ | euca-tools --debug provides no more debug than without, so im stuck | 15:13 |
creiht | jonesy: cool... I'll review it shortly | 15:14 |
jonesy | creiht: thx! | 15:14 |
*** adrian_otto has joined #openstack | 15:15 | |
anm_ | is there a way to put more debug output from the euca tools at runtime? | 15:16 |
*** adrian_otto1 has joined #openstack | 15:17 | |
*** adrian_otto has quit IRC | 15:19 | |
anm_ | this is the error from nova-api: http://etherpad.openstack.org/NPNjekyFvv | 15:20 |
anm_ | so i think the error is in nova, not euca-tools the more I look | 15:21 |
anm_ | should I report this as a bug in launchpad? | 15:21 |
*** poli has joined #openstack | 15:21 | |
anm_ | and do I file it against openstack or nova in launchpad? nova doesnt appear to have any bugs in launchpad | 15:23 |
*** stewart has quit IRC | 15:23 | |
*** lbieber has quit IRC | 15:23 | |
mtaylor | anm_: then you'll be the first! | 15:25 |
mtaylor | anm_: (file against nova) | 15:25 |
creiht | mtaylor: evidently swift doesn't have a mailing list | 15:26 |
creiht | and good morning :) | 15:26 |
anm_ | DONE! https://bugs.launchpad.net/nova/+bug/609203 | 15:26 |
uvirtbot | Launchpad bug 609203 in nova "Error trying to start instance" [Undecided,New] | 15:26 |
anm_ | my first nova bug :D | 15:26 |
*** justinsheehy has quit IRC | 15:27 | |
mtaylor | creiht: yes. working on it :( | 15:27 |
mtaylor | creiht: and good morning to you! | 15:27 |
*** naren has joined #openstack | 15:29 | |
creiht | mtaylor: hope you had a good trip, and that we didn't ruin it too much | 15:31 |
mtaylor | creiht: trip was great - and really what's better than hacking on stuff while at the beach :) | 15:32 |
*** ttx has quit IRC | 15:33 | |
creiht | jonesy: btw... have you signed the CLA? | 15:36 |
jonesy | er. | 15:36 |
uvirtbot | New bug: #609203 in nova "Error trying to start instance" [Undecided,New] https://launchpad.net/bugs/609203 | 15:36 |
jonesy | I'm gonna say no, creiht | 15:36 |
jonesy | since I don't know what/where it is. | 15:36 |
creiht | it isn't a big deal | 15:37 |
creiht | http://wiki.openstack.org/HowToContribute | 15:37 |
jonesy | creiht: loading now. :) | 15:38 |
*** jeromatron has joined #openstack | 15:40 | |
jonesy | that is a pretty cool e-sign tool. | 15:40 |
jonesy | creiht: ok, I'm all signed and verified. | 15:41 |
creiht | jonesy: awesome | 15:41 |
creiht | jonesy: one more small change needed... sorry | 15:41 |
jonesy | np! | 15:42 |
silassewell | so swift-init, do we think its going to stay pretty much the same as it is now for the near future? | 15:45 |
creiht | silassewell: I don't expect it to change | 15:46 |
*** dendrobates is now known as dendro-afk | 15:46 | |
creiht | silassewell: Are you asking in the context of using it for init scripts? | 15:46 |
*** JordanRinke has joined #openstack | 15:47 | |
*** pvo has quit IRC | 15:47 | |
silassewell | creiht: yeah, just trying to figure out the best way to integrate it with the fedora/el init stuff | 15:48 |
creiht | silassewell: you might ask exlt as he has been doing the same with the debian packaging | 15:49 |
silassewell | creiht: yeah, I've looked at the debian stuff which basically just calls it directly, fedora likes some extra information. trying to decide if it would be easier to rewrite it in bash or just patch what's there | 15:50 |
exlt | silassewell: it's on my honey-do list as soon as I can get to it - proper debian inits | 15:51 |
*** Predominant has quit IRC | 15:52 | |
*** jeromatron has left #openstack | 15:53 | |
silassewell | exlt: do you think you're going to update swift-init or just write them from scratch? looking over swift-init there looks like there are some cases where things might fail and there is no way for an init script to know | 15:55 |
*** jsgotangco has joined #openstack | 15:55 | |
*** ratasxy has quit IRC | 15:56 | |
jonesy | tornado released 1.0 last night, | 15:57 |
exlt | silassewell: mtaylor and I talked about this yesterday - I think leaving swift-init there for source install use is great - the debian inits will use the standard start-stop-daemon with PID files, etc in the same locations | 15:58 |
silassewell | exlt: ok, I'll just plan on doing that as well | 15:59 |
exlt | time is my only factor, atm - not a difficult branch, just tedious | 15:59 |
silassewell | exlt: definitely | 15:59 |
mtaylor | exlt: you saw the packaging split went through, yeah? | 16:01 |
*** sketchup has left #openstack | 16:04 | |
jonesy | creiht: tried again. Someone else chimed in too :) | 16:06 |
allsystemsarego | what Linux distributions is openstack already packaged for? | 16:06 |
chmouel | i think fedora is working on it | 16:07 |
creiht | allsystemsarego: we have some basic ubuntu packaging, but it needs some work | 16:07 |
silassewell | allsystemsarego: https://bugzilla.redhat.com/show_bug.cgi?id=617632 -- its still rough through | 16:08 |
uvirtbot | bugzilla.redhat.com bug 617632 in Package Review "Review Request: swift - OpenStack Object Storage (swift)" [Medium,New] | 16:08 |
rbergeron | thanks, silassewell :) | 16:09 |
notmyname | jonesy: I was the "someone else" | 16:12 |
*** Predominant has joined #openstack | 16:12 | |
*** tobym has quit IRC | 16:14 | |
silassewell | exlt: are you creating a user for each service or just one for all? don't really know enough about how the services interact yet | 16:15 |
creiht | silassewell: We run it as just one user for everything | 16:17 |
silassewell | creiht: ok | 16:17 |
*** randybias has joined #openstack | 16:20 | |
exlt | silassewell: that user/group is set up in debian/swift.postinst | 16:22 |
*** mark___ has joined #openstack | 16:23 | |
silassewell | exlt: thanks, must have missed that (still figuring out the debian packaging stuff) | 16:23 |
*** mark___ is now known as sparkycollier | 16:24 | |
* JordanRinke waves to sparkycollier | 16:24 | |
sparkycollier | Good day JordanRinke + world | 16:26 |
*** jsgotangco has quit IRC | 16:28 | |
*** bretpiatt has joined #openstack | 16:29 | |
*** sparkycollier has quit IRC | 16:29 | |
*** christiekoehler has joined #openstack | 16:31 | |
*** Predominant has quit IRC | 16:32 | |
*** jdmaturen has joined #openstack | 16:34 | |
*** justinsheehy_ has joined #openstack | 16:34 | |
*** jdmaturen has joined #openstack | 16:34 | |
*** ewan_ has quit IRC | 16:35 | |
*** randybias has quit IRC | 16:36 | |
*** bretpiat1 has joined #openstack | 16:36 | |
*** bretpiatt has quit IRC | 16:36 | |
*** Predominant has joined #openstack | 16:37 | |
*** tobym has joined #openstack | 16:37 | |
*** lbieber has joined #openstack | 16:38 | |
*** justinsheehy_ is now known as justinsheehy | 16:39 | |
*** mjmac has joined #openstack | 16:43 | |
*** tupshin has quit IRC | 16:46 | |
*** poli has quit IRC | 16:49 | |
*** bretpiat1 has quit IRC | 16:50 | |
*** lbieber has quit IRC | 16:50 | |
*** lbieber has joined #openstack | 16:52 | |
*** blp has joined #openstack | 16:53 | |
*** blp is now known as bretpiatt | 16:53 | |
*** jbd has quit IRC | 16:54 | |
*** justinsheehy has quit IRC | 16:57 | |
*** arthurc has quit IRC | 16:58 | |
jonesy | notmyname: well thanks for the input! | 16:58 |
*** pdelgallego has quit IRC | 17:01 | |
*** poli has joined #openstack | 17:01 | |
*** jsmith-away is now known as jsmith | 17:03 | |
*** silassewell has quit IRC | 17:03 | |
lbieber | anm_ file bugs against nova - https://bugs.launchpad.net/nova | 17:05 |
PiotrSikora | wreese: did you have time to upload your slides yet? | 17:06 |
* jonesy 's still waiting for the vid to be posted. | 17:12 | |
*** lbieber has quit IRC | 17:12 | |
*** silassewell has joined #openstack | 17:12 | |
*** bretpiatt has quit IRC | 17:12 | |
*** justinsheehy has joined #openstack | 17:14 | |
*** lbieber has joined #openstack | 17:15 | |
*** bretpiatt has joined #openstack | 17:17 | |
* jaypipes wishes for autumn...this summer heat is killing him... | 17:19 | |
jonesy | jaypipes: are you in TX? | 17:20 |
jonesy | they don't have autumn there. They just have summer and February. | 17:20 |
jaypipes | jonesy: hehe, no, in Columbus Ohio. I thought it was hot in Austin last week..got back, it's the same here ;) | 17:20 |
jonesy | heh | 17:20 |
bretpiatt | jaypipes, yeah I was looking forward to a cool weekend in NYC for DrupalCamp, it's going to be as hot there as it is in San Antonio :( | 17:21 |
jonesy | ...but more humid :( | 17:21 |
jaypipes | bretpiatt, jonesy: more humid than Austin?! | 17:22 |
jonesy | no, than SA | 17:22 |
jaypipes | ah | 17:23 |
bretpiatt | more excitement about swift today, the folks from nasuni pulled down code and got it running in their lab in an hour, they're happy to finally have a scalable backend for testing in their own lab | 17:25 |
*** lbieber has quit IRC | 17:26 | |
jaypipes | bretpiatt: we have to work on that "in an hour" thing ;) 5 minute installs FTW. | 17:26 |
* jaypipes happy to see termie's Makefile in nova...made my blog post irrelevant in, like, 1 day ;) | 17:27 | |
*** JordanRinke has quit IRC | 17:27 | |
*** justinsheehy has quit IRC | 17:27 | |
*** _0x44 is now known as rootkit | 17:27 | |
jaypipes | bretpiatt: It would be super-excellent to get someone to code up a custom installer for openstack as a whole, that would download and install each/all components of the stack and lead someone through the various setup stuffs... | 17:29 |
*** rootkit is now known as _0x44 | 17:30 | |
jonesy | jaypipes: what? I was planning to use that blog post, like, next week. I also don't see a Makefile. Where's that? | 17:30 |
bretpiatt | jaypipes, jordanrinke put some additional docs in the wiki, we're talking to the fedora team about packages there, soren or ? is working on ubuntu packages, we have a few debian packages guys involved too (not sure who specifically is on it there) | 17:31 |
jaypipes | jonesy: termie put in a merge prop on it a little while ago...not in trunk yet. | 17:31 |
*** talli has joined #openstack | 17:31 | |
jonesy | oh | 17:31 |
jaypipes | jonesy: you can still follow my blog post, though. termie's work simply lays out the dependencies properly. | 17:31 |
jaypipes | lemme grab a link for ya... | 17:31 |
bretpiatt | jaypipes, we will need long term an installer app though that lets people pick which components to install to ensure they get a fully functioning system | 17:32 |
jaypipes | jonesy: https://code.launchpad.net/~termie/nova/testing_dependencies/+merge/30738 | 17:32 |
jaypipes | bretpiatt: yup. | 17:32 |
jonesy | heh - I was already there :) | 17:32 |
jaypipes | bretpiatt: especially with so much necessary configuration/setup work and so many components. | 17:32 |
bretpiatt | jaypipes, greenisus and adrian cole may be able to sort that out together after we have packages, they're going to integrate chef into the mobile and web apps, part of those apps could have a screen for "install new deployment" | 17:33 |
jaypipes | nice :) | 17:34 |
*** Predominant has quit IRC | 17:34 | |
*** lbieber has joined #openstack | 17:35 | |
*** hugod has quit IRC | 17:36 | |
*** tobym has quit IRC | 17:37 | |
*** _jcsmith has joined #openstack | 17:37 | |
*** dpa has joined #openstack | 17:39 | |
*** lbieber has quit IRC | 17:44 | |
*** novas0x2a has quit IRC | 17:45 | |
*** Predominant has joined #openstack | 17:47 | |
*** dpa has quit IRC | 17:49 | |
*** christiekoehler has quit IRC | 17:52 | |
*** KnuckleSangwich has joined #openstack | 17:55 | |
creiht | jonesy: sadly, they may not be posting video... I will let you know if that changes though :/ | 17:55 |
jonesy | oh man :( | 17:55 |
jonesy | nice "open source" conference :-/ | 17:56 |
creiht | jaypipes: Re: install, there are already chef scripts and bash scripts in the wild | 17:56 |
creiht | for the SAIO at least | 17:58 |
*** christiekoehler has joined #openstack | 17:58 | |
*** bretpiatt has quit IRC | 17:58 | |
*** dpa has joined #openstack | 18:00 | |
*** Predominant has quit IRC | 18:00 | |
*** tobym has joined #openstack | 18:00 | |
jaypipes | creiht: sure, understood, I was talking more about an OSAIO, though :) | 18:01 |
creiht | for sure... I'm working on a deployment document right now that should help someone start on an effort like that | 18:04 |
*** bretpiatt has joined #openstack | 18:07 | |
*** justinsheehy has joined #openstack | 18:09 | |
*** poli has quit IRC | 18:11 | |
*** poli has joined #openstack | 18:11 | |
*** Predominant has joined #openstack | 18:14 | |
*** justinsheehy has quit IRC | 18:16 | |
*** pvo has joined #openstack | 18:18 | |
*** pvo has joined #openstack | 18:18 | |
*** naren has left #openstack | 18:21 | |
anm_ | jaypipes: is setuptools robust enough to do a setup with? | 18:23 |
anm_ | ive never actually tried to build a setup using it | 18:23 |
anm_ | or maybe easy_install? | 18:23 |
*** justinsheehy has joined #openstack | 18:23 | |
creiht | anm_: easy_install is powered by setuptools | 18:23 |
anm_ | i see | 18:23 |
creiht | setuptools will get python packages installed, and even dependencies if configured right | 18:24 |
creiht | most ops people don't like managing python packages with setup.py though and prefer os packaging in stead | 18:25 |
creiht | So for deployment, there are two steps. 1. is getting packages installed (and dependencies) 2. Configuration of the cluster | 18:25 |
creiht | 1. is almost there with OS packaging | 18:25 |
creiht | 2. can be done with some automation | 18:26 |
creiht | at this point, I'm a little leary of automation, as there are a lot of decisions to make when setting up a production cluster | 18:26 |
justinsb | anm_: I commented on your bug... I think it's an auth issue between the objectstore and the compute node if you're running trunk | 18:27 |
creiht | I don't want people to think they can just throw some servers together with some drives, and then install the software and think they are done | 18:27 |
anm_ | justinsb: is that something I have control of? I just do . novarc to get all the keys loaded into env vars | 18:29 |
mcgurrin | I might be able to help with any efforts to make a mac installer | 18:30 |
*** stewart has joined #openstack | 18:30 | |
*** KnuckleSangwich has quit IRC | 18:30 | |
mcgurrin | I think that would not be terribly difficult using the mac package installing system and a post install script for setup | 18:31 |
jaypipes | creiht: that's why I suggested an installer that walked the user through those steps.. | 18:31 |
mtaylor | or debconf questions | 18:31 |
*** JordanRinke has joined #openstack | 18:33 | |
mcgurrin | what is the minimum cluster size btw? | 18:33 |
creiht | mcgurrin: for swift? | 18:33 |
mcgurrin | yeah | 18:33 |
termie | my makefile stuff, btw, is specifically for testing for development | 18:33 |
termie | dinner time though, bbiab | 18:34 |
creiht | well you can run everything on 1 server, but that wouldn't be very failure resistant :) | 18:34 |
mcgurrin | ok, it will accept that though like for testing? | 18:34 |
creiht | mcgurrin: yeah that is how we do development | 18:34 |
anm_ | justinsb: saw the notes, I'm running from soren's PPA debs. | 18:34 |
mcgurrin | ok, thats good, I thought dev might be through a few VM's or something | 18:35 |
creiht | If you want to do some performance testing though, I recommend a more realistic cluster of at least 4-5 nodes | 18:35 |
*** KnuckleSangwich has joined #openstack | 18:35 | |
creiht | mcgurrin: yeah we do dev on one vm with all the services running | 18:35 |
*** KnuckleSangwich has quit IRC | 18:35 | |
mcgurrin | I'm looking for quick usability testing and making sure it will work for my purposes, not for speed | 18:35 |
creiht | then we have a small dev cluster that we can push to inorder to do more realistic testing | 18:35 |
anm_ | i may go back to pulling source to stay on trunk, but the docs seem more sketchy on running from trunk | 18:36 |
mcgurrin | can anyone guide me through a setup if I pull the code? | 18:36 |
*** mcgurrin has quit IRC | 18:36 | |
*** mcgurrin has joined #openstack | 18:36 | |
creiht | mcgurrin: http://swift.openstack.org/development_saio.html | 18:36 |
*** bretpiatt has quit IRC | 18:36 | |
creiht | is the instructions to set up a dev vm | 18:37 |
creiht | and several swift devs are on this channel if you have any questions | 18:37 |
mcgurrin | cool, thanks, I will take a look into this in a few minutes, I just need to see about freeing up disk space | 18:38 |
*** hugod has joined #openstack | 18:38 | |
*** stewart has quit IRC | 18:38 | |
*** patrickshouse has quit IRC | 18:41 | |
*** aliguori has quit IRC | 18:41 | |
*** xtoddx has joined #openstack | 18:44 | |
justinsb | anm_: I wrote those docs, so thanks! :-) | 18:45 |
justinsb | anm_: I never got it to work from the PPAs. Others may have... | 18:45 |
anm_ | ok, I'll try building from source | 18:46 |
anm_ | I'm doing that now using the installfest doc | 18:46 |
justinsb | anm_: I think trunk, with use_s3=False, should work | 18:46 |
anm_ | ok | 18:46 |
justinsb | anm_: Certainly people will be more able to help you if you're closer to trunk! | 18:46 |
anm_ | true | 18:47 |
*** mtaylor has quit IRC | 18:47 | |
*** christiekoehler has quit IRC | 18:48 | |
justinsb | anm_: Use these instructions: http://wiki.openstack.org/NovaInstallFestInstructions, not the etherpad. Both are mine, but I updated the wiki and not the etherpad. | 18:49 |
anm_ | justinsb: Yeah, I m using the wiki - need to fix a few things on it, is that ok? | 18:49 |
anm_ | like: | 18:50 |
anm_ | bzr init-repo . | 18:50 |
anm_ | cd nova | 18:50 |
anm_ | bzr branch lp:nova | 18:50 |
anm_ | nova doesnt exist yet | 18:50 |
*** pvo has left #openstack | 18:50 | |
justinsb | anm_: Definitely... be my guest! | 18:50 |
*** poli has quit IRC | 18:56 | |
*** aliguori has joined #openstack | 18:57 | |
*** tobym has quit IRC | 19:13 | |
*** tobym has joined #openstack | 19:15 | |
*** _jcsmith has quit IRC | 19:22 | |
*** dpa has quit IRC | 19:25 | |
*** hugod has quit IRC | 19:26 | |
*** christiekoehler has joined #openstack | 19:28 | |
*** joshuamckenty has joined #openstack | 19:28 | |
*** hugod has joined #openstack | 19:33 | |
*** JordanRinke has quit IRC | 19:34 | |
*** Predominant has quit IRC | 19:36 | |
*** mtaylor has joined #openstack | 19:41 | |
*** ChanServ sets mode: +v mtaylor | 19:41 | |
*** jsmith is now known as jsmith-away | 19:41 | |
*** JordanRinke has joined #openstack | 19:42 | |
*** Predominant has joined #openstack | 19:43 | |
soren | mtaylor: Just the man I wanted to talk to! | 19:48 |
*** jhalogen has joined #openstack | 19:48 | |
mtaylor | soren: yay! | 19:48 |
mtaylor | soren: well here I am! | 19:48 |
soren | mtaylor: I forget if you answered this already, but are you intending to set up automatic ppa uploads for nova? | 19:48 |
mtaylor | soren: yes. it's done already for swift ... I hadn't gotten to it for nova yet | 19:49 |
soren | mtaylor: Excellent. | 19:49 |
jaypipes | http://arstechnica.com/open-source/news/2010/07/nasa-and-rackspace-part-the-clouds-with-open-source-project.ars | 19:49 |
mtaylor | soren: actually, I've got launchpad doing nightlies for swift with the new daily-ppa-build stuff they added | 19:49 |
jaypipes | mtaylor: ping, any reason my recent blog post didn't show up on openstack planet? | 19:50 |
mtaylor | jaypipes: because you suck? | 19:50 |
jaypipes | mtaylor: well, there's that... | 19:50 |
mtaylor | :) | 19:50 |
mtaylor | looking | 19:50 |
mtaylor | jaypipes: btw - it's being run by a hudson job... | 19:50 |
mtaylor | jaypipes: http://hudson.openstack.org/job/Planet%20OpenStack/1626/console | 19:51 |
soren | jaypipes: The feed url you provided is not an rss feed, afaict. | 19:51 |
mtaylor | jaypipes: I agree with soren | 19:51 |
soren | jaypipes: Add "/feed/" to the end, and you should be golden. | 19:51 |
mtaylor | jaypipes: you wanted http://www.joinfu.com/category/openstack/feed/ | 19:51 |
mtaylor | jinx | 19:51 |
soren | snap | 19:51 |
mtaylor | soren: those daily ppas they have would be _great_ if you could configure it to occur on branch push ... | 19:52 |
soren | mtaylor: Yes. Yes, they would. | 19:52 |
soren | mtaylor: I think having a package in the PPA 20 minutes after a bzr push is *sexy*. | 19:53 |
mtaylor | soren: agree | 19:53 |
mtaylor | soren: until the get branch-push, we can have hudson upload them for us :) | 19:53 |
mtaylor | right now, /me is filing bugs against tarmac... | 19:53 |
mtaylor | rockstar is going to hate me | 19:53 |
jaypipes | mtaylor: hmm...ok, I'll change it. thx. | 19:55 |
*** abecc has joined #openstack | 19:55 | |
*** poli has joined #openstack | 19:58 | |
*** pdelgallego has joined #openstack | 19:58 | |
*** abecc has quit IRC | 19:59 | |
*** randybias has joined #openstack | 19:59 | |
*** phildoc is now known as philaway | 20:05 | |
*** randybias has quit IRC | 20:13 | |
*** hugod has quit IRC | 20:16 | |
mtaylor | jaypipes: woot. you're on | 20:16 |
mtaylor | jaypipes: man - you need to remove that white box from your hackergotchi | 20:16 |
mtaylor | jaypipes: love the pic though | 20:16 |
*** hugod has joined #openstack | 20:17 | |
jaypipes | hmm....white box eh? | 20:18 |
mtaylor | jaypipes: yeah - replace with transparent and you'll be gold | 20:19 |
* soren sees it too | 20:19 | |
jaypipes | mtaylor, soren: ya, I see it...hmm. | 20:19 |
jaypipes | Bring out the Gimp. | 20:19 |
mtaylor | nice | 20:20 |
*** jonesy_ has quit IRC | 20:27 | |
*** Predominant has quit IRC | 20:28 | |
*** adrian_otto1 has quit IRC | 20:29 | |
*** poli has quit IRC | 20:29 | |
*** christiekoehler has quit IRC | 20:30 | |
mtaylor | soren: https://bugs.edge.launchpad.net/launchpad/+bug/609298 | 20:40 |
uvirtbot | Launchpad bug 609298 in launchpad "daily PPA system should be able to be triggered by branch push" [Undecided,New] | 20:40 |
*** xfrogman5 has joined #openstack | 20:44 | |
soren | mtaylor: Make that two cookies. | 20:44 |
mtaylor | soren: :) | 20:45 |
*** gasbakid has joined #openstack | 20:47 | |
rackerhacker | happy friday afternoon, folks http://wewt.me/8c | 20:53 |
notmyname | it's techno viking! | 20:53 |
rackerhacker | you betcha | 20:54 |
glange | it's sara palin! | 20:54 |
notmyname | every day must be friday for him | 20:54 |
*** hugod has joined #openstack | 20:58 | |
*** jhalogen has quit IRC | 21:03 | |
anm_ | http://www.servercorps.com/images/stories/haters-gonna-hate.gif | 21:07 |
glange | http://chzsomuchpun.files.wordpress.com/2010/04/129165187334369262.jpg | 21:09 |
*** randybias has joined #openstack | 21:15 | |
*** jsmith-away is now known as jsmith | 21:17 | |
*** jtdowney has joined #openstack | 21:17 | |
*** lbieber has joined #openstack | 21:21 | |
*** joshuamckenty has quit IRC | 21:21 | |
*** joshuamckenty has joined #openstack | 21:22 | |
*** xfrogman5 has quit IRC | 21:24 | |
*** tobym has quit IRC | 21:27 | |
*** mcgurrin has quit IRC | 21:31 | |
*** randybias has quit IRC | 21:31 | |
uvirtbot | New bug: #609312 in nova "Comments on raw-disk-image branch" [Undecided,New] https://launchpad.net/bugs/609312 | 21:36 |
*** joshuamckenty has quit IRC | 21:36 | |
*** joshuamckenty has joined #openstack | 21:37 | |
*** mcgurrin has joined #openstack | 21:48 | |
*** allsystemsarego has quit IRC | 22:01 | |
*** algorist has joined #openstack | 22:01 | |
*** johnpur has quit IRC | 22:03 | |
*** pdelgallego has quit IRC | 22:04 | |
*** jsmith is now known as jsmith-away | 22:06 | |
*** binaryWarrior has quit IRC | 22:06 | |
*** mtaylor has quit IRC | 22:14 | |
*** joshuamckenty has quit IRC | 22:28 | |
creiht | And a parting gift for the weekend | 22:34 |
creiht | http://swift.openstack.org/deployment_guide.html | 22:34 |
creiht | It is just a start, but should be enough for those who are interested in deploying real swift cluster to start thinking | 22:35 |
creiht | and looks like wreese's presentation can be downloaded here: http://www.oscon.com/oscon2010/public/schedule/detail/15634 | 22:37 |
redbo | the slides? | 22:38 |
creiht | yes | 22:38 |
creiht | sorry... sounds like there will not be video after all :/ | 22:38 |
exlt | I love that comment: # disable TIME_WAIT.. wait.. | 22:38 |
creiht | :) | 22:38 |
exlt | glad it made it in there | 22:38 |
exlt | creiht: that looks really good so far | 22:39 |
creiht | exlt: thanks! If you think of other sections that need to be there, let me know | 22:40 |
redbo | or submit a patch :P | 22:40 |
*** _jcsmith has joined #openstack | 22:40 | |
creiht | indeed :) | 22:40 |
cw | creiht: why use a partion table at all? | 22:45 |
cw | also, 1024 byte inodes are quite large ... did you really profile the amount of data in these and determine it's really better than 512 most of the time? | 22:46 |
cw | using less slab is quite desirable for millions of objects | 22:47 |
glange | cw: you'll need to get with creiht later, he's driving home right now :) | 22:47 |
redbo | yeah, 1k inodes are nice, our metadata regularly pushes close to that | 22:48 |
cw | you have that many xattrs? | 22:48 |
redbo | yeah | 22:48 |
cw | w/ attr2 they pack quite nicely so it's probably something like 950 bytes of them raw then | 22:49 |
redbo | It's a serialized python dictionary split into xattrs of 256 bytes each, I think. | 22:50 |
redbo | kind of horrible, but if anyone xattr goes over 256 bytes, it goes into an extent | 22:51 |
redbo | *any one | 22:51 |
*** xtoddx has quit IRC | 22:51 | |
cw | 8-bit size when packed into inods | 22:51 |
cw | inodes | 22:51 |
redbo | Some small percentage of objects don't fit in the inode, but 1024 gets most of them. Obviously that can vary a bit by use case. | 22:54 |
*** christiekoehler has joined #openstack | 22:54 | |
cw | redbo: is the account_stats table global? ie. can it be used to get basic billing details? | 22:55 |
cw | no, i guess it's not | 22:57 |
redbo | No, there's one per account. We don't have a list of all accounts easily accessible, I imagine we'll need to build something along those lines soon. | 22:57 |
redbo | We use our auth system for the list of accounts, basically. | 22:58 |
redbo | and go through that list and do a HEAD request to get the storage used for each account, basically | 22:58 |
_jcsmith | are those counters from the beginning of time and you do deltas, or do they reset at some point? | 22:59 |
_jcsmith | like for operations | 23:00 |
_jcsmith | number of api requests | 23:00 |
redbo | There's some stats collecting/reporting stuff that's being cleaned up and made less Rackspace billing-specific come out soon. | 23:01 |
*** lbieber has quit IRC | 23:01 | |
redbo | It works by processing log files, that's how we bill ops. We just have counters for amount of storage used. | 23:02 |
_jcsmith | would would be the best way to roll up bytes transfered ? the log files? | 23:02 |
_jcsmith | I see | 23:02 |
redbo | yeah | 23:02 |
redbo | I don't know as much about those parts, notmyname may be able to provide more information. | 23:03 |
_jcsmith | I saw some stats stuff in the code that writes to a csv, is that just for like getting metrics on the systems workload? | 23:03 |
redbo | That's more for our operations to know the state of the cluster. It reports on what percentage of things are where they're supposed to be, to give you an idea of how healthy replication is. | 23:05 |
*** algorist has quit IRC | 23:05 | |
_jcsmith | I wonder if it would make sense to allow http byte range requests through the proxy? | 23:07 |
*** Predominant has joined #openstack | 23:07 | |
redbo | We already allow requests with a Range header. Is that what you mean? | 23:07 |
_jcsmith | yeah, oh cool, I didnt realize that | 23:08 |
_jcsmith | like being able to request a part of a file | 23:08 |
redbo | yeah, the proxy just passes that along to the object servers, which know how to handle it. | 23:09 |
*** jsmith-away is now known as jsmith | 23:16 | |
*** algorist has joined #openstack | 23:20 | |
*** JordanRinke has quit IRC | 23:21 | |
*** talli has left #openstack | 23:21 | |
*** christiekoehler has quit IRC | 23:27 | |
*** jsgotangco has joined #openstack | 23:30 | |
*** silassewell has quit IRC | 23:32 | |
*** JordanRinke has joined #openstack | 23:37 | |
*** jsgotangco has quit IRC | 23:40 | |
*** silassewell has joined #openstack | 23:48 | |
*** _jcsmith has left #openstack | 23:48 | |
*** gasbakid has quit IRC | 23:49 | |
*** jsgotangco has joined #openstack | 23:53 | |
*** jsgotangco has quit IRC | 23:59 | |
*** justinsheehy has quit IRC | 00:03 | |
*** sirp1 has quit IRC | 00:06 | |
*** justinsheehy has joined #openstack | 00:08 | |
notmyname | there were questions on stats and billing in swift? | 00:08 |
*** algorist has quit IRC | 00:09 | |
*** JordanRinke has quit IRC | 00:11 | |
*** jhalogen has joined #openstack | 00:15 | |
*** cytrak has joined #openstack | 00:20 | |
*** cytrak has quit IRC | 00:23 | |
notmyname | cw: did you have some questions about tracking account stats or usage? | 00:24 |
*** stefano_s has quit IRC | 00:25 | |
*** stefano_s has joined #openstack | 00:32 | |
*** jsmith is now known as jsmith-away | 00:35 | |
creiht | cw: yeah redbo covered it pretty well. The default inode size is 256 bytes, and an average file seems to use about 100 bytes. An average object adds about 500 bytes, and then we leave a little head room for user defined metadata | 00:40 |
creiht | If I remember correctly, upping inode increased performance by a large amount | 00:41 |
notmyname | indeed it did | 00:44 |
*** Predominant has quit IRC | 00:47 | |
*** binaryWarrior has joined #openstack | 00:49 | |
redbo | One app that uses us regularly goes over what'd fit in 512 byte inodes just with object names. | 00:53 |
*** tonywolf has quit IRC | 00:57 | |
*** algorist has joined #openstack | 01:05 | |
*** jdmaturen has quit IRC | 01:22 | |
*** silassewell has quit IRC | 01:22 | |
*** tobym has joined #openstack | 01:26 | |
*** algorist has quit IRC | 01:38 | |
*** abecc has joined #openstack | 01:49 | |
*** tobym has quit IRC | 01:49 | |
*** tobym has joined #openstack | 01:53 | |
*** jdmaturen has joined #openstack | 01:54 | |
*** jtdowney has quit IRC | 01:59 | |
*** poli has joined #openstack | 02:02 | |
*** poli has joined #openstack | 02:02 | |
*** mtaylor has joined #openstack | 02:03 | |
*** ChanServ sets mode: +v mtaylor | 02:03 | |
*** Predominant has joined #openstack | 02:13 | |
*** aliguori has quit IRC | 02:22 | |
*** tobym has quit IRC | 02:24 | |
*** gnrfan has joined #openstack | 02:29 | |
*** nettok has joined #openstack | 02:59 | |
*** binaryWarrior has quit IRC | 03:03 | |
*** gnrfan has quit IRC | 03:03 | |
PiotrSikora | hm... this tweet got me thinking | 03:11 |
PiotrSikora | what are the "upstream core components" ? :) | 03:11 |
notmyname | which tweet? | 03:12 |
notmyname | I guess what I'm asking is, "what's the context?" | 03:13 |
wreese | PiotrSikora: slides are up now. http://www.oscon.com/oscon2010/public/schedule/detail/15634 | 03:13 |
pandemicsyn | notmyname: http://twitter.com/OpenStack/status/19378455506 | 03:13 |
PiotrSikora | notmyname: http://twitter.com/OpenStack/statuses/19378455506 | 03:13 |
PiotrSikora | pandemicsyn: you've won :P | 03:14 |
pandemicsyn | ;) | 03:14 |
PiotrSikora | wreese: cheers, thx :) | 03:14 |
notmyname | but pandemicsyn's link doesn't work :-) | 03:14 |
rbergeron | someone should get out the stopwatch next time.... | 03:14 |
PiotrSikora | notmyname: ah... guess i won then :) | 03:15 |
wreese | his link worked for me | 03:15 |
PiotrSikora | hmm... it seems that both links works | 03:16 |
notmyname | I'm guessing the upstream core components have more to do with nova than swift (read: I don't know). maybe things like the hypervisors? | 03:16 |
notmyname | huh. now it's working for me | 03:16 |
notmyname | I got a timeout the first time I clicked it | 03:16 |
PiotrSikora | notmyname: i don't know, i'm the one asking :) | 03:16 |
notmyname | heh | 03:16 |
PiotrSikora | who's posting under @OpenStack anyway? | 03:16 |
wreese | notmyname: you seemed surprised. it's twitter :P | 03:18 |
notmyname | ya, well I'm optimistic :-) | 03:18 |
PiotrSikora | wreese: hmm... i just had neat idea for swift while reading slides | 03:19 |
PiotrSikora | maybe its already implemented, but i dont know that ;) | 03:19 |
PiotrSikora | do you think it would be cool to replicate "hot" (highly requested) content on more nodes? | 03:20 |
PiotrSikora | to scale I/O | 03:20 |
*** binaryWarrior has joined #openstack | 03:20 | |
pandemicsyn | highly requested , like many reads (GET's) ? | 03:21 |
PiotrSikora | yeah... | 03:21 |
*** binaryWarrior has quit IRC | 03:21 | |
PiotrSikora | if (reads_in_last_10mins > XXX) { number_of_replicatas++; } | 03:22 |
*** binaryWarrior has joined #openstack | 03:22 | |
notmyname | that implies that the number of replicas is set per object | 03:22 |
PiotrSikora | yes, it does | 03:22 |
notmyname | which isn't supported now, but may be in the future for reduced redundancy type things | 03:23 |
PiotrSikora | k | 03:23 |
PiotrSikora | i just thought i'll share the idea :) | 03:23 |
wreese | what he said and you would need to randomize the read requests across servers as well. :) | 03:23 |
PiotrSikora | i did something like this for web backends last year | 03:24 |
PiotrSikora | ngx_supervisord - it is bringing backend servers depending on current load | 03:24 |
notmyname | I think the way it works now is to request from each server in sequence and return the data from the first object server that doesn't error | 03:24 |
PiotrSikora | so if there are many request to the website it would bring up to max backends | 03:24 |
notmyname | rather than concurrently request from all copies and return the first one | 03:24 |
PiotrSikora | but it would cool-down to just 1 when there were no loads | 03:25 |
PiotrSikora | (nginx module) | 03:25 |
PiotrSikora | wreese, notmyname: yeah... but maybe that will change by the time for 2.0 :) | 03:25 |
notmyname | it's not a bad idea. but it would require some changes to the way things work. (patches welcome!) | 03:25 |
notmyname | we've talked about how to do arbitrary redundancy stuff (or talked about wanting to talk about it) | 03:26 |
*** lbieber has joined #openstack | 03:26 | |
gholt | Currently, a CDN layer provides the requested-a-lot niche. But, obviously only for that particular use case. | 03:26 |
notmyname | if we supported it, a "hot files" concept might be a performance improvement. I'd want to see tests, though | 03:27 |
wreese | not to mention a single server can handle a few thousand requests per second (depending on file size, hardware, network) | 03:27 |
pandemicsyn | could do something on the proxy's like ye'old henhouse cache...didn't work out so well there though | 03:28 |
PiotrSikora | wreese: agreed, but i guess when we're talking about 100PBs of data this might become bottleneck at one point ;) | 03:29 |
wreese | true, but we haven't run into that yet. but we have limelight to thank for that :) | 03:29 |
PiotrSikora | :) | 03:30 |
PiotrSikora | notmyname: well, it would be performance improvement for sure, there is no reason why it wouldn't be... but i guess you wouldn't notice it under "normal" circumstances | 03:33 |
notmyname | ya. that's all I mean. would the extra complexity give real world improvements for enough use cases to justify it? | 03:34 |
PiotrSikora | for few use cases, maybe... definiately not for many | 03:35 |
PiotrSikora | but like i said... it was "cool idea" :) | 03:35 |
notmyname | definitely. I do like it | 03:35 |
notmyname | and it may be one of those things that may be more appropriate as a tunable feature that can be set per-deployment. so nasa can set it high for their mars image viewer that has a swift backend, and cloud files can turn it down for the more general use case | 03:36 |
PiotrSikora | agreed | 03:37 |
rbergeron | where's the "awesome ideas for the future" wiki / mailing list / how to capture ideas that will be great down the road spot? | 03:41 |
PiotrSikora | rbergeron: i just started one on the paper on my desk :P | 03:42 |
notmyname | I'm no launchpad expert, but is that something that can be captured there? in blueprints or bugs or something? | 03:42 |
rbergeron | yeah - i'm not sure. | 03:42 |
* rbergeron usually follows the "it's a wiki, be bold" rule - and at least have a place to say "if you have ideas, here's what to do with them" | 03:42 | |
rbergeron | even if it's a pointer elsewhere | 03:43 |
* rbergeron goes and looks at launchpad | 03:43 | |
rbergeron | oh, airplane battery fail. why is my cord in cargo.... | 03:43 |
notmyname | why not add a page to wiki.openstack.org? | 03:44 |
rbergeron | that's an excellent suggestion. | 03:44 |
* rbergeron will do that... probably when she gets home, unless someone beats her to it, since my laptop is about to go byebye. | 03:45 | |
* rbergeron wonders if that is what a blueprint is | 03:46 | |
mtaylor | short version from /me ... | 03:46 |
mtaylor | bug == what - blueprint == how | 03:47 |
rbergeron | yeah | 03:47 |
mtaylor | this is the meme I keep trying to sell people on - of course, I could be wrong :) | 03:47 |
rbergeron | well - do you think of it as bug - or "feature request" | 03:48 |
PiotrSikora | mtaylor: but features arent bugs | 03:48 |
rbergeron | well. | 03:48 |
* rbergeron grins | 03:48 | |
rbergeron | i think noting that kind of thing in the wiki at least as pointing to the blueprints spot is a reasonable idea | 03:49 |
* rbergeron will wield her wiki skillz | 03:50 | |
*** littleidea has quit IRC | 03:53 | |
PiotrSikora | wreese: nice slides, got me few ideas :) | 03:54 |
PiotrSikora | other then the "hot files" feature | 03:54 |
PiotrSikora | wouldn't it be a good idea to off-load serving files to "real webserver"? | 03:55 |
PiotrSikora | so python proxy sould only send X-sendfile with file location to nginx/lighttp/etc and it would server the content itself | 03:55 |
PiotrSikora | i don't think that serving GB+ files on through any high level language is good idea | 03:56 |
mshadle | exactly | 03:56 |
PiotrSikora | Hi Mike :) | 03:57 |
mshadle | yoyo | 03:57 |
mshadle | i need to make sure my drupal install uses x-accel-redirect | 03:57 |
*** binaryWarrior has quit IRC | 03:59 | |
*** littleidea has joined #openstack | 04:00 | |
wreese | PiotrSikora: the python/eventlet servers have no trouble saturating the 10GbE links on the proxies, much less the 1GbE links on the storage nodes. x-sendfile in c with zero-copy would be more efficient, but you would be optimizing where optimization isn't needed. | 04:08 |
PiotrSikora | but it still reads (copies) content of the file to the userspace | 04:10 |
PiotrSikora | and sendfile is handled on the kernel level | 04:10 |
PiotrSikora | but yeah... it probably isn't a bottleneck ;) | 04:11 |
wreese | right, but the cpu on those servers is nearly idle, while everything is waiting on disk and network. there really isn't a need to optimize there, at least for any use case we've seen. | 04:11 |
PiotrSikora | ACK | 04:12 |
pandemicsyn | SYN | 04:12 |
PiotrSikora | btw: would you mind sharing some numbers on what swift handles in production? | 04:13 |
notmyname | our product guy really doesn't want us to do that :-) | 04:13 |
PiotrSikora | hehe | 04:13 |
PiotrSikora | but are we talking about few PB and GBps or hundreds? | 04:13 |
PiotrSikora | Gbps* | 04:13 |
PiotrSikora | just the order of magnitude :) | 04:14 |
wreese | we worked very hard to be as objective as possible everytime we were tempted to introduce an optimization. it's better to keep things simple and manageable | 04:14 |
PiotrSikora | wreese: thats great | 04:15 |
PiotrSikora | ...but in this case this depends a lot on your background | 04:15 |
PiotrSikora | for me using proxy as "traffic router" (tell me where the data is) would be easier :) | 04:16 |
pandemicsyn | how come ? | 04:16 |
wreese | that's basically what the system does, but sometimes the data isn't where you think it is and you need to take a different course of action. that's why you can't hand of the request to a "real webserver". | 04:17 |
wreese | the proxy has to use many techniques to ensure it's able to locate and succesfully send the correct data to the client | 04:18 |
PiotrSikora | pandemicsyn: erm... i meant it on the design/conceptual level | 04:18 |
PiotrSikora | wreese: but couldn't real webserver do this as well? | 04:18 |
PiotrSikora | or to put it differently | 04:18 |
creiht | What problem would you be trying to solve by putting a *real* webserver in front of it? | 04:19 |
pandemicsyn | seems like you'd just be recreating the proxy-server as a nginx module | 04:19 |
PiotrSikora | there is a moment when proxy already used all those techniques to ensure that data is there | 04:19 |
PiotrSikora | and instead of sending data it would sent x-sendfile header | 04:20 |
PiotrSikora | creiht: no problem, it was just a thought | 04:20 |
creiht | ahh | 04:20 |
creiht | If it aint broke, don't fix it :) | 04:20 |
PiotrSikora | pre-optimialization, if you will ;) | 04:21 |
creiht | on the other side of things there are still a lot of interesting things to figure out | 04:21 |
creiht | Like container meta data | 04:21 |
notmyname | that's a good one. let's do that | 04:21 |
creiht | :) | 04:21 |
wreese | we will be :) | 04:21 |
creiht | wreese: I guess we should talk about that on monday to plan a real roadmap :) | 04:22 |
wreese | yeah. right after I complete your reviews :P | 04:22 |
notmyname | wreese: are you ready for the endless stream of "hey, Will"? | 04:22 |
creiht | haha | 04:22 |
redbo | We did play around with using sendfile(2), but we were pretty hard-pressed to make it show any speed improvements over the read/send loop. The performance characteristics are sort of bound by disk i/o. | 04:22 |
PiotrSikora | redbo: ah, ok | 04:24 |
wreese | and that's coming from the guy that always wants to rewrite certain sections in c ;) | 04:24 |
PiotrSikora | hehe :) | 04:24 |
PiotrSikora | i know the type ;) | 04:25 |
redbo | when I rewrite things in C, nobody will ever look at it :( | 04:26 |
wreese | haha | 04:26 |
*** gustavomzw has quit IRC | 04:31 | |
*** lbieber has quit IRC | 04:42 | |
*** nettok has quit IRC | 04:46 | |
*** tjyang has quit IRC | 04:53 | |
*** rajijoom has joined #openstack | 05:07 | |
PiotrSikora | hmm.. what is ozone subproject? | 05:14 |
redbo | I think that name has been retired. It was the group/project working on rewriting Cloud Servers in Rackspace. At least the project part has been retired in favor of nova. | 05:15 |
PiotrSikora | thx | 05:16 |
*** z0 has quit IRC | 05:19 | |
*** jhalogen has quit IRC | 05:21 | |
*** bretpiatt has joined #openstack | 05:22 | |
*** justinsheehy has quit IRC | 05:26 | |
*** rajijoom has quit IRC | 05:27 | |
*** justinsheehy has joined #openstack | 05:32 | |
*** tobym has joined #openstack | 05:38 | |
*** norm_ has joined #openstack | 05:41 | |
*** z0 has joined #openstack | 05:43 | |
norm_ | what would be some good books for getting familiar with cloud computing? I admin VMware VMs, and am not clear on what a cloud would look like | 05:43 |
*** chelz has joined #openstack | 05:46 | |
*** sirp2 has joined #openstack | 05:48 | |
chelz | i just watched the video from OSCON about OpenStack and the parts asking for input on how to best "do" open source reminded me of this great talk by Brian Fitzpatrick and Ben Collins-Sussman called "The Myth of the Genius Programmer" from Google I/O 2009 about collaboration and related fun stuff. it's 55min long: http://www.youtube.com/watch?v=0SARbwvhupQ | 06:01 |
chelz | they've also given two talks that are available on either or both google video and youtube on the topic of "poisonous people". all very good. | 06:02 |
mshadle | openstack should be redone in javascript. and use node.js! i'd be able to contribute easier then | 06:03 |
mshadle | or go the php route as previously suggested! | 06:03 |
mshadle | mtaylor: get onit | 06:03 |
*** tobym has quit IRC | 06:06 | |
*** norm_ has quit IRC | 06:06 | |
*** littleidea has quit IRC | 06:23 | |
*** sirp2 has quit IRC | 06:25 | |
*** jdmaturen has quit IRC | 06:28 | |
*** jxta has joined #openstack | 06:38 | |
*** yoko has joined #openstack | 06:46 | |
*** jxta has quit IRC | 06:47 | |
*** rajijoom has joined #openstack | 06:54 | |
*** allsystemsarego has joined #openstack | 07:17 | |
*** abecc has quit IRC | 07:22 | |
*** abecc has joined #openstack | 07:22 | |
*** viewlogic has quit IRC | 07:29 | |
*** bretpiatt has left #openstack | 07:33 | |
*** yoko has quit IRC | 07:33 | |
*** yoko has joined #openstack | 07:34 | |
*** pdelgallego has joined #openstack | 07:36 | |
*** abecc has quit IRC | 07:45 | |
*** yoko has quit IRC | 08:30 | |
*** poli has quit IRC | 08:30 | |
*** chelz has quit IRC | 10:05 | |
*** justinsheehy has quit IRC | 10:48 | |
*** justinsheehy has joined #openstack | 10:53 | |
*** tjyang has joined #openstack | 11:00 | |
*** littleidea has joined #openstack | 11:07 | |
*** littleidea has quit IRC | 11:17 | |
*** jtdowney has joined #openstack | 11:36 | |
*** silassewell has joined #openstack | 11:59 | |
*** tjyang has quit IRC | 12:29 | |
*** jsmith-away is now known as jsmith | 12:46 | |
*** justinsheehy has quit IRC | 13:09 | |
*** justinsheehy has joined #openstack | 13:14 | |
*** Predominant has quit IRC | 13:46 | |
*** jtdowney has quit IRC | 13:47 | |
*** jtdowney has joined #openstack | 13:47 | |
*** lbieber has joined #openstack | 14:02 | |
*** lbieber has quit IRC | 14:09 | |
*** binaryWarrior has joined #openstack | 14:32 | |
*** littleidea has joined #openstack | 14:56 | |
*** allsystemsarego has quit IRC | 15:02 | |
*** rajijoom has quit IRC | 15:04 | |
*** _anm has joined #openstack | 15:15 | |
_anm | I'm getting a: nova.exception.NotFound: None | 15:18 |
_anm | when trying to upload a bundle to a new bucket | 15:18 |
*** Predominant has joined #openstack | 15:19 | |
_anm | it actually dumps a huge html exception file | 15:19 |
_anm | to th console | 15:19 |
*** sirp2 has joined #openstack | 15:26 | |
*** justinsheehy has quit IRC | 15:40 | |
*** littleidea has quit IRC | 15:44 | |
*** justinsheehy has joined #openstack | 15:45 | |
*** lbieber has joined #openstack | 15:56 | |
*** gasbakid has joined #openstack | 16:12 | |
*** jsmith is now known as jsmith-away | 16:19 | |
*** jtdowney has quit IRC | 16:27 | |
*** poli has joined #openstack | 16:33 | |
*** sirp2 has quit IRC | 16:38 | |
*** Predominant has quit IRC | 16:39 | |
*** Predominant has joined #openstack | 16:41 | |
*** dendro-afk is now known as dendrobates | 16:44 | |
*** lbieber has quit IRC | 16:48 | |
*** Predominant has quit IRC | 16:53 | |
*** bretpiatt has joined #openstack | 16:57 | |
*** Predominant has joined #openstack | 16:58 | |
*** silassewell has quit IRC | 17:09 | |
*** jeromatron has joined #openstack | 17:10 | |
*** jsmith-away is now known as jsmith | 17:10 | |
*** Predominant has quit IRC | 17:14 | |
*** Predominant has joined #openstack | 17:16 | |
mtaylor | mshadle: wha? | 17:22 |
*** Predominant has quit IRC | 17:31 | |
*** blp has joined #openstack | 17:31 | |
*** bretpiatt has quit IRC | 17:31 | |
*** blp is now known as bpiatt | 17:31 | |
*** bpiatt is now known as bretpiatt | 17:31 | |
PiotrSikora | mtaylor: i guess mshadle forgot to use the <sarcasm /> sign ;) | 17:35 |
mtaylor | PiotrSikora: :) | 17:35 |
*** poli has quit IRC | 17:39 | |
*** clayg has quit IRC | 17:43 | |
*** clayg- has joined #openstack | 17:44 | |
*** jhalogen has joined #openstack | 17:51 | |
*** jonesy_ has joined #openstack | 17:59 | |
*** sander__ has quit IRC | 18:16 | |
*** jdmaturen has joined #openstack | 18:17 | |
*** jdmaturen has quit IRC | 18:18 | |
*** snowboarder04 has quit IRC | 18:26 | |
*** PiotrSikora has quit IRC | 18:31 | |
*** sander__ has joined #openstack | 18:31 | |
*** PiotrSikora has joined #openstack | 18:32 | |
_anm | can these wikis be edited? http://swift.openstack.org/deployment_guide.html. I'd like to clean up some typos and grammer | 18:37 |
_anm | or are these built from source? | 18:37 |
*** PiotrSikora has quit IRC | 18:37 | |
*** ewan_ has joined #openstack | 18:39 | |
eday | _anm: those are built from source, see swift/doc/source | 18:40 |
*** littleidea has joined #openstack | 18:40 | |
eday | _anm: you can just fix/propose merges on launchpad just like other src changes | 18:41 |
bretpiatt | _anm wiki.openstack.org is the doc repository that can be edited | 18:41 |
bretpiatt | but we really should fix in swift/doc/source as it'll be more detailed than the stuff on wiki | 18:42 |
_anm | ok, i'll submit patches to launchpad. | 18:44 |
eday | _anm: thanks! | 18:45 |
_anm | my project contributios usually revolve around usability, docs, and setup issues, end-user admin stuff | 18:45 |
*** ewan_ has quit IRC | 18:46 | |
bretpiatt | awesome, building rome doesn't help without a guide book for people to avoid getting lost in it :) | 18:46 |
_anm | so, without contrib priveleges, do I just submit issues with patches attached? I'm very new to bzr/launchapad - I'm reading the wiki about it now | 18:49 |
*** jdmaturen has joined #openstack | 18:50 | |
*** PiotrSikora has joined #openstack | 18:50 | |
eday | _anm: You should follow the branch/merge propsal for just as you would for source. Once your changes are proposed, folks can review and approve them to get merged into trunk | 18:50 |
eday | _anm: at that point we have automated systems to merge and update the docs | 18:51 |
_anm | ok, thanks | 18:51 |
*** clayg- is now known as clayg | 18:51 | |
mtaylor | yay automated systems! | 18:53 |
eday | yay mtaylor for setting them all up! | 18:53 |
eday | mtaylor: you still on vacation? | 18:53 |
mtaylor | eday: nope. I'm back now | 18:53 |
*** WaterDog is now known as DogWater | 18:54 | |
*** PiotrSikora has quit IRC | 18:54 | |
eday | mtaylor: ahh. feeling mroe relaxed? you were in Cozumel correct? | 18:55 |
mtaylor | eday: yes. it was a very useful vacation - I'm feeling great now | 18:56 |
*** copilot_ has joined #openstack | 18:57 | |
*** jsmith is now known as jsmith-away | 18:57 | |
*** copilot has quit IRC | 18:58 | |
*** PiotrSikora has joined #openstack | 18:59 | |
*** binaryWarrior has quit IRC | 19:03 | |
*** binaryWarrior has joined #openstack | 19:06 | |
*** snowboarder04 has joined #openstack | 19:09 | |
*** sirp2 has joined #openstack | 19:10 | |
*** jonesy_ has quit IRC | 19:17 | |
*** bretpiatt has quit IRC | 19:17 | |
*** ewan_ has joined #openstack | 19:22 | |
notmyname | _anm: be sure to sign the CLA too. we can't accept your patches until you do | 19:27 |
*** sirp2 has quit IRC | 19:30 | |
*** littleidea has quit IRC | 19:35 | |
*** binaryWarrior has quit IRC | 19:38 | |
*** littleidea has joined #openstack | 19:40 | |
*** PiotrSikora has quit IRC | 19:42 | |
*** PiotrSikora has joined #openstack | 19:44 | |
*** allsystemsarego has joined #openstack | 19:51 | |
*** PiotrSikora has quit IRC | 19:52 | |
*** snowboarder04 has quit IRC | 19:53 | |
*** jeromatron has left #openstack | 19:55 | |
*** lbieber has joined #openstack | 19:56 | |
*** jonesy_ has joined #openstack | 19:57 | |
*** PiotrSikora has joined #openstack | 19:59 | |
* mtaylor really needs to get the CLA signing integrated into the merge-request system... | 19:59 | |
*** jonesy_ has quit IRC | 20:01 | |
*** PiotrSikora has quit IRC | 20:04 | |
*** PiotrSikora has joined #openstack | 20:06 | |
jtimberman | notmyname: how much flak have you guys gotten about CLAs? | 20:07 |
jtimberman | we've gotten a little bit (on chef) | 20:07 |
creiht | So far it hasn't been that bad | 20:09 |
creiht | The e signing is very nice | 20:09 |
jtimberman | yes. echosign was awesome for us. | 20:09 |
jtimberman | lot higher rate of CLA signing w/ it | 20:09 |
*** PiotrSikora has quit IRC | 20:10 | |
*** PiotrSikora has joined #openstack | 20:13 | |
notmyname | what creiht said | 20:13 |
creiht | hehe | 20:13 |
notmyname | mtaylor: would that tie the name with the lp account. cause when I see a merge request, I don't really have a good way to map the two | 20:15 |
mtaylor | notmyname: yes. | 20:16 |
mtaylor | notmyname: the eventual goal there will be that the system will just not let you file a merge request unless you've signed | 20:17 |
notmyname | ooohh. fancy | 20:17 |
mtaylor | it's on my todo list of patches to write for launchpad :) | 20:17 |
notmyname | I've realized that I have a problem not ending questions with "?". I need to work on that. | 20:18 |
*** snowboarder04 has joined #openstack | 20:18 | |
mtaylor | nah. why bother ;) | 20:18 |
*** jhalogen has quit IRC | 20:23 | |
*** lbieber has quit IRC | 20:28 | |
*** ewan_ has quit IRC | 20:44 | |
*** littleidea has quit IRC | 20:49 | |
mtaylor | http://hudson.openstack.org/job/swift-coverage/cobertura | 20:53 |
*** jdmaturen has quit IRC | 20:54 | |
notmyname | that's pretty neat | 20:56 |
mtaylor | will get updated on every merge | 20:56 |
*** dendrobates is now known as dendro-afk | 20:57 | |
_anm | mtaylor: do you have a link to the CLA | 20:59 |
_anm | ? | 20:59 |
*** gasbakid has quit IRC | 21:00 | |
notmyname | _anm: http://wiki.openstack.org/HowToContribute | 21:00 |
notmyname | linked on that page | 21:00 |
*** gasbakid has joined #openstack | 21:02 | |
_anm | notmyname: DONE! | 21:04 |
notmyname | mtaylor: now that's what I'm talking about. how do I match _anm with someone on the approved contributors list? /whois says his name is "real real", and I don't see that on the list ;-) | 21:06 |
_anm | do i need to use my real name here? | 21:06 |
notmyname | no | 21:06 |
notmyname | just giving mtaylor a hard time | 21:06 |
_anm | thsi IS IRC, you know :) | 21:06 |
_anm | i need a bit of anonymity around these parts | 21:07 |
mtaylor | hah | 21:07 |
notmyname | hey, my name is "notmyname" | 21:07 |
mtaylor | well.. you _do_ need to set up your bzr whoami to be some name that matches what you signed the CLA as | 21:07 |
mtaylor | and honestly - if we're going to be doing this, we should probably require signed commits... | 21:08 |
notmyname | (of course, I put my real name in whois, but I find it useful in openstack stuff) | 21:08 |
_anm | yeah, IRC is where i claim "non-attribution" on anything i say | 21:08 |
_anm | mtaylor: Ok | 21:08 |
mtaylor | but we don't require signing | 21:08 |
mtaylor | I'm just faffing about atm | 21:08 |
mtaylor | notmyname: hey! swift is pretty clean pep8-wise | 21:09 |
notmyname | ya, creiht and gholt beat me into submission about that :-) | 21:09 |
mtaylor | oh -nevermind | 21:10 |
mtaylor | I should have run with --repeat to show all instances of a particular violation :) | 21:10 |
notmyname | hehe | 21:10 |
mtaylor | still, only 67 violations - not nearly as bad as pylint ;) | 21:10 |
notmyname | mtaylor: in reference to the whoami matching the CLA, how does that work with name versus display name in launchpad? for example, I've got my real name as display name, but notmyname as my name (launchpad.net/~notmyname) | 21:15 |
mtaylor | notmyname: not fully sure ... we may just have to eyeball make some sort of mapping list until we get actual integration going | 21:16 |
notmyname | ok | 21:16 |
notmyname | also, how does signing work? does it use the OpenPGP key that lp has a field for? | 21:17 |
mtaylor | signing commits? | 21:17 |
notmyname | ya | 21:17 |
mtaylor | that's actually a local thing - you can either run "bzr sign-my-commits" | 21:18 |
mtaylor | and it'll go through and sign all commits in the tree that you made | 21:18 |
mtaylor | or, you can put create_signatures = always in the [DEFAULT] section of ~/.bazaar/bazaar.conf and it'll sign each commit as you make it | 21:19 |
mtaylor | nothing really consumes this at the moment- but I personally think it would be cool to eventually put in a filter in the merge testing thing to make sure all commits being merged are signed by a person we know | 21:20 |
*** ewan_ has joined #openstack | 21:20 | |
rbergeron | how does the cla work wrt wiki, documentation, etc? ie - non-code type contributions? same story? | 21:20 |
rbergeron | must sign to edit? | 21:20 |
* rbergeron nearly started wiki-ing away until she thought otherwise re: cla signing | 21:21 | |
mtaylor | right now it's just related to things in the source tree | 21:21 |
_anm | i was making comnits to the wiki with justinsb yesterday, sooo not sure | 21:22 |
mtaylor | I have heard nothing that indicate we want a CLA for wiki edits - if the lawyers think that's a good idea, I think they're crazy | 21:22 |
mtaylor | so go for it :) | 21:23 |
_anm | wiki.openstack.org wiki, not the src built wikis | 21:23 |
mtaylor | yes | 21:23 |
mtaylor | well, the other docs aren't wikis as much as they're just html documenation - but yes | 21:23 |
notmyname | mtaylor: so sign-my-commits works (that is --dry-run works). but what key is it using to sign? | 21:25 |
mtaylor | notmyname: it's using whatever your primary gpg key is ... | 21:25 |
mtaylor | it _should_ be matching your key id with the email listed in the committer in the commit message I think | 21:25 |
mtaylor | looking at source... | 21:26 |
mtaylor | yeah. that seems to be correct | 21:28 |
termie | mtaylor: heya, if i want to make a new launchpad project to test some bzr stuff, how do i go about that? | 21:29 |
notmyname | hmm. forgive my ignorance, but I'm not even sure that I have gpg installed. I guess I must.. | 21:29 |
termie | i see this ~termie/+junk/<new project> stuff but i'd prefer to have something that was more like an actual project | 21:29 |
mtaylor | termie: hrm - are you just wanting to poke around with bzr things? or else what sort of project things do you want to poke at? | 21:30 |
mtaylor | termie: (just trying to understand so I can give good advice) | 21:30 |
*** allsystemsarego has quit IRC | 21:31 | |
termie | mtaylor: i've just found 'staging.launchpad.net', reading through it, what i am doing is testing a git<--> bzr bridge | 21:31 |
termie | mtaylor: and want to put it through some paces | 21:31 |
mtaylor | termie: ah. k. | 21:31 |
notmyname | mtaylor: here's a bzr bug: bzr sign-my-commits --dry-run works even if gpg isn't installed (bzr sign-my-commits fails without the --dry-run option). this makes it kinda hard to play with | 21:32 |
mtaylor | notmyname: heh. well - oh- so I sort of misspoke earlier | 21:33 |
mtaylor | notmyname: regarding order of things... | 21:33 |
*** jtdowney has joined #openstack | 21:33 | |
mtaylor | it looks to see what your bzr whoami is - then it looks for revision where the committer matches that | 21:33 |
mtaylor | then it tries to sign that | 21:34 |
mtaylor | so the list of revisions reported by sign-my-commits should still be what it's expecting to be able to sign | 21:34 |
notmyname | ok | 21:35 |
*** gnrfan has joined #openstack | 21:36 | |
notmyname | I see that, but it doesn't help me figure out what to set up or do for signing until I actually want to do it | 21:36 |
mtaylor | hrm. ah, I see what you're saying | 21:37 |
termie | mtaylor: are you affiliated with launchpad in any way? i assumed so from your comment before but i may be mistaken | 21:38 |
*** lbieber has joined #openstack | 21:38 | |
termie | mtaylor: don't want to keep bugging you with launchpad issues if you're not ;) | 21:38 |
*** jhalogen has joined #openstack | 21:40 | |
mtaylor | termie: sort of - I'm with rackspace, but one of the things I do for the project is hack on launchpad ... so I'm the right person to bug here | 21:42 |
mtaylor | termie: I think there may be some mix up though in what a project is and what a branch is ... | 21:43 |
mtaylor | termie: I'd use the terms "~termie/+junk/<new branch>" above- because a project is really just a collection of a bunch of things like bug trackers and possibly branches | 21:44 |
mtaylor | termie: but if you're looking to test out git/bzr integration, then all you need is a branch or two, really | 21:44 |
termie | mtaylor: aye, i'll see where it goes | 21:44 |
termie | mtaylor: staging.launchpad.net times out on loading a project, btw | 21:45 |
*** binaryWarrior has joined #openstack | 21:45 | |
mtaylor | unlike git, there is nothing special server side needed for a branch (you don't have to do a remote repo initialization ... so the step you need in github of creating a place to stick something just isn't relevant here ) | 21:45 |
mtaylor | termie: excellent | 21:45 |
mtaylor | BUT - there is of course nothing wrong with creating things and poking at them :) | 21:45 |
mtaylor | other than timeouts | 21:45 |
termie | mtaylor: does rackspace's side of openstack have anybody in the european timezones? | 21:48 |
termie | i notice the timezone feature on lp but haven't yet figured out how to use it | 21:48 |
mtaylor | termie: soren | 21:48 |
mtaylor | termie: soren is in denmark | 21:48 |
termie | (or tried to figure out, i am mostly in the asking you all the questions that pop to mind mode :p) | 21:49 |
mtaylor | hehe | 21:49 |
*** snowboarder04 has quit IRC | 21:49 | |
*** metoikos has joined #openstack | 21:49 | |
mtaylor | termie: if you look at a team page, such as: https://edge.launchpad.net/~nova | 21:50 |
*** ewan_ has quit IRC | 21:50 | |
mtaylor | termie: you'll see a location map with pins in it for team members who have indicated their location | 21:51 |
*** jdmaturen has joined #openstack | 21:51 | |
termie | ah, nifty :) | 21:51 |
*** snowboarder04 has joined #openstack | 21:56 | |
*** justinsheehy has quit IRC | 22:26 | |
*** justinsheehy has joined #openstack | 22:32 | |
*** lbieber has quit IRC | 22:32 | |
*** jtdowney has quit IRC | 22:40 | |
*** jtdowney has joined #openstack | 22:40 | |
mtaylor | creiht: hey - what platform are you on? | 22:45 |
mtaylor | creiht: you had that problem a while back about build_sphinx not working and I had to change something - I was going to try to track down what the actual problem was | 22:45 |
*** silassewell has joined #openstack | 22:47 | |
mtaylor | soren: hey. I got coverage set up in hudson for swift, but it doesn't work with nova | 22:54 |
mtaylor | soren: running run_tests.py through coverage causes tests to fail :( | 22:54 |
gasbakid | i have the same problem with the build_sphinx | 22:54 |
mtaylor | gasbakid: oh you do? great! what sort of problem are you seeing? and what version of python-sphinx do you have installed? | 22:55 |
gasbakid | it tells me that its an invalid command | 22:59 |
gasbakid | the build sphinx one | 22:59 |
mtaylor | ah. well that ones easier - you need to have sphinx installed ... do you? | 22:59 |
gasbakid | yes | 22:59 |
mtaylor | hrm. what versoin? | 22:59 |
mtaylor | sphinx-build --version 2>&1 | head -1 | 23:01 |
gasbakid | 0.6. | 23:02 |
gasbakid | 0.6.4 | 23:02 |
gasbakid | i think i was not on the right version | 23:04 |
gasbakid | it passes when i update the package and i dont know on what version i was | 23:05 |
*** jhalogen has quit IRC | 23:08 | |
mtaylor | heh. well that's ok - at least it's working for you :) | 23:11 |
mtaylor | http://hudson.openstack.org/job/nova-pep8/violations/? | 23:11 |
*** jdmaturen has quit IRC | 23:11 | |
gasbakid | yeahh great thankssssssssssss | 23:11 |
*** justinsheehy has quit IRC | 23:27 | |
*** justinsheehy has joined #openstack | 23:32 | |
*** poli has joined #openstack | 23:44 | |
creiht | mtaylor: I'm on Ubuntu 10.04, and I can remember what the error that I had was | 23:45 |
mtaylor | 10.04 -- that's karmic, right? | 23:45 |
mtaylor | oh wait - no, that's lucid. hrm | 23:45 |
creiht | yup lucid | 23:51 |
mtaylor | hrm. weird | 00:01 |
mtaylor | oh well | 00:01 |
* creiht votes that nova should change their docs directory to doc :) | 00:02 | |
mtaylor | heh | 00:03 |
mtaylor | I suppose I could also submit a branch that does that... | 00:03 |
mtaylor | OR - I could just change both of you to an independent third approach | 00:04 |
creiht | hah | 00:04 |
creiht | hrm | 00:04 |
creiht | so in the new approach, there is no makefile for the docs | 00:04 |
creiht | (or at least in your branch) | 00:04 |
mtaylor | yeah... is it really useful anymore? | 00:05 |
creiht | If you want to build something other than html | 00:05 |
mtaylor | you mean like : python setup.py build_sphinx -b latex | 00:05 |
creiht | oh heh | 00:05 |
mtaylor | :) | 00:05 |
mtaylor | actually - I was thinking - I should build the latex docs and publish the pdf too... | 00:06 |
creiht | is there a reason the build directory is named _build trather than build? | 00:07 |
creiht | or is that just a sphinx default? | 00:07 |
mtaylor | guh... erm... gah... | 00:07 |
creiht | hah | 00:07 |
mtaylor | ah | 00:08 |
mtaylor | "You have two options for placing the build directory for Sphinx output. | 00:08 |
mtaylor | Either, you use a directory "_build" within the root path, or you separate | 00:08 |
mtaylor | "source" and "build" directories within the root path. | 00:08 |
mtaylor | " | 00:08 |
creiht | ahh | 00:08 |
mtaylor | so it seems that they picked one of the branches and you picked the other :) | 00:08 |
creiht | right | 00:08 |
creiht | if we keep it this way, we should change .bzrignore to ignore the docs dir | 00:09 |
mtaylor | mmm. good call | 00:09 |
creiht | I mean the _build dir | 00:09 |
*** lbieber has joined #openstack | 00:11 | |
creiht | mtaylor: and the README has to change | 00:11 |
*** silassewell has quit IRC | 00:12 | |
*** polvi has joined #openstack | 00:12 | |
mtaylor | creiht: k. pushed those changes | 00:12 |
mtaylor | creiht: mind you - I'm not convinced this is the right thing to do | 00:13 |
creiht | heh | 00:13 |
creiht | I guess I don't have a strong opinion, I kinda preferred the separate dirs, but I could live either way | 00:13 |
mtaylor | lemme make the mirror patch for nova... | 00:14 |
creiht | I'll make a comment on the merge request, and see what the other guys | 00:15 |
creiht | have to say | 00:15 |
*** jdmaturen has joined #openstack | 00:24 | |
mtaylor | creiht: ok. I think I like your approach with the source/build dirs better - but I they've got some nice config stuff done | 00:37 |
*** pdelgallego has quit IRC | 00:41 | |
*** lbieber has quit IRC | 00:45 | |
*** jdmaturen has quit IRC | 00:47 | |
*** littleidea__ has joined #openstack | 00:48 | |
*** gasbakid has quit IRC | 00:57 | |
*** littleidea_ has joined #openstack | 01:01 | |
*** littleidea__ has quit IRC | 01:05 | |
creiht | mtaylor: what does superceded mean? | 01:09 |
creiht | superseded even | 01:09 |
mtaylor | creiht: it means I clicked "resubmit proposal" | 01:09 |
creiht | ahh | 01:10 |
mtaylor | sort of a way to trigger that I had just overwritten the tree and stuff | 01:10 |
creiht | I see now | 01:11 |
creiht | mtaylor: I like that better, though I get a warning when building the docs about the intersphinx stuff | 01:16 |
mtaylor | oh do you? | 01:16 |
mtaylor | creiht: what warning do you get? | 01:16 |
mtaylor | and what version of sphinx are you running? | 01:16 |
mtaylor | sphinx-build --version 2>&1 | head -1 | 01:17 |
creiht | mtaylor: http://pastebin.com/8AwjKPdL | 01:17 |
mtaylor | FASCINATING | 01:17 |
creiht | 0.6.5 | 01:17 |
mtaylor | ah | 01:17 |
mtaylor | it's becomming clear to me now | 01:17 |
* mtaylor has 1.0b1 installed | 01:17 | |
creiht | ahh | 01:18 |
mtaylor | I betcha this is the cause of our previous issue too... | 01:18 |
creiht | intersphinx was available in .5+, but maybe the use of it changed | 01:18 |
mtaylor | yeah - the config dict it gets is differnet now | 01:18 |
mtaylor | different even | 01:18 |
mtaylor | mine is installed from the drizzle-developers ppa. bother | 01:19 |
creiht | oh interesting | 01:19 |
mtaylor | I guess I should spin up a vm that has the version of sphinx other folks have | 01:19 |
creiht | easy_install -U sphinx gives me 1.0 now | 01:19 |
mtaylor | OR - we could require 1.0 if someone wants to build the docs... | 01:19 |
creiht | hehe | 01:20 |
creiht | hrm | 01:20 |
* mtaylor is torn | 01:20 | |
mtaylor | one of them is less work for /me | 01:20 |
creiht | sphinx website says current version is 0.6.7 | 01:20 |
creiht | not sure if it just got updated | 01:21 |
creiht | the otherside is, docs still build with the other version, they just get that warning (and no crosslinking) | 01:21 |
creiht | so I'm thinking that requiring 1.0 to build may be alright | 01:21 |
creiht | heh... I reload the sphinx page, and current version is now 1.0 :) | 01:22 |
mtaylor | hahaha | 01:22 |
creiht | So I'm fine with the 1.0 req | 01:22 |
mtaylor | cool. cause yeah - it's not a _hard_ req | 01:22 |
creiht | right | 01:23 |
* creiht goes to look at what is new in 1.0 :) | 01:23 | |
creiht | the intersphinx stuff is cool... I hadn't known about that before | 01:24 |
*** lbieber has joined #openstack | 01:28 | |
mtaylor | morning lbieber ! | 01:30 |
notmyname | creiht: mtaylor: I don't have strong feelings about the docs. I think they need to be buildable and accessible. beyond that....meh. | 01:33 |
creiht | hehe | 01:33 |
*** littleidea_ has quit IRC | 01:34 | |
notmyname | so how much would I be in trouble if I blogged about the undocumented features in swift on my personal blog? | 01:35 |
notmyname | the code's out there, so it's not really a secret any more | 01:35 |
*** lbieber has quit IRC | 01:38 | |
PiotrSikora | notmyname: you wouldn't be in trouble | 01:41 |
PiotrSikora | you've got my word :P | 01:41 |
notmyname | thanks :-) | 01:43 |
*** pdelgallego has joined #openstack | 01:43 | |
mtaylor | notmyname: agree | 01:44 |
mtaylor | first rule of running an open source department in a company - don't get people in trouble for blogging. anything. | 01:44 |
redbo | Just wait until gholt starts a blog. He has a way of making trouble. | 01:46 |
creiht | haha | 01:46 |
* mtaylor should introduce gholt and stewart then... | 01:47 | |
*** dendro-afk is now known as dendrobates | 02:06 | |
*** pdelgallego has quit IRC | 02:13 | |
*** justinsheehy has quit IRC | 02:14 | |
*** justinsheehy has joined #openstack | 02:19 | |
*** pvo has joined #openstack | 02:33 | |
*** lbieber has joined #openstack | 02:35 | |
*** binaryWarrior has quit IRC | 02:53 | |
*** copilot_ has quit IRC | 03:06 | |
*** copilot_ has joined #openstack | 03:07 | |
*** littleidea has joined #openstack | 03:13 | |
*** lbieber has quit IRC | 03:17 | |
*** jtdowney has quit IRC | 03:17 | |
*** pvo has quit IRC | 03:29 | |
*** binaryWarrior has joined #openstack | 03:37 | |
*** binaryWarrior has joined #openstack | 03:38 | |
dendrobates | This is new to rackspace, there will probably be some initial issues as they get used to transparency. | 03:44 |
dendrobates | I will back up anyone who needs it, as long as they did not blog about embargoed unfixed security issues. | 03:44 |
dendrobates | btw, is anyone at rackspace on vendorsec? | 03:45 |
*** binaryWarrior has quit IRC | 03:46 | |
mtaylor | dendrobates: not I | 03:52 |
mtaylor | dendrobates: did you see the new coverage reports? | 03:52 |
*** jdmaturen has joined #openstack | 03:54 | |
dendrobates | mtaylor: no | 03:57 |
mtaylor | dendrobates: http://hudson.openstack.org/job/swift-coverage/ | 03:59 |
mtaylor | dendrobates: nova tests break when coverage is turned on, so there's a little more work to do there | 03:59 |
*** metoikos has quit IRC | 04:00 | |
mtaylor | dendrobates: sorry ... http://hudson.openstack.org/job/swift-coverage/cobertura is a better link | 04:00 |
*** lbieber has joined #openstack | 04:01 | |
*** jdmaturen has quit IRC | 04:02 | |
*** silassewell has joined #openstack | 04:03 | |
dendrobates | mtaylor: awesome, thanks | 04:09 |
*** abecc has joined #openstack | 04:09 | |
dendrobates | I love hudson | 04:10 |
mtaylor | ++ | 04:10 |
termie | ... and success | 04:12 |
termie | http://github.com/termie/git-bzr-ng | 04:12 |
termie | also, the sun has risen | 04:13 |
creiht | interesting | 04:15 |
creiht | notmyname: -^ | 04:15 |
rbergeron | mmmmm, code coverage. delicious | 04:16 |
* creiht is embarrased by the couple of scripts that don't have any coverage and brings the whole percentage down | 04:17 | |
PiotrSikora | mtaylor: i never understood this approach, why do you create separate *-coverage, *-pylint, etc instead of enabling all those features in master build? | 04:20 |
redbo | creiht: psh... not me. Some of those uncovered files aren't important, or were being knocked out the last day before our deadline, and get covered by probe and func tests anyway. We can backfill unit tests as we work on it. | 04:22 |
creiht | indeed | 04:23 |
creiht | having a shiny web page that points it out makes it more evident though :) | 04:23 |
redbo | So long as we don't accept any code without it being tested, it'll go in the right direction. | 04:26 |
creiht | yup | 04:27 |
redbo | I compared our object server against nginx last night in the lab. | 04:34 |
creiht | oh yeah? | 04:35 |
redbo | nginx is almost twice as fast on a cached file | 04:35 |
redbo | but the difference is very small compared to the time it takes to pull an uncached file off disk | 04:36 |
*** jhalogen has joined #openstack | 04:36 | |
*** littleidea has quit IRC | 04:37 | |
PiotrSikora | redbo: told you! :D | 04:37 |
redbo | told me what? | 04:37 |
PiotrSikora | that you should use sendfile from real webserver | 04:37 |
PiotrSikora | yesterday, afaik ;) | 04:37 |
redbo | I know nginx is faster than a python webserver. It just doesn't matter. | 04:37 |
PiotrSikora | afair* even | 04:37 |
creiht | redbo: Was that cached in memory for nginx? | 04:39 |
PiotrSikora | it was cached by the operating system for any process, if i got this right :) | 04:39 |
redbo | I did cached file for both, and uncached file for both. Yeah, just talking page cache. | 04:40 |
creiht | k | 04:40 |
creiht | just curious | 04:40 |
PiotrSikora | redbo: but even if this doesn't matter performance wise | 04:40 |
creiht | PiotrSikora: I don't think we ever argued that nginx wouldn't be faster, but what we had been doing was fast enough for our use cases | 04:41 |
creiht | Plus the fact that we optimize for writes | 04:41 |
PiotrSikora | then you should take into account the fact that nginx won't copy that data from kernel memory into user space | 04:41 |
PiotrSikora | which means that virtually all memory will be used for kernel file cache | 04:41 |
PiotrSikora | and with python webserver only half will be used (more or so) | 04:42 |
PiotrSikora | so you'll get more cache-hits with sendfile | 04:42 |
redbo | yeah, like that 64k of memory we use per file is really putting a lot of pressure on our page cache. | 04:42 |
PiotrSikora | redbo: erm... right :) | 04:43 |
creiht | PiotrSikora: The other side is that in our use case, we don't get a lot of concurrent or subsequent reads to the same object | 04:43 |
creiht | the CDN integration has handled that | 04:43 |
PiotrSikora | creiht: true, i always forget that you're behind limelight | 04:43 |
PiotrSikora | :) | 04:43 |
redbo | Actually, right now we kill the page cache after serving a file. | 04:43 |
redbo | so we can cache more inodes and directoryies | 04:44 |
redbo | but that will likely change as we have different use cases | 04:44 |
PiotrSikora | by "we" do you mean your internal tools or is this enabled in swift by default? | 04:44 |
redbo | that's swift right now | 04:44 |
PiotrSikora | ah, that's probably not-so-good | 04:44 |
redbo | it's good for us | 04:45 |
redbo | but like I said, it'll probably change when people want to use swift for other stuff | 04:45 |
PiotrSikora | yeah... i know, i meant that its ofn by default | 04:45 |
PiotrSikora | on* | 04:45 |
*** littleidea_ has joined #openstack | 04:45 | |
PiotrSikora | just out of curiosity, does single request pull data from cludfiles to limelight and spreads it to all CDN nodes? | 04:46 |
redbo | so anyway... like 95% of the time it takes to download a file is disk access. nginx would just cut the other 5% in half. Not really worth another moving part for a 2.5% speedup. | 04:46 |
PiotrSikora | or does each request to unique CDN node pulls it straight from cloudfiles? | 04:47 |
*** silassewell has quit IRC | 04:47 | |
creiht | PiotrSikora: either way, it is easy enough to make into a config value if it becomes an issue | 04:47 |
redbo | PiotrSikora: there's a cache between them | 04:47 |
creiht | PiotrSikora: There is a server that the file gets fed to, that then feeds the edge networks | 04:47 |
creiht | yeah | 04:47 |
PiotrSikora | cool, thx :) | 04:47 |
creiht | PiotrSikora: btw, that piece is supposed to be open sourced as well | 04:48 |
redbo | We couldn't use nginx like that anyway. Maybe lighttpd would work, I haven't looked at it. | 04:48 |
PiotrSikora | that piece? i thought it was done on the limelight side of network? | 04:48 |
creiht | Well there is a piece that is needed for their integration | 04:49 |
redbo | Yeah, that cache is part of llnw. | 04:49 |
creiht | sorry if I made that sounds confusing... I was more talking about the llnw integration in general | 04:49 |
PiotrSikora | creiht: ah, ok :) | 04:49 |
PiotrSikora | redbo: just for the sake of argument, why you couldn't use nginx but lighttpd would work? | 04:50 |
redbo | well, I guess we could have different ports for uploading and downloading | 04:50 |
redbo | downloading goes through nginx and uploading goes directly to object server | 04:50 |
redbo | basically, you can't proxy big uploads through nginx | 04:51 |
PiotrSikora | redbo: that isn't true | 04:51 |
redbo | okay you can, but they get spooled to disk | 04:52 |
creiht | redbo: we can't in swift, since it spools requests locally | 04:52 |
PiotrSikora | redbo: you can just set buffers bigger than they are by default | 04:52 |
PiotrSikora | so it wouldn't spool so fast | 04:52 |
PiotrSikora | you can tune pretty much everything with nginx | 04:53 |
creiht | it still doesn't matter.. we don't want it to spool | 04:53 |
PiotrSikora | then you just set buffer big enough ;) | 04:53 |
redbo | sure, we just need 5GB of ram dedicated to nginx | 04:54 |
creiht | more than that if we want to support concurrent uploads :) | 04:54 |
redbo | actually you can turn off the file writing completely, but uploads just spill over into swap and it's even worse | 04:54 |
creiht | PiotrSikora: Don't get us wrong... we are big fans of nginx, it just wouldn't work for our use case | 04:55 |
redbo | Yeah, it's cool for most people. It's just bad as a proxy for big uploads. | 04:55 |
PiotrSikora | creiht: yeah, i get this, i was just wondering about the reasons :) | 04:55 |
PiotrSikora | redbo: well, IMHO nginx buffers uploads for very good reason | 04:56 |
redbo | right, so it can retry if the backend has an error | 04:56 |
PiotrSikora | yup | 04:56 |
redbo | but it bottlenecks the network when we tested using it as a load balancer | 04:56 |
*** bretpiatt has joined #openstack | 04:57 | |
PiotrSikora | single nginx? then yes, probably it was :) | 04:57 |
*** bretpiatt is now known as blpiatt | 04:58 | |
PiotrSikora | JFYI: i'm not trying to convince you to nginx or pretend that i know your use case better than you | 04:59 |
PiotrSikora | i just like the discussion :) | 04:59 |
*** blpiatt has quit IRC | 05:01 | |
redbo | We evaluated nginx pretty closely for our load balancing, and it just couldn't get anywhere near 10gbps for uploads. We'd have to have a crazy disk setup. | 05:01 |
*** blpiatt has joined #openstack | 05:01 | |
PiotrSikora | or huge amount of RAM ;) | 05:02 |
redbo | How does nginx deal with the fact that sendfile() can block? Does it do some readahead() magic, or does it just put it in a separate thread? | 05:06 |
PiotrSikora | AFAIK it uses asynchronous interface to sendifle, so it cannot block | 05:07 |
PiotrSikora | let me double-check that | 05:08 |
redbo | or does it mess with the horrible aio api | 05:08 |
PiotrSikora | why horrible? | 05:10 |
PiotrSikora | under FreeBSD is same syscall | 05:10 |
PiotrSikora | it just return EAGAIN | 05:10 |
PiotrSikora | returns* | 05:10 |
redbo | well, I know you can do sendfile nonblocking for the socket, but I'm pretty sure it still blocks on disk access. | 05:11 |
PiotrSikora | erm... it shouldn't | 05:12 |
PiotrSikora | that would be just stupid implementation IMHO | 05:12 |
*** clayg has quit IRC | 05:12 | |
*** clayg- has joined #openstack | 05:13 | |
PiotrSikora | SF_NODISKIO. This flag causes any sendfile() call which would | 05:13 |
PiotrSikora | block on disk I/O to instead return EBUSY. | 05:13 |
PiotrSikora | (from the manpage) | 05:14 |
PiotrSikora | but thats tunable only in FreeBSD | 05:14 |
PiotrSikora | to be honest i have no idea what it would behave under Linux | 05:15 |
PiotrSikora | but answering your original question, nginx uses this flag, so it wouldn't block | 05:15 |
redbo | I can't find any documentation on it, but I've read stuff like http://brad.livejournal.com/2228488.html that says sendfile will block. | 05:15 |
pquerna | on freebsd. | 05:15 |
PiotrSikora | at least not on FreeBSD | 05:15 |
PiotrSikora | redbo: please note that this post is 4 years old | 05:15 |
redbo | true | 05:16 |
PiotrSikora | a lot changed since then :) | 05:16 |
redbo | but | 05:16 |
pquerna | not in regards to sendilfe really. | 05:16 |
redbo | it's the only thing I have to go on | 05:16 |
creiht | AIO on linux has very poor documentation, and it is very difficult to find many sources that say weather or not it even works very well | 05:17 |
creiht | most references that I found when first researching it, indicated that it could be very problematic | 05:17 |
creiht | (mostly related to AIO and file access) | 05:17 |
cw | oracle uses it | 05:17 |
redbo | The thing I hate about aio, there's no way to use it in like a select or epoll event loop in a reasonable way. | 05:17 |
cw | aio interfaces are unpleasant to say the least, but it works fairly well | 05:18 |
PiotrSikora | creiht: you could always switch to FreeBSD + ZFS :D | 05:18 |
redbo | don't tempt him | 05:18 |
cw | redbo: check out mason's aiocp for something simplish | 05:18 |
creiht | hah | 05:18 |
pquerna | actually, don't. | 05:18 |
cw | or the fio code | 05:18 |
pquerna | well, sorry, actually, do | 05:18 |
PiotrSikora | redbo: erm... what? AIO is exactly to be used with event loops | 05:18 |
pquerna | freebsd + zfs is awesome, but freebsd + zfs + sendfile is actually kinda broken right now | 05:18 |
pquerna | for www.apache.org we had to turn off sendfile | 05:19 |
creiht | I went down the path once of setting up a freebsd vm with zfs to test, but just didn't have time to keep messing with it | 05:19 |
PiotrSikora | pquerna: ouch, i didn't know that... i assume they didnt fix it yet? | 05:20 |
pquerna | http://mail-archives.apache.org/mod_mbox/httpd-dev/201007.mbox/%3CAANLkTil_WxKF6FqwNN9BksS1xVzNYBPdn0vLdUa2alSq@mail.gmail.com%3E | 05:20 |
*** clayg- is now known as clayg | 05:20 | |
*** blpiatt has quit IRC | 05:21 | |
PiotrSikora | ah, its there... kern/141305 | 05:21 |
creiht | and then when you add that we are doing this with python, it adds an extra layer of complexity with using aio | 05:22 |
PiotrSikora | pquerna: thanks for the detailed info :) | 05:22 |
PiotrSikora | pquerna: you don't know whatever sendfile works reliably with (open)solaris's zfs? | 05:23 |
*** blpiatt has joined #openstack | 05:23 | |
pquerna | PiotrSikora: it worked on solaris | 05:23 |
creiht | http://code.google.com/p/twisted-linux-aio/wiki/WorkInProgress | 05:23 |
creiht | Has a lot of links similar to some of the issues that I found | 05:23 |
pquerna | PiotrSikora: until 2 weeks ago apache.org was on solaris 10, and it had sendfile on w/ a zfs array | 05:23 |
redbo | man, why do I always feel like I'm trailblazing when I want to make asynchronous io work? | 05:24 |
pquerna | two weeks ago we switched to fbsd8.1 + zfs, and had to turn off sendfile | 05:24 |
creiht | hehe | 05:24 |
PiotrSikora | pquerna: ah... that's disapointing :( | 05:24 |
PiotrSikora | disappointing* | 05:24 |
creiht | I really hope someone can figure out how to make async file io work well on linux and then also with python :) | 05:24 |
pquerna | its called threading :x | 05:25 |
pquerna | (a small number of threads) | 05:25 |
*** clayg has quit IRC | 05:25 | |
pquerna | (not 1:1 with clients, but.. anyways) | 05:25 |
redbo | he said it had to work with python | 05:25 |
pquerna | hahaha | 05:25 |
creiht | hah | 05:25 |
PiotrSikora | redbo, creiht: well, but if you would use real webserver, then you wouldn't have to do AIO in python :P | 05:25 |
*** clayg has joined #openstack | 05:26 | |
PiotrSikora | but going back to python, you could probably could use stackless? | 05:26 |
creiht | *only if we run on freebsd | 05:26 |
redbo | actually my test of nginx vs our server was bad. nevermind the results, I'll do a better test later. | 05:26 |
PiotrSikora | with 1:1 clients:greenlets ratio | 05:26 |
*** lbieber has quit IRC | 05:26 | |
creiht | haha | 05:26 |
creiht | PiotrSikora: eventlet/greenlet gives us most of the advantages of stackless in regular python | 05:27 |
*** clayg has quit IRC | 05:27 | |
creiht | There is a huge advantage for being able to use the standard python interpreter | 05:27 |
PiotrSikora | erm... so why do you need AIO? | 05:27 |
creiht | We don't *need* it, it just could have been nice | 05:28 |
redbo | disk io can block the whole process | 05:28 |
*** clayg- has joined #openstack | 05:28 | |
creiht | yeah | 05:28 |
PiotrSikora | redbo: yeah... i just checked eventlet/greenlet | 05:28 |
PiotrSikora | i thought they worked like stackless (lots of microthreads) | 05:28 |
pquerna | eventlet/greenlet don't farm out to a thread pool for file io? | 05:29 |
creiht | stackless doesn't do anything for you in the way of async file io | 05:29 |
pquerna | I had assumed they did | 05:29 |
PiotrSikora | so even if they would block it wouldn't matter | 05:29 |
creiht | pquerna: that is left up to the dev | 05:29 |
pquerna | (like how node.js interacts with libeio) | 05:29 |
PiotrSikora | creiht: exactly, because you dont need it | 05:29 |
*** clayg- is now known as clayg | 05:29 | |
creiht | All of the above said, we've managed to work around it | 05:30 |
PiotrSikora | you can just create new microthread for each client and dont worry about it blocking on i/o | 05:30 |
PiotrSikora | creiht: yeah... i guess we went a bit off-topic from the nginx discussion :) | 05:30 |
creiht | PiotrSikora: If you do any disk file operation, it will block the whole process until that file operation is complete | 05:31 |
creiht | hehe | 05:31 |
redbo | disk io will block a whole stackless process as well. greenlet is literally the stackless part of stackless. | 05:31 |
PiotrSikora | creiht: whole process or whole thread? | 05:31 |
redbo | well, it's actually a whole bunch more stacks. but whatever. | 05:31 |
creiht | PiotrSikora: there are no real threads in stackless, it is all cooperative, so it blocks the whole process (and in turn all microthreads you have running) | 05:32 |
creiht | In the end, chunking your reads/writes works out pretty well | 05:32 |
PiotrSikora | that's disappointing | 05:32 |
creiht | PiotrSikora: That's why async file io would be nice if it worked, because then it wouldn't block | 05:34 |
PiotrSikora | i know :) | 05:35 |
creiht | Even with all of that said, we can pretty easy saturate the disk and network io, so it isn't a huge issue | 05:35 |
cw | creiht: how much concurrnent 'aio' do you want in fight? | 05:35 |
redbo | I guess what we'd probably have to do is read in another thread. Or at least posix_fadvise/readahead in another thread. | 05:36 |
cw | a few hundred threads would probably keep most IO subsystems busy assuming 24 spindles | 05:36 |
cw | and threads aren't that expensive (though they are pretty ugly) | 05:36 |
creiht | cw: well you can't even get that far, because (at least on gig-E) you run out of network real qucik | 05:36 |
cw | creiht: assuming a LOT of small very seeky IO you could fave a few 1000s IOs in flight | 05:37 |
redbo | yeah, but someday I'd like downloads of uncached files to not interfere with each other if they're on the same process. | 05:37 |
cw | creiht: small ones would git in GbE | 05:37 |
creiht | redbo: indeed | 05:37 |
creiht | cw: true | 05:37 |
creiht | cw: Things actually work well enough as is, I would still like to test AIO if I could figure out how it works, just to see if it would allow for us to handle more concurrency | 05:38 |
*** jhalogen has quit IRC | 05:38 | |
cw | creiht: it might be fun to play with, but my gut feeling is that it's a premature optimization right now | 05:38 |
*** clayg has quit IRC | 05:38 | |
cw | creiht: maybe 48-96 disk systems with 10GbE ports and you'll be closer to a load where it might help | 05:39 |
creiht | cw: yeah that is what we concluded back when I initially looked into it | 05:39 |
cw | syscalls are cheap, and context switches also | 05:39 |
cw | cpus are fast | 05:39 |
cw | (nowdays) | 05:39 |
*** clayg- has joined #openstack | 05:39 | |
redbo | well, I do want to work on our CPU usage, but file downloads are far from the biggest culprit. | 05:40 |
cw | wrt to zones, the recommended number is 5 not less | 05:40 |
creiht | And there are a lot of features to add that are a lot more important :) | 05:40 |
cw | is there any reason to think 3 zones would actually cause issues for a small deployment? | 05:40 |
creiht | cw: That is true for adequately handling failures | 05:41 |
redbo | It'd be sweet if you could strap an atom processor to 8 drives and have that be a node. | 05:41 |
creiht | 3 is the absolute minimum | 05:41 |
cw | right, but 3 != 5 | 05:41 |
creiht | cw: deployment for testing, or for use? | 05:41 |
redbo | cw: I think the 5 node thing is an old superstition. | 05:41 |
redbo | er 5 zone | 05:41 |
cw | use | 05:41 |
cw | initial deployment of three phyiscal jbod style machines for very small deployments | 05:42 |
cw | creiht: are you going to the meetup next week? | 05:42 |
creiht | meetup? | 05:42 |
creiht | cw: As the code is right now, handoffs are sent to other zones | 05:42 |
creiht | so if you only have 3 zones, then it can't handoff | 05:43 |
redbo | oh, good point | 05:43 |
wreese | redbo: I think we have logic that tries 3 zones, and picks a 4th on failure (not pick a 3rd that's not already chosen) | 05:43 |
creiht | Everything still works, but the system can't work around failures | 05:43 |
cw | creiht: handoff? | 05:43 |
creiht | cw: say a drive has failed, and you want to write to it | 05:43 |
creiht | the system detects the failure, and writes to another node instead, (in another zone) | 05:44 |
creiht | When the drive is replaced, that data gets replicated back | 05:44 |
cw | creiht: and this is orthogonal to having 3+ copies? | 05:44 |
creiht | It provides greater availability | 05:44 |
redbo | did we ever fix that problem where handoffs were clumping on even-numbered nodes? that seems like something we could fix. | 05:44 |
creiht | redbo: yeah we need to look closer into that | 05:44 |
wreese | yeah. gholt fixed that | 05:44 |
creiht | oh cool | 05:44 |
* creiht didn't know that | 05:45 | |
creiht | cw: We try as much as possible to always have 3 copies (even in the event of failure) | 05:45 |
wreese | well, we should ask him to be sure, but I'm almost positive he fixed it :P | 05:45 |
creiht | for example in the above example, at least 2 writes have to complete successfully for the system to return success | 05:45 |
redbo | I'd like in the future for the handoff node for a partition to become a real node for it if one of the others fails. Then the stuff is already where we would have moved it. | 05:46 |
cw | creiht: meetup i think is more focused on Nova for now ... next week | 05:46 |
creiht | cw: where is the meetup at? | 05:46 |
cw | cloudkick in san francisco | 05:46 |
creiht | ahh... Then I guess probably not | 05:46 |
cw | lee bieber might be there | 05:46 |
creiht | redbo: that is an interesting idea | 05:47 |
cw | creiht: how do you know a write has completed? | 05:47 |
cw | it could be cached ... even w/ O_SYNC | 05:47 |
wreese | yeah. it's non-deterministic atm | 05:47 |
cw | so there is a small racy | 05:47 |
cw | err, race | 05:47 |
creiht | The write is proxied through the proxy server, and streamed to 3 servers | 05:47 |
creiht | 3 object servers | 05:47 |
redbo | we do fsync. That's the best we can do :) | 05:47 |
creiht | those object servers return to the proxy success or failure | 05:48 |
creiht | hehe | 05:48 |
*** abecc_ has joined #openstack | 05:48 | |
cw | ah, fsync on xfs will work quite well actually | 05:48 |
cw | even on cheapo disks | 05:48 |
*** jdmaturen has joined #openstack | 05:48 | |
creiht | cw: yeah... we go out of our way to make sure things get to disk | 05:48 |
wreese | and we ensure disk caches are disabled | 05:48 |
cw | creiht: you might wan to play with delay logging at some point ... if you have metadata heavy loads (though fsync heavy it won't help a lot) | 05:48 |
*** abecc has quit IRC | 05:48 | |
cw | delayed logging | 05:48 |
cw | god im retarded | 05:48 |
*** abecc_ is now known as abecc | 05:49 | |
* creiht googles | 05:49 | |
cw | i have beer ... but it's near emtpy and the kids are in bed | 05:49 |
cw | how can i possibly get another one then | 05:49 |
redbo | also, all 3 nodes would have to fail for the file to be lost, presumably | 05:49 |
creiht | looks like a fairly new feature? | 05:49 |
cw | would it be wrong to throw things at the wall and wake my eldest so she can get me a beer? | 05:49 |
cw | surely not | 05:49 |
cw | wreese: disabling the disk caches shouldn't be necessary | 05:50 |
*** jdmaturen has quit IRC | 05:50 | |
cw | wreese: i would advise against it in some cases ... if you fsync you get things to the oxide anyhow | 05:50 |
cw | w/o caches you're slowing other writes down between the critical barriers | 05:50 |
redbo | I'm intrigued by ext4 with write ordering instead of fsyncing (for consistency), and relying on multiple nodes for durability. | 05:50 |
PiotrSikora | hm... so you're streaming in the real-time while its being uploaded? | 05:50 |
PiotrSikora | instead of post-upload spreading? | 05:50 |
creiht | PiotrSikora: yes | 05:50 |
PiotrSikora | that's really, really cool | 05:51 |
creiht | :) | 05:51 |
wreese | many drives/controllers noop the barriers and fsync just gets it out of the kernel | 05:51 |
PiotrSikora | i don't think i saw "distributed fs" that did this | 05:51 |
cw | you use zeus in front of the proxies right? to get the needed perf? | 05:51 |
PiotrSikora | creiht: tricky question, do you also compare checksums post-upload? | 05:51 |
PiotrSikora | to verify that each node got the same file and it isnt corrupted? | 05:52 |
creiht | currently we use zeus for load balancing to get the most possible out of SSL and 10g | 05:52 |
redbo | PiotrSikora: we don't re-read the file, but we check the checksums that each node received and make sure they match. | 05:52 |
cw | redbo: do you ever in production see checksum failures and invalidate? | 05:52 |
PiotrSikora | redbo: but where do you get the checksum from? | 05:53 |
cw | doing some math on that tells me they should be very rare even on cheap hardware | 05:53 |
redbo | as each node is receiving the data, it adds it to an md5, then each node sends the checksum back with its response, and the proxy makes sure they all match | 05:53 |
PiotrSikora | cw: amazon had this once, one of their load balancers was introducing errors | 05:54 |
creiht | PiotrSikora: The client can also send the expected checksum in the header, and it will be checked against that | 05:54 |
redbo | we also have a process that randomly checks files on disk to be sure their checksum when uploaded matches what's on disk | 05:54 |
wreese | cw: we have auditors that scan for that, and I don't think they've come across any bit rot yet | 05:54 |
PiotrSikora | redbo: ah, ok... seems reasonable enough :) | 05:55 |
creiht | cw: do you have a link for xfs delayed logging? | 05:55 |
cw | wreese: you've injected errors to verify the auditors DTRT right? | 05:55 |
cw | creiht: it's in recent mainline | 05:55 |
cw | i think dave got it into '33 | 05:55 |
cw | for stock ubuntu won't have it | 05:55 |
creiht | cw: k | 05:55 |
cw | or do you just want docs? | 05:56 |
creiht | docs would be cool | 05:56 |
creiht | :) | 05:56 |
cw | Documentation/filesystems/xfs-delayed-logging-design.txt | 05:56 |
cw | creiht: do you know off hand if many of your stored objects span multiple extents? | 05:57 |
cw | on disk i mean ... ie. delalloc doesn't work so well if htey are written out slowly and there is memory pressure | 05:57 |
cw | so a linear read of larger objects will cause seeking | 05:58 |
redbo | very few should have multiple extents | 05:58 |
wreese | cw: I believe we did, but gholt would know for sure. the code reads the file and computes the checksum, then compares to the checksum in the xattr data. if they differ, the file is quarantined and replicated back in place | 05:58 |
*** abecc_ has joined #openstack | 05:58 | |
*** abecc has quit IRC | 05:58 | |
*** abecc_ is now known as abecc | 05:58 | |
redbo | if the user gives us an upload size, we fallocate the space first | 05:58 |
cw | redbo: there is some EOF preallocation that's trimmed on close ... but even so slow writes would tend to cause fragmentation | 05:58 |
creiht | cw: We preallocate the filesize if we know beforehand | 05:58 |
creiht | yeah what redbo said | 05:58 |
cw | creiht: using the ioctl? | 05:58 |
cw | sorry, i should just check the code :-) | 05:58 |
cw | we have swift up in the lab, i'm being lazy | 05:59 |
creiht | hehe | 05:59 |
redbo | if only we could fallocate directories, those are worse | 05:59 |
cw | fragmentation there? | 06:00 |
redbo | yeah | 06:00 |
cw | you can tweak the allocator sometimes a little actually | 06:00 |
redbo | listdirs take forever | 06:00 |
cw | redbo: how many entries? | 06:00 |
cw | xfs readdir has pretty good readahead unlike everything else | 06:00 |
redbo | um... hundreds to a few thousand | 06:00 |
cw | that's nothing | 06:00 |
cw | so slow growth here right? | 06:01 |
creiht | yeah | 06:01 |
creiht | that is the issue | 06:01 |
cw | so they are fragmenting to 20+ extents? | 06:01 |
redbo | you say that. but it's normal to have a listdir that takes 10 or 20 seconds. | 06:01 |
*** clayg- is now known as clayg | 06:01 | |
creiht | or even up to a minute in some cases | 06:01 |
cw | redbo: xfs_bmap -vp and see what it looks like | 06:01 |
cw | that sounds grossly excessive | 06:01 |
cw | even with a LOT of other IO going on | 06:02 |
redbo | let's see... "time ls" -> real 0m18.162s | 06:04 |
cw | how many extents? | 06:04 |
redbo | only 10 extents | 06:04 |
cw | 5500 entries here and only 22 extents | 06:04 |
redbo | 2696 entries | 06:04 |
cw | ls is doing a stat | 06:04 |
creiht | cw: I think the problem is with the extents being all over the disk | 06:04 |
cw | for every entry i bet | 06:04 |
redbo | oh, true | 06:04 |
cw | ls is retarded | 06:04 |
cw | /bin/ls | 06:04 |
creiht | yeah | 06:04 |
cw | no options | 06:04 |
cw | it's not the dir reading, it's the seeking from ls | 06:04 |
cw | you can optimize that a lot if you care | 06:05 |
creiht | right... but for our purposes, we need the stat info | 06:05 |
cw | what do you need from it? | 06:05 |
creiht | listing just gives the name right? | 06:05 |
cw | name, inode and type | 06:06 |
redbo | yeah. well, we were using os.walk a lot, which does stat everything | 06:06 |
creiht | for example the biggest issue was rsync crawling the partitions to make their changelists | 06:06 |
cw | though type isn't in the old ABI | 06:06 |
creiht | cw: btw, this is why we have hashlists for the partitions to help get around this issue | 06:06 |
cw | creiht: ok ... i'm very intersted in that | 06:07 |
cw | the issue is rsync doing the read + stat kills perf? | 06:07 |
creiht | cw: and I would be interested in hearing how to make the stat calls better | 06:07 |
creiht | It made replication take too long | 06:07 |
redbo | too bad xfs doesn't support the d_type field in dirents. | 06:07 |
cw | it does | 06:07 |
cw | man getdents | 06:07 |
cw | iirc | 06:08 |
redbo | hrm... when I looked up the docs a while back, it said it didn't. I'll check it out. | 06:08 |
redbo | that would make file system walking considerably faster. | 06:09 |
cw | well, it depends what you want | 06:09 |
cw | python also doesn't have all the interfaces exposed | 06:09 |
redbo | to know if something's a directory or file, mostly | 06:09 |
cw | ok, that it has | 06:09 |
cw | is this the python code that needs this or rsync? | 06:09 |
cw | rsync is horrible for lots of little files | 06:09 |
cw | i would just say "don't do that" | 06:09 |
creiht | hah | 06:09 |
cw | but ... if you really must use it, well, there are things you can do to make it better | 06:10 |
redbo | I don't know anything about rsync internals, but we use os.walk for stuff, and that gets pretty slow with all the stat calls. | 06:10 |
creiht | Well now we use the hashes to determine if something has changed, and if something has changed, then we do run rsync | 06:10 |
cw | i have bulkstat working at over 100,000 inodes/s a while back | 06:10 |
cw | actually, i think it peaked at over 260,000 inodes/s cold-cache and over 1M hot-cache | 06:10 |
cw | os.walk isn't very clean internally last i did an strace of it | 06:10 |
redbo | no, it's horrible, but it's simple :) | 06:11 |
cw | yeah | 06:11 |
cw | the python interface to it is very nice | 06:11 |
cw | i've broken down and used it in a lot of place, i even have: | 06:11 |
creiht | some of the code we manually do listdirs, and then walk manually | 06:11 |
cw | # don't cross mountpoints | 06:11 |
cw | dirs[:] = [d for d in dirs if not os.path.ismount(os.path.join(root, d))] | 06:11 |
cw | which is nice to write but nasty when you look at what happens under the hood | 06:12 |
redbo | yeah, but if we could have an interface that was as fast as listdirs but has the interface of os.walk, we'd be sitting pretty | 06:12 |
redbo | yeah, ismount stats the file and its parent to see if they have the same device id | 06:12 |
cw | os.walk also doesn't iterate the way i would expect it too | 06:12 |
redbo | there's an option to make it sane, I think | 06:13 |
cw | in so far as it reads a LOT of entries then next()... gives details | 06:13 |
cw | it's not a fine grained as i expected | 06:13 |
cw | though a lot of this is python 2.4 & 2.5 experience | 06:13 |
PiotrSikora | well, if python wrappers are so unoptimized | 06:14 |
PiotrSikora | cant you embed C syscalls within python? | 06:15 |
cw | redbo: don't you know roughly what to sync when you havr to sync it? | 06:15 |
cw | rather than making rsync work it out? | 06:15 |
redbo | yeah, we're doing okay now | 06:15 |
redbo | we use hashes of directory contents to decide better what needs to actually rsync at a pretty granularity that gives a pretty good balance | 06:16 |
redbo | er.. not pretty granularity | 06:16 |
redbo | but there's still parts of our code where execution time is ruled primarily by listdirs. So it's something to go after. | 06:17 |
*** mattf has quit IRC | 06:18 | |
cw | the reason to use rsync is because it's prety robust and easy? | 06:28 |
cw | something less to embed in the code? | 06:29 |
creiht | It works | 06:29 |
creiht | :) | 06:29 |
creiht | We tried to reuse tools where we could (until we found them to not work at our scale) | 06:29 |
cw | it requires some config you could avoid though | 06:30 |
*** jdmaturen has joined #openstack | 06:30 | |
redbo | avoid config by rewriting rsync? | 06:31 |
creiht | :) | 06:31 |
cw | heh, that came out wrong | 06:31 |
*** jdmaturen has quit IRC | 06:32 | |
*** mcgurrin_ has joined #openstack | 06:34 | |
PiotrSikora | fast question (sorry, i'm being lazy) but does the object server/proxy support byte-ranges and/or gzipping? | 06:35 |
redbo | ranges, yes. gzipping, no. | 06:35 |
*** mcgurrin has quit IRC | 06:35 | |
*** mcgurrin_ is now known as mcgurrin | 06:35 | |
PiotrSikora | redbo: thx | 06:36 |
PiotrSikora | i had a feeling that would be the answer | 06:36 |
PiotrSikora | gzipping probably isn't best business idea when you charge for transfer ;) | 06:36 |
redbo | right, letting users use more CPU to pay less wasn't the best idea | 06:37 |
redbo | we actually had it in the code at some point, but took it out | 06:37 |
PiotrSikora | well, you aren't CPU-bound so this probably wasn't big problem | 06:38 |
redbo | I'm sure we could put it back in as optional, there's not much to it. | 06:38 |
redbo | touche | 06:38 |
PiotrSikora | i was just going to ask you about doing that :) | 06:38 |
*** mattf has joined #openstack | 06:39 | |
PiotrSikora | oohhh... i just remembered | 06:40 |
redbo | Also we'll have to do like vary: accept-encoding so caches will know to grab different versions for people who accept or don't accept gzip and test all that. | 06:40 |
PiotrSikora | one of my friend is working for a portal that uses cloudfiles | 06:40 |
PiotrSikora | and he told me that little % of uploaded files reguraly disappear | 06:41 |
PiotrSikora | do you know about such issue? | 06:41 |
creiht | PiotrSikora: how is he uploading them? | 06:42 |
PiotrSikora | creiht: to be honest? i don't know | 06:42 |
PiotrSikora | he just mentioned this over a drink | 06:42 |
PiotrSikora | that they have to re-upload few files almost everyday | 06:42 |
PiotrSikora | i'll get details out of him when he wakes up | 06:43 |
redbo | I haven't seen that. If he could give us an example file that disappeared, we could find out a lot. | 06:43 |
redbo | well, account and filename | 06:43 |
creiht | If he was using a third party tool, then lots of things can go wrong there :) | 06:44 |
creiht | If he was uploading directly using an api, then it would also be interesting to hear what return codes he sees when he uploads | 06:44 |
*** clayg has quit IRC | 06:45 | |
PiotrSikora | i'll ask him about details | 06:46 |
*** ewan_ has joined #openstack | 06:46 | |
*** clayg- has joined #openstack | 06:46 | |
*** BartVB has joined #openstack | 06:50 | |
*** blpiatt has quit IRC | 07:00 | |
*** allsystemsarego has joined #openstack | 07:11 | |
*** clayg- is now known as clayg | 07:12 | |
*** ewan_ has quit IRC | 07:25 | |
*** ewan_ has joined #openstack | 07:29 | |
*** gasbakid has joined #openstack | 07:33 | |
*** jhalogen has joined #openstack | 07:41 | |
*** ewan_ has quit IRC | 07:51 | |
PiotrSikora | one more tricky question, what happens when 1 or 2 (or n < total number of replicas) nodes go offline during file upload? | 07:52 |
PiotrSikora | is this handled gracefully by the proxy? and is file replicated right away to another node after upload completes? | 07:52 |
*** jhalogen has quit IRC | 07:58 | |
*** gasbakid has quit IRC | 08:08 | |
*** littleidea_ has quit IRC | 08:11 | |
*** abecc has quit IRC | 08:17 | |
*** gasbakid has joined #openstack | 08:27 | |
*** avi has quit IRC | 08:35 | |
redbo | if a majority of the uploads fail, it'll fail for the user. If one fails, it'll have to wait for replication to be put back in place. | 08:52 |
redbo | well, it'll get replicated to a fourth node, then when the drive is replaced or removed from the ring, the file will go where it's supposed to. | 08:57 |
*** _anm has quit IRC | 08:59 | |
PiotrSikora | redbo: thx :) | 09:15 |
*** metoikos has joined #openstack | 10:00 | |
PiotrSikora | redbo, creiht: i just talked with my friend | 10:09 |
PiotrSikora | they didn't log those lost files | 10:10 |
PiotrSikora | but he told me that he'll enable process in their infrastrucutre | 10:10 |
PiotrSikora | so this will be logged next time it will happen | 10:10 |
*** ewan_ has joined #openstack | 10:11 | |
*** justinsheehy has quit IRC | 10:27 | |
PiotrSikora | enable logging* | 10:28 |
*** justinsheehy has joined #openstack | 10:32 | |
*** alexh__ has quit IRC | 11:09 | |
*** jtdowney has joined #openstack | 11:14 | |
*** ewan_ has quit IRC | 11:53 | |
*** jk0 has quit IRC | 12:22 | |
*** antonym has quit IRC | 12:22 | |
*** jk0 has joined #openstack | 12:27 | |
*** antonym has joined #openstack | 12:27 | |
*** ChanServ sets mode: +v jk0 | 12:27 | |
*** blpiatt has joined #openstack | 12:27 | |
*** antonym is now known as Guest1494 | 12:27 | |
*** StephanL has joined #openstack | 12:31 | |
*** tjyang has joined #openstack | 12:34 | |
*** justinsheehy has quit IRC | 12:49 | |
*** StephanL has quit IRC | 12:52 | |
*** justinsheehy has joined #openstack | 12:54 | |
*** tjyang has quit IRC | 13:28 | |
*** justinsheehy has quit IRC | 13:33 | |
*** pvo has joined #openstack | 13:34 | |
*** pvo has joined #openstack | 13:34 | |
*** justinsheehy has joined #openstack | 13:38 | |
uvirtbot | New bug: #609749 in nova "nova.process singleton isn't working, causing "Someone released me too many times: too many tokens!" assertion" [Undecided,In progress] https://launchpad.net/bugs/609749 | 13:46 |
*** silassewell has joined #openstack | 14:04 | |
*** _anm has joined #openstack | 14:12 | |
uvirtbot | New bug: #609757 in nova "nova-compute does not accept images_path on the command line" [Undecided,Fix committed] https://launchpad.net/bugs/609757 | 14:36 |
*** blpiatt has quit IRC | 14:42 | |
pvo | Ewan is cranking stuff out on a Sunday afternoon. : ) | 14:44 |
*** pvo has quit IRC | 14:45 | |
*** littleidea_ has joined #openstack | 14:47 | |
*** sandy has joined #openstack | 14:55 | |
*** sandy is now known as Guest29615 | 14:55 | |
allsystemsarego | has anyone pitted the nova/swift duo against hadoop/mapreduce feature-wise? | 14:56 |
*** Guest29615 has quit IRC | 15:02 | |
*** Guest29615 has joined #openstack | 15:02 | |
*** Guest1494 is now known as antonym | 15:09 | |
*** antonym has quit IRC | 15:09 | |
*** antonym has joined #openstack | 15:09 | |
*** ChanServ sets mode: +v antonym | 15:09 | |
*** joker[avd] has joined #openstack | 15:29 | |
_anm | allsystemsarego: that sounds like an apples/oranges comparison to me | 15:54 |
_anm | hadoop/mapreduce is a distributed processing and storage system, whereas nova/swift are cloud computing a dnblock storage platforms, mot akin to amazon ec2 and S3 | 15:56 |
*** tjyang has joined #openstack | 16:00 | |
*** silassewell has quit IRC | 16:01 | |
*** binaryWarrior has joined #openstack | 16:05 | |
*** rajijoom has joined #openstack | 16:15 | |
*** silassewell has joined #openstack | 16:16 | |
*** mattf has quit IRC | 16:16 | |
*** poli has quit IRC | 16:26 | |
*** Guest29615 has left #openstack | 16:29 | |
*** Guest29615 has joined #openstack | 16:29 | |
*** Guest29615 has left #openstack | 16:34 | |
*** Guest29615 has joined #openstack | 16:34 | |
uvirtbot | New bug: #609791 in nova "AMQPChannelException when running instance" [Undecided,In progress] https://launchpad.net/bugs/609791 | 16:36 |
*** rajijoom has quit IRC | 16:41 | |
*** rajijoom has joined #openstack | 16:41 | |
*** binaryWarrior has quit IRC | 16:44 | |
*** tjyang has quit IRC | 16:47 | |
uvirtbot | New bug: #609792 in nova "Hardcoded exchange name should be replaced with flags.control_exchange" [Undecided,Fix committed] https://launchpad.net/bugs/609792 | 16:50 |
*** blpiatt has joined #openstack | 17:01 | |
*** dendrobates is now known as dendro-afk | 17:09 | |
*** dendro-afk is now known as dendrobates | 17:16 | |
*** rajijoom has quit IRC | 17:17 | |
*** lbieber has joined #openstack | 17:30 | |
*** pvo has joined #openstack | 17:37 | |
*** pvo has joined #openstack | 17:37 | |
*** ChanServ sets mode: +v pvo | 17:37 | |
*** lbieber has quit IRC | 17:45 | |
*** rajijoom has joined #openstack | 17:46 | |
*** ChanServ sets mode: -s | 17:51 | |
*** dendrobates is now known as dendro-afk | 17:54 | |
*** pvo has quit IRC | 17:57 | |
*** algorist has joined #openstack | 18:06 | |
*** algorist has quit IRC | 18:09 | |
uvirtbot | New bug: #609816 in nova "euca_deregister fails with error and objectstore logs "exceptions.AttributeError: ImageResource instance has no attribute 'get_argument'"" [Undecided,In progress] https://launchpad.net/bugs/609816 | 18:11 |
*** dendro-afk is now known as dendrobates | 18:18 | |
*** blpiatt has quit IRC | 18:23 | |
*** jdmaturen has joined #openstack | 18:23 | |
*** jdmaturen has quit IRC | 18:30 | |
*** rajijoom has quit IRC | 18:39 | |
*** z0 has quit IRC | 19:00 | |
*** jonesy has quit IRC | 19:17 | |
*** lbieber has joined #openstack | 19:24 | |
*** jhalogen has joined #openstack | 19:26 | |
*** algorist has joined #openstack | 19:28 | |
*** algorist has quit IRC | 19:32 | |
*** algorist has joined #openstack | 19:48 | |
*** littleidea_ has quit IRC | 19:49 | |
*** littleidea_ has joined #openstack | 19:53 | |
*** lbieber has quit IRC | 19:54 | |
*** anm` has joined #openstack | 19:55 | |
*** _anm has quit IRC | 19:56 | |
*** anm` is now known as _anm | 19:56 | |
*** Guest29615 has left #openstack | 19:58 | |
*** littleidea_ has quit IRC | 20:08 | |
*** littleidea has joined #openstack | 20:10 | |
*** algorist has quit IRC | 20:32 | |
*** jhalogen has quit IRC | 20:40 | |
*** kirkland has quit IRC | 20:51 | |
*** kirkland has joined #openstack | 20:54 | |
*** blpiatt has joined #openstack | 21:00 | |
*** kirkland has quit IRC | 21:20 | |
*** kirkland has joined #openstack | 21:21 | |
*** z0 has joined #openstack | 21:33 | |
*** mattf has joined #openstack | 21:43 | |
*** blpiatt is now known as blp-dinner | 21:59 | |
*** allsystemsarego has quit IRC | 22:07 | |
*** lbieber has joined #openstack | 22:14 | |
*** lbieber has quit IRC | 22:20 | |
*** blp-dinner has quit IRC | 22:44 | |
*** miclorb has joined #openstack | 22:45 | |
*** jhalogen has joined #openstack | 22:56 | |
*** algorist has joined #openstack | 22:56 | |
*** lbieber has joined #openstack | 23:05 | |
*** pvo has joined #openstack | 23:13 | |
*** pvo has joined #openstack | 23:13 | |
*** ChanServ sets mode: +v pvo | 23:13 | |
*** silassewell has quit IRC | 23:23 | |
*** lbieber has quit IRC | 23:28 | |
*** henriquetft has joined #openstack | 23:31 | |
*** pvo has quit IRC | 23:32 | |
*** pvo has joined #openstack | 23:32 | |
*** ChanServ sets mode: +v pvo | 23:32 | |
*** adjohn has joined #openstack | 23:33 | |
*** dmd17 has joined #openstack | 23:34 | |
*** justinsheehy has quit IRC | 23:53 | |
*** dmd17 has quit IRC | 23:56 | |
*** justinsheehy has joined #openstack | 23:58 | |
*** algorist has quit IRC | 00:00 | |
*** nettok has joined #openstack | 00:01 | |
*** lbieber has joined #openstack | 00:13 | |
*** ianl has quit IRC | 00:15 | |
*** comstud has quit IRC | 00:16 | |
*** jdmaturen has joined #openstack | 00:25 | |
*** lbieber has quit IRC | 00:25 | |
*** jdmaturen has quit IRC | 00:27 | |
*** metoikos has quit IRC | 00:34 | |
*** zul has joined #openstack | 00:39 | |
*** jdmaturen has joined #openstack | 00:44 | |
*** jdmaturen has quit IRC | 00:51 | |
*** alexh__ has joined #openstack | 00:52 | |
*** metoikos has joined #openstack | 01:08 | |
*** justinsheehy has quit IRC | 01:19 | |
*** justinsheehy has joined #openstack | 01:25 | |
*** alexh__ has left #openstack | 01:36 | |
*** silassewell has joined #openstack | 01:40 | |
*** jtdowney has quit IRC | 01:57 | |
*** gasbakid has quit IRC | 01:59 | |
*** blp-dinner has joined #openstack | 02:03 | |
*** metoikos has quit IRC | 02:06 | |
*** blp-dinner is now known as blpiatt | 02:12 | |
*** lbieber has joined #openstack | 02:23 | |
*** kashyapc has joined #openstack | 02:25 | |
*** joker[avd] has quit IRC | 02:30 | |
*** tjyang has joined #openstack | 02:33 | |
*** lbieber has quit IRC | 02:37 | |
*** blpiatt has quit IRC | 02:45 | |
*** silassewell has quit IRC | 03:03 | |
*** pvo has quit IRC | 03:29 | |
*** pvo has joined #openstack | 03:30 | |
*** pvo has joined #openstack | 03:30 | |
*** ChanServ sets mode: +v pvo | 03:30 | |
*** _anm has quit IRC | 03:33 | |
*** gnrfan has quit IRC | 03:38 | |
*** pvo has quit IRC | 03:39 | |
*** pvo has joined #openstack | 03:40 | |
*** pvo has joined #openstack | 03:40 | |
*** ChanServ sets mode: +v pvo | 03:40 | |
*** pvo has quit IRC | 03:44 | |
*** littleidea has quit IRC | 03:46 | |
*** pvo has joined #openstack | 03:46 | |
*** ChanServ sets mode: +v pvo | 03:46 | |
*** jsgotangco has joined #openstack | 03:50 | |
*** pvo has quit IRC | 03:56 | |
*** pvo has joined #openstack | 03:58 | |
*** ChanServ sets mode: +v pvo | 03:58 | |
*** pvo has quit IRC | 04:03 | |
*** littleidea has joined #openstack | 04:04 | |
*** pvo has joined #openstack | 04:06 | |
*** ChanServ sets mode: +v pvo | 04:06 | |
*** adjohn_ has joined #openstack | 04:11 | |
*** rajijoom has joined #openstack | 04:11 | |
*** adjohn has quit IRC | 04:11 | |
*** adjohn_ is now known as adjohn | 04:11 | |
*** rajijoom has quit IRC | 04:13 | |
*** tjyang has quit IRC | 04:15 | |
*** lbieber has joined #openstack | 04:26 | |
*** rajijoom has joined #openstack | 04:35 | |
*** stewart has joined #openstack | 04:35 | |
*** jdmaturen has joined #openstack | 04:36 | |
*** jsgotangco has quit IRC | 04:42 | |
*** lbieber has quit IRC | 04:53 | |
*** al-maisan has joined #openstack | 04:55 | |
al-maisan | hello .. starting nova-api fails with "ImportError: No module named daemon" .. I guess I am missing a dependency | 04:56 |
al-maisan | ? | 04:57 |
al-maisan | I guess it will be "python-daemon" | 04:57 |
*** pvo has quit IRC | 05:00 | |
dendrobates | al-maisan: yep | 05:00 |
dendrobates | al-maisan: nice to see you in our channel | 05:01 |
al-maisan | :-) | 05:01 |
*** justinsheehy has quit IRC | 05:17 | |
*** justinsheehy has joined #openstack | 05:22 | |
*** poisonarms has joined #openstack | 05:25 | |
*** poisonarms has left #openstack | 05:28 | |
*** algorist has joined #openstack | 05:29 | |
*** henriquetft has quit IRC | 05:32 | |
*** rajijoom has quit IRC | 06:00 | |
*** algorist has quit IRC | 06:04 | |
*** nettok_ has joined #openstack | 06:05 | |
*** z0 has quit IRC | 06:06 | |
*** jhalogen has quit IRC | 06:08 | |
*** nettok has quit IRC | 06:08 | |
al-maisan | When running "python run_tests.py" all but 2 pass: http://pastebin.ubuntu.com/469181/ | 06:11 |
al-maisan | Any idea what's going wrong here? | 06:11 |
*** sophiap has quit IRC | 06:17 | |
al-maisan | hmm .. probably an old tornado version | 06:20 |
*** ttx has joined #openstack | 06:24 | |
al-maisan | ah, apparently the 'HTTPHeaders' class was moved from httpserver to httputil in tornado | 06:25 |
al-maisan | this change is all it takes to make the last 2 tests pass: http://pastebin.ubuntu.com/469186/ | 06:28 |
*** nettok_ has quit IRC | 06:44 | |
*** jdmaturen has quit IRC | 06:55 | |
*** calavera has joined #openstack | 07:18 | |
*** allsystemsarego has joined #openstack | 07:48 | |
*** rajijoom has joined #openstack | 07:56 | |
*** metoikos has joined #openstack | 07:56 | |
*** miclorb has quit IRC | 08:00 | |
*** miclorb_ has joined #openstack | 08:00 | |
*** truff has left #openstack | 08:16 | |
*** miclorb_ has quit IRC | 08:18 | |
*** littleidea has quit IRC | 08:25 | |
*** al-maisan is now known as almaisan-away | 08:25 | |
*** almaisan-away is now known as al-maisan | 08:25 | |
*** adjohn_ has joined #openstack | 09:02 | |
*** adjohn has quit IRC | 09:04 | |
*** adjohn_ is now known as adjohn | 09:04 | |
*** allsystemsarego has quit IRC | 09:12 | |
*** allsystemsarego has joined #openstack | 09:13 | |
*** jtdowney has joined #openstack | 09:15 | |
*** bytee has quit IRC | 09:26 | |
*** bytee has joined #openstack | 09:48 | |
*** nefo has joined #openstack | 10:12 | |
*** miclorb_ has joined #openstack | 10:52 | |
*** sophiap has joined #openstack | 11:01 | |
*** miclorb_ has quit IRC | 11:11 | |
*** bobnormal has joined #openstack | 11:15 | |
uvirtbot | New bug: #610035 in nova "New admin create caused bug by using LDAP" [Undecided,New] https://launchpad.net/bugs/610035 | 11:15 |
bobnormal | openstack suitable for use with LXC / vserver / openvz virtualisation? | 11:20 |
*** jaker has quit IRC | 11:27 | |
*** justinsheehy has quit IRC | 11:27 | |
*** justinsheehy has joined #openstack | 11:27 | |
*** jaker has joined #openstack | 11:27 | |
*** terje has quit IRC | 11:27 | |
*** ericography has quit IRC | 11:27 | |
*** silassewell has joined #openstack | 11:27 | |
*** terje has joined #openstack | 11:27 | |
bobnormal | (mostly interested in compute cloud with container-based virt for IO bandwidth preservation vs. hypervisors) | 11:31 |
*** avsm has joined #openstack | 11:41 | |
*** nefo has quit IRC | 11:41 | |
*** jonesy has joined #openstack | 11:50 | |
*** jtdowney has quit IRC | 12:00 | |
*** jtdowney has joined #openstack | 12:04 | |
*** silassewell has quit IRC | 12:09 | |
*** al-maisan is now known as almaisan-away | 12:12 | |
*** poli has joined #openstack | 12:14 | |
*** jonesy_ has joined #openstack | 12:14 | |
*** silassewell has joined #openstack | 12:14 | |
*** justinsheehy has quit IRC | 12:16 | |
termie | when i look at bzr missing when i get conflicts, some commits have 'branch nick' in them... how do i set that? is there a way to assign a nick when i run bzr branch XXX? | 12:18 |
*** avsm has quit IRC | 12:20 | |
redbo | it's the name of the directory where you make your branch | 12:21 |
termie | if i do bzr branch init-repo foo; cp foo; bzr branch lp:nova nova it seems to result in no branch nick | 12:25 |
termie | at least when i do bzr missing | 12:25 |
termie | the commits i am trying to import have a branch nick, but the local ones conflict because they have none | 12:26 |
termie | s/local/local copies of the original commits at lp:nova/ | 12:27 |
redbo | that's more than I know how to deal with | 12:27 |
termie | i'll try a standalone repo and see if it changes things | 12:27 |
*** almaisan-away is now known as al-maisan | 12:27 | |
*** bahadir has joined #openstack | 12:34 | |
termie | hmm, so it seems that the branch nick isn't causing the conflict | 12:37 |
termie | http://pastebin.com/epxBcnk1 | 12:38 |
*** justinsheehy has joined #openstack | 12:38 | |
*** bahadir has quit IRC | 12:38 | |
termie | there must be something fishy in the mapping between commit "numbers" | 12:38 |
termie | does bzr have a secret commit id somewhere? | 12:39 |
termie | the revno, message, and so forth are all the same | 12:39 |
termie | i just don't know bzr well enough to know how it determines conflicts | 12:39 |
termie | the commits are obviously the same in all ways, but there must be a secret properties | 12:40 |
*** pvo has joined #openstack | 12:40 | |
*** ChanServ sets mode: +v pvo | 12:40 | |
*** rackerhacker has quit IRC | 12:44 | |
*** rackerhacker has joined #openstack | 12:45 | |
al-maisan | termie: try one of the other merge algorithms (e.g. --lca) ; they may yield better results, see also 'bzr help merge" | 12:46 |
al-maisan | s/'/"/ | 12:46 |
*** justinsheehy has quit IRC | 12:47 | |
*** aliguori has joined #openstack | 12:53 | |
aliguori | hrm, imap still down? | 12:53 |
aliguori | i guess it's not 9am EST yet.. | 12:53 |
*** metoikos has quit IRC | 12:53 | |
*** pvo has quit IRC | 12:55 | |
termie | a-ha! | 12:58 |
termie | bzr fast-export uses a different format that git's | 12:58 |
termie | s/that/than/ | 12:58 |
termie | and it is solvable by file munging :) | 12:59 |
*** al-maisan is now known as almaisan-away | 13:00 | |
*** croteb has quit IRC | 13:03 | |
*** justinsheehy has joined #openstack | 13:08 | |
*** pvo has joined #openstack | 13:11 | |
*** pvo has joined #openstack | 13:11 | |
*** ChanServ sets mode: +v pvo | 13:11 | |
*** z0 has joined #openstack | 13:13 | |
*** JordanRinke_ has joined #openstack | 13:23 | |
*** Phil_FL has joined #openstack | 13:26 | |
*** silassewell has quit IRC | 13:35 | |
*** silassewell has joined #openstack | 13:36 | |
*** adjohn has quit IRC | 13:39 | |
*** drewlande has joined #openstack | 13:42 | |
*** kashyapc has quit IRC | 13:43 | |
*** jhalogen has joined #openstack | 13:47 | |
termie | successfully fixed all the (known) bugs :) but now github is partially down so nobody can see :p | 13:56 |
*** MAR3007 has joined #openstack | 13:57 | |
*** MAR3007 is now known as MaxR_ | 13:58 | |
MaxR_ | Hello, I've downloaded, installed the SAIO and was interested in checking out Swift. I'm fairly new to cloud technologies in general and have been doing quite a bit of reading anywhere I could. | 13:59 |
*** avsm has joined #openstack | 14:00 | |
MaxR_ | As best I can tell, Swift is installed and functioning properly.. However, I can't help but feel I'm missing something here in my knowledge of how to utilize it. It appears that it has an HTTP API interface..so I'd guess there is some other management-type of software that I'd need in order to utilize it? | 14:00 |
*** philaway is now known as phildoc | 14:00 | |
MaxR_ | If someone could point me in the right directions, software package, docs, website, anything; I'd greatly appreciate it :) | 14:00 |
MaxR_ | I understand SAIO was released as more of a developer preview, however I have a firm believe that OpenStack is the future of cloud computing, and wanted to jump in and start gaining experience utilizing it | 14:01 |
notmyname | MaxR_: glad you got it up and running | 14:02 |
*** murkk has joined #openstack | 14:02 | |
notmyname | all the docs for swift are at swift.openstack.org | 14:02 |
notmyname | as far as taking to swift, you can use the Rackspace Cloud Files developer guide and/or any of the cloud files language bindings at github.com/rackspace | 14:03 |
MaxR_ | Unless I've missed a page somewhere, I've read through everything there. The docs seem to be written on how to configure Swift, and explaining the different pieces and applications... | 14:03 |
MaxR_ | there we go | 14:03 |
MaxR_ | that's what I was looking for (= | 14:03 |
MaxR_ | thank you | 14:03 |
notmyname | ya, we should probably make that more clear, I guess | 14:04 |
notmyname | did you have any issues or questions when setting up the SAIO? | 14:04 |
MaxR_ | As someone new to cloud computing, that part is very unclear from the website/wiki. And unfortunately it's hard to tell what other information in cloud computing is compatible with OpenStack :) | 14:04 |
MaxR_ | But I'm determined nevertheless to make this technology work hehe :) | 14:04 |
MaxR_ | Not at all.. the instructions were very clear for me with SAIO... | 14:05 |
MaxR_ | The only other comment I'd say.. | 14:05 |
notmyname | you can also use the st program that is included in the swift code | 14:05 |
MaxR_ | it may be helpful to provide another guide that explains how to set it up, but with actual servers instead of SAIO. As it stands, I'll have to extrapolate out what to do based on the SAIO and the other various pages that explain how the pieces work | 14:05 |
notmyname | or look at client.py for a callable python module | 14:05 |
notmyname | we are working on some deployment guide docs | 14:06 |
MaxR_ | I love a challenge, but I can imagine it may turn others away from checking out the technology if it isnt sufficiently easy for them to get it up and running | 14:06 |
MaxR_ | I must say though.. | 14:06 |
MaxR_ | I give the programmers some serious credit for their implementation of the various ring building apps and whatnot | 14:07 |
MaxR_ | it appears to be very well thought out and easy to use in general | 14:07 |
notmyname | thanks :-) | 14:07 |
MaxR_ | but retaining the power of many options as well | 14:07 |
bobnormal | sorry im using an oldschool irc client with short scrollback buffer .. i was asking about using openstack with container-based virtualisation systems (lxc/vserver/openvz)... anyone done this / have some basic info on direction to get it happening? | 14:08 |
MaxR_ | Okay, so my nxt step is to read the Rackspace Cloud Files developer guide, check out the cloud files language bindings at github.com/rackspace, the st program included in the code, and the client.py python module | 14:08 |
notmyname | that should keep you busy for a while :-) | 14:08 |
*** avsm has quit IRC | 14:08 | |
MaxR_ | hehe yeah, but I'm addicted to new technologies.. this is fun stuff to have the privilege of 'toying' with :) | 14:09 |
MaxR_ | I hope others see this as I do, the future of not only cloud computing, but computing in general | 14:09 |
MaxR_ | Anyway, thank you for the guidance notmyname. :) I greatly appreciate it | 14:09 |
MaxR_ | I'm sure I'll probably end up back here with more questions at some point | 14:10 |
notmyname | if you need any help or advice as you deploy to real servers, let us know. we've walked through a lot of that already for our clusters | 14:10 |
MaxR_ | Fair enough.. I'm doing my best to take notes as I go through this and learn it from a "newbie" perspective so that I can pass it along for others in the same situation | 14:10 |
notmyname | great. and we welcome patches to both the code and docs | 14:11 |
*** gustavomzw has joined #openstack | 14:11 | |
MaxR_ | Thank you again for providing such a gem to the world. | 14:12 |
* MaxR_ waves | 14:12 | |
*** MaxR_ has quit IRC | 14:12 | |
notmyname | bobnormal: I'm not ignoring you, I just don't have an answer | 14:12 |
bobnormal | notmyname: ok. | 14:13 |
bobnormal | notmyname: hoping to build an efficient video transcoding cluster using this stuff... otherwise will have to write something from scratch thats less useful / hardware-sharable for other tasks = not ideal | 14:15 |
notmyname | I"m sure someone can answer your questions, but I'm a dev on the storage side. You probably know more than I do about virtualization, etc | 14:15 |
bobnormal | notmyname: storage system in openstack... how useful for VERY large files with guaranteed IO throughput requirements? | 14:16 |
bobnormal | notmyname: the globalised replication concept would be super useful .. im just worried about performance with huge files (high resolution video) | 14:17 |
bobnormal | notmyname: kind of assuming it's not a good match .. would like to hear otherwise :) | 14:17 |
notmyname | swift supports 5GB files out of the box. but that is simply a constant in the code. we've talked about support for arbitrary sized files, but for now the answer is to split on the client side | 14:17 |
bobnormal | notmyname: why do you encourage a split? | 14:18 |
notmyname | IO throughput is a function of the cluster size | 14:18 |
notmyname | we have a few reasons for the 5GB limit: 1) larger uploads have a higher chance of failure, necessitation a complete re-upload 2) cluster balance is determined by the ratio of max file size to drive size | 14:19 |
notmyname | necessitation, nice new word. s/necessitation/necessitating | 14:20 |
notmyname | however, if you are using a swift cluster for strictly internal purposes and have a reliable connectivity to it, you could get rid of excuse 1 | 14:21 |
notmyname | and reason 2 is more of a cost/benefit analysis for your use case. for example, if you had only (or mostly) 20GB files, the balance wouldn't be as big of an issue. in our case (cloud files), we have to support both large and small files | 14:22 |
notmyname | (and replace 20GB files with any filesize you want) | 14:23 |
notmyname | the only limit on filesize is the size of the independent drives | 14:23 |
notmyname | so if you make a cluster with 2T drives, you will have an upper bound of 2T for your max file size. (not that I recommend doing that) | 14:24 |
notmyname | bobnormal: and back to the throughput question, it is also determined by the IO speed of a single disk spindle. aggregate throughput is based on the size of the cluster. single operation throughput is based on the speed of a single disk spindle | 14:25 |
*** littleidea_ has joined #openstack | 14:26 | |
*** dhdenny has left #openstack | 14:32 | |
*** almaisan-away is now known as al-maisan | 14:33 | |
bobnormal | notmyname: thanks, thats a lot clearer. | 14:33 |
*** bobnormal has quit IRC | 14:34 | |
anm_ | Hello, back at it on a MOnday - justinsb walked with through the installfest directions he wrote on instaling from Source, does anyone object on these instructions becoming a standalon page on the wiki? It's by far the most complete set to date | 14:35 |
anm_ | I'll be glad to build it and link it from the home page | 14:36 |
anm_ | on the wiki | 14:36 |
*** xfrogman5 has joined #openstack | 14:36 | |
*** croteb has joined #openstack | 14:37 | |
anm_ | also, after going through the excellent installfest instructions, when uploaded my first image to a bucket, I get a HUGE html error back from euca-upload-bundle | 14:38 |
*** silassewell has quit IRC | 14:38 | |
anm_ | i copied it ito my browser, and it says: 'nova.exception.NotFound' | 14:38 |
*** sirp2 has joined #openstack | 14:47 | |
*** silassewell has joined #openstack | 14:52 | |
*** phildoc has quit IRC | 14:57 | |
*** phillyd has joined #openstack | 14:58 | |
*** phillyd is now known as cursed | 14:58 | |
*** cursed is now known as phildoc | 14:58 | |
*** metoikos has joined #openstack | 15:01 | |
*** kashyapc has joined #openstack | 15:01 | |
*** adrian_otto has joined #openstack | 15:02 | |
*** rajijoom has quit IRC | 15:08 | |
*** dendrobates is now known as dendro-afk | 15:15 | |
*** kashyapc has quit IRC | 15:21 | |
*** sirp2 has quit IRC | 15:21 | |
*** jtdowney has quit IRC | 15:21 | |
*** pvo_ has joined #openstack | 15:28 | |
*** pvo_ has joined #openstack | 15:28 | |
*** ChanServ sets mode: +v pvo_ | 15:28 | |
*** pvo has quit IRC | 15:30 | |
*** pvo_ is now known as pvo | 15:30 | |
*** pvo has joined #openstack | 15:31 | |
*** ChanServ sets mode: +v pvo | 15:31 | |
*** dendro-afk is now known as dendrobates | 15:31 | |
*** croteb has quit IRC | 15:32 | |
*** croteb has joined #openstack | 15:32 | |
*** croteb has joined #openstack | 15:32 | |
*** croteb has joined #openstack | 15:33 | |
*** BartVB has quit IRC | 15:38 | |
*** lbieber has joined #openstack | 15:38 | |
*** kashyapc has joined #openstack | 15:39 | |
*** BartVB has joined #openstack | 15:44 | |
*** kashyapc has quit IRC | 15:44 | |
*** MaxR_ has joined #openstack | 15:53 | |
*** pvo_ has joined #openstack | 15:54 | |
*** ChanServ sets mode: +v pvo_ | 15:54 | |
anm_ | does Openstack have a pastebin of its own? | 15:55 |
*** calavera has quit IRC | 15:56 | |
anm_ | i have a BIG error message I'd like to pb if someone has a sec to look at it | 15:56 |
g0rdy | anm_: good idea | 15:56 |
creiht | anm_: Not yet | 15:56 |
creiht | mtaylor: -^ :) | 15:56 |
anm_ | :) | 15:56 |
mtaylor | creiht: :) thanks - somehow it fell off of my todo list. | 15:56 |
anm_ | etherpad would probably work, but does no markup | 15:56 |
* mtaylor uses pastebin.com | 15:57 | |
*** pvo has quit IRC | 15:58 | |
*** pvo_ is now known as pvo | 15:58 | |
pvo | have a etherpad.opstack.org | 15:59 |
MaxR_ | Okay.. so Swift allows me to store data into an HTTP-driven object store... through a tip earlier here, I read the Cloud Files Developer Guide which mentions a few windows apps that give me FTP-like access to the storage space...but I'm not clear on how the data stored into Swift would be accessible without an HTTP call? | 15:59 |
pvo | er, that'd be etherpad.openstack.org | 15:59 |
notmyname | MaxR_: the trick is setting the auth url | 16:00 |
MaxR_ | Like if let's say, I wanted to have my backup software store directly to it...or if I wanted to run a VMWare container off it? | 16:00 |
notmyname | MaxR_: for example, duplicity (I'm told) allows you to change the auth url. CyberDuck doesn't | 16:00 |
*** jtdowney has joined #openstack | 16:00 | |
*** kashyapc has joined #openstack | 16:00 | |
creiht | MaxR_: My guess on the ftp software is that it acts as a proxy, converting the ftp requests to swift | 16:01 |
notmyname | since swift == cloud files, the API works. but, of course, it can't work if it doesn't know how to talk to your swift cluster | 16:01 |
MaxR_ | yeah.. I found out Cloud Files Manager is hard coded to work with Rackspace's cloud | 16:01 |
MaxR_ | already sent the developer an email.. he had never even heard of openstack.. I informed him and pointed to website | 16:01 |
notmyname | that surprises you? ;-0 | 16:01 |
notmyname | ;-) | 16:01 |
MaxR_ | not really.. hehe :) I feel like I've ventured into the wild wild west | 16:01 |
creiht | hopefully now that swift is open source, more of those tools will allow for configurable servers | 16:02 |
LoRez | is the guy that gave the OSCON keynote about openstack in here? | 16:02 |
notmyname | ah, that's not what I thought you were referring to | 16:02 |
creiht | LoRez: -> wreese | 16:02 |
creiht | oh | 16:02 |
notmyname | unless he meenw lew | 16:02 |
notmyname | means | 16:02 |
creiht | LoRez: you talking about for swift or openstack in general? | 16:02 |
creiht | yeah | 16:02 |
MaxR_ | But that still doesn't explain how to access the data without one of those applications | 16:02 |
MaxR_ | those apps are great if all I want to do is store data into an FTP-like account... | 16:03 |
LoRez | there was a swift keynote? | 16:03 |
notmyname | MaxR_: I thought you were referring to the rackspace cloud files manager (in the rackspace control panel) | 16:03 |
MaxR_ | but doesnt help if I need to access that same data directly | 16:03 |
creiht | LoRez: there was a talk at oscon about how we built swift | 16:03 |
MaxR_ | no.. www.cloudfilesmanager.com | 16:03 |
LoRez | a talk or a keynote? | 16:03 |
creiht | talk | 16:03 |
creiht | I read talk when you said keynote :) | 16:03 |
MaxR_ | I did read that iSCSI is in the works.. that would easily solve my question.. | 16:04 |
LoRez | I'm speaking about the keynote, although someone probably made the same mistake in your talk | 16:04 |
MaxR_ | but there must be a way to do it now, as this code is already in use? | 16:04 |
notmyname | MaxR_: how do you want to access it? if you use a 3rd party tool, you should be able to access it through the same tool. are you looking for something like public access? | 16:04 |
LoRez | it's irc.freenode._*NET*_, not ._*COM*_ | 16:04 |
creiht | LoRez: Feel free to ask your question here, as someone may be able to answer | 16:05 |
notmyname | MaxR_: I'm a little confused on what you mean about accessing the data. it's not write-only :-) | 16:05 |
MaxR_ | Okay let me give an example what I mean | 16:05 |
creiht | MaxR_: Perhaps if you let us know a little more about what you want to do, then we can point you in the right direction | 16:05 |
*** jbryce has joined #openstack | 16:06 | |
MaxR_ | The simplest example I can think of.. is tar backups on a linux server.. people often want to store their backups offsite. If I wanted to provide someone space that they could map on their Linux server.. so that their swift-stored data shows up as a folder on their linux server | 16:07 |
mcgurrin | MaxR, are you asking how to access it with a more standard access protocol that has been around for longer like FTP, SFTP, etc. instead of the API? | 16:07 |
notmyname | MaxR_: swift doesn't map directly to a file system | 16:07 |
mcgurrin | oh, it does with software | 16:07 |
notmyname | MaxR_: swift is a blob store, not a block store | 16:08 |
MaxR_ | That's the part I'm trying to figure out.. if it doesnt map directly to a file system.. that means there is no way to directly access it without some sort of interface.. correct? | 16:08 |
mcgurrin | I saw a cloudfiles FUSE driver a while ago | 16:08 |
notmyname | mcgurrin: yes. redbo wrote that | 16:08 |
*** pvo has quit IRC | 16:08 | |
mcgurrin | I think that might solve this issue | 16:08 |
notmyname | MaxR_: the way to access the data is through the REST API. you could write a wrapper for it or use other products (like duplicity) | 16:09 |
MaxR_ | alright... then let's move on to example #2 | 16:09 |
mcgurrin | I have a question or 2 once this is solved | 16:09 |
creiht | MaxR_: If it helps conceptually, swift has similar functionality to S3 | 16:09 |
*** pvo has joined #openstack | 16:09 | |
*** ChanServ sets mode: +v pvo | 16:09 | |
MaxR_ | creiht: I'm about as new as new gets to cloud computing in general. I thought I understood the basics, lol, but I'm beginning to see I apparently didn't even have that. hehe | 16:10 |
MaxR_ | Okay.. if let's say I get Nova up and running.. between Nova and Swift, I'd *guess* that there should be some way to run a VPS container...? Or is that a bad assumption? | 16:11 |
creiht | MaxR_: So in the above example, swift might be used to store nova images, but to run a container (if I understand the terminology correctly) I think you are looking for block storage | 16:12 |
creiht | Which swift is not | 16:12 |
mcgurrin | given that this is opensource software I would assume I have the legal ability if I do it correctly but I would like to avoid angering people, can I put up a ovf format virtual appluance set up by following the SAIO guide for public download? | 16:12 |
notmyname | I think that's a bad assumption (for now). I would hope that would be a possibility for the future. but you can't run an image stored in swift | 16:12 |
creiht | mcgurrin: by all means! :) | 16:12 |
notmyname | mcgurrin: go for it! | 16:12 |
mcgurrin | ok, I will do that once the ubunto download finishes | 16:13 |
*** xfrogman5 has quit IRC | 16:13 | |
MaxR_ | I know this is kind of an ambiguous question here.. but what do most people use this kind of object store for? | 16:13 |
mcgurrin | I will send out the link here once its up | 16:13 |
notmyname | MaxR_: 2 main use cases: backups and static web content | 16:13 |
creiht | MaxR_: the biggest use cases are for storing data, and backups | 16:13 |
creiht | hehe | 16:13 |
MaxR_ | Okay.. so for static web content.. Is there a way to set it up so that the content can be pulled from the object store, without authorization? Like if let's say I put an .ISO up on it, and wanted to link to it from my website for others to download... | 16:14 |
creiht | MaxR_: That is coming | 16:14 |
notmyname | currently, that only works on cloud files through the cdn | 16:15 |
creiht | In the current cloud files from Rackspace, that would be done through the CDN integration | 16:15 |
notmyname | but, what creiht said | 16:15 |
creiht | hah | 16:15 |
notmyname | ok, only one of us should answer him :-) | 16:15 |
* creiht hands of to notmyname | 16:15 | |
creiht | :) | 16:15 |
notmyname | but I want to go to lunch! | 16:15 |
creiht | me too :) | 16:15 |
MaxR_ | okay, I think I'm understanding this now. Just like the docs say, this is a new method of storing.. | 16:16 |
creiht | MaxR_: Short answer is, public containers are high on the list | 16:16 |
notmyname | MaxR_: so any other questions before we go to lunch? | 16:16 |
creiht | as is the piece that works with the CDN | 16:16 |
MaxR_ | Nope, you guys have been a great help for me understanding this. :) | 16:16 |
MaxR_ | Thanks for being so approachable and willing to help me understand | 16:17 |
notmyname | any time (unless we are at lunch) | 16:17 |
MaxR_ | lol lunch sounds good | 16:17 |
MaxR_ | I think I'll follow your lead on that one | 16:17 |
creiht | and feel free to leave any questions on IRC, and we can answer them when we get back from lunch :) | 16:17 |
*** MaxR_ is now known as MaxR_AWY | 16:17 | |
mtaylor | lunch. bah | 16:18 |
jaypipes | mtaylor: yeah...coffee++ | 16:18 |
mtaylor | mmm. coffee... | 16:19 |
mcgurrin | I should have lunch to, I'ss see everyone later | 16:19 |
*** jsmith-away is now known as jsmith | 16:20 | |
*** gnrfan has joined #openstack | 16:25 | |
*** kashyapc has quit IRC | 16:31 | |
*** jdmaturen has joined #openstack | 16:37 | |
*** jdmaturen has joined #openstack | 16:38 | |
*** kashyapc has joined #openstack | 16:45 | |
uvirtbot | New bug: #610140 in nova "r164 - ApiEc2TestCase failures - HTTPHeaders" [Undecided,New] https://launchpad.net/bugs/610140 | 16:46 |
*** croteb has quit IRC | 16:46 | |
jaypipes | anyone know Ewan Mellors' | 16:47 |
jaypipes | IRC nick? | 16:47 |
_0x44 | jaypipes: It was originally ewanmellor it may now be ewan_ | 16:50 |
jaypipes | _0x44: k, cheers :) | 16:50 |
_0x44 | :) | 16:50 |
jaypipes | _0x44: didn't know if it was something obscure like _0x44 ;) | 16:51 |
*** joshuamckenty has joined #openstack | 16:51 | |
*** kashyapc has quit IRC | 16:51 | |
_0x44 | jaypipes: I could change to rootkit like on Ryder, then everyone would know who I am :P | 16:51 |
jaypipes | _0x44: :) | 16:52 |
*** al-maisan is now known as almaisan-away | 16:52 | |
* mtaylor still wishes that lp had an irc network so that his lp id would match his irc id .... | 16:53 | |
* mtaylor realizes others probably do not share his wish | 16:53 | |
lbieber | mtaylor: any updates on getting the swift mailing list going? | 16:53 |
mtaylor | lbieber: nope. lemme go poke someone again | 16:54 |
lbieber | mtaylor: please do, its taking way too long | 16:54 |
*** pvo has quit IRC | 16:57 | |
mtaylor | lbieber: https://answers.edge.launchpad.net/launchpad-registry/+question/117954 and https://bugs.edge.launchpad.net/launchpad-registry/+bug/608785 in case you want to follow along | 16:59 |
uvirtbot | Launchpad bug 608785 in launchpad-registry "OOPSes due to renaming of teams with mailing lists/privacy switch" [Low,Incomplete] | 16:59 |
joshuamckenty | mtaylor: is there somewhere I can grab logs of this channel? | 17:08 |
joshuamckenty | I keep missing interesting things while I'm travelling and don't have scrollback | 17:08 |
joshuamckenty | I can log in from a server if needed, just checking | 17:08 |
mtaylor | joshuamckenty: I think someone setup a logger... but I've been meaning to get something more official going | 17:09 |
_0x44 | joshuamckenty: Didn't _cerberus_ set you guys up with accounts on his bip server? | 17:09 |
mtaylor | joshuamckenty: lemme add that to my list of things to get done here RSN | 17:09 |
joshuamckenty | thanks | 17:09 |
joshuamckenty | _0x44: negative | 17:09 |
joshuamckenty | what's bip? | 17:09 |
joshuamckenty | build in progress? | 17:10 |
joshuamckenty | I have lots of servers, that's not an issue | 17:10 |
_0x44 | joshuamckenty: It's an IRC proxy that does scroll back | 17:10 |
joshuamckenty | ooh, nice | 17:10 |
_0x44 | http://bip.t1r.net/ | 17:10 |
joshuamckenty | why am I not running that? | 17:10 |
_0x44 | I use screen+irssi, but most of the ozone guys use bip+colloquy | 17:10 |
*** aliguori has quit IRC | 17:11 | |
joshuamckenty | Yeah, I'm not very good with irssi :( | 17:11 |
joshuamckenty | I use screen for everything else, though | 17:11 |
joshuamckenty | multi-user screen is the best form of pair programming on the planet | 17:11 |
_0x44 | joshuamckenty: I can give you a tar of the logs since the room was opened. | 17:11 |
_0x44 | Wholly agree about that. | 17:11 |
joshuamckenty | that would be awesome | 17:11 |
_0x44 | Email address you'd like it sent to? | 17:13 |
*** algorist has joined #openstack | 17:13 | |
joshuamckenty | jmckenty@gmail.com - my nasa email is full | 17:14 |
joshuamckenty | :( | 17:14 |
joshuamckenty | 400MB email box suckz | 17:14 |
_0x44 | Oh wow... one sec. | 17:14 |
*** kashyapc has joined #openstack | 17:16 | |
mcgurrin | joshuamckenty, I have a cool solution you can use to avoid size hassles for email | 17:16 |
joshuamckenty | is it called POP3? | 17:17 |
mcgurrin | fetchmail+sendmail+procmail+mail server | 17:17 |
* joshuamckenty is married to IMAP | 17:17 | |
joshuamckenty | yeah, I used to run my own mail server for this sort of thing | 17:17 |
_0x44 | joshuamckenty: You should have mail | 17:17 |
mcgurrin | no, download email through fetchmail to a different mail server | 17:17 |
joshuamckenty | before SPF came out | 17:17 |
joshuamckenty | ah, hmm | 17:17 |
joshuamckenty | yeah, that would work | 17:17 |
mcgurrin | then access through imap from there | 17:17 |
joshuamckenty | that's kind of insane | 17:17 |
joshuamckenty | :) | 17:17 |
joshuamckenty | _0x44: thank you | 17:18 |
mcgurrin | I use it to collect my 3+ different accounts into one | 17:18 |
_0x44 | joshuamckenty: You're welcome :) | 17:18 |
mcgurrin | it also bypasses limits on connections per hour when you have a lot of devices like me | 17:18 |
joshuamckenty | K, I'm offline again for a while - driving to Seattle to get a flight to Zurich. | 17:18 |
jbryce | _0x44: can i get a copy of the logs too? jbryce@rackspace.com | 17:18 |
*** jtdowney has quit IRC | 17:18 | |
joshuamckenty | Will consider the fetchmail solution | 17:18 |
joshuamckenty | and get bip set up shortly | 17:19 |
*** BartVB has quit IRC | 17:19 | |
mcgurrin | I can help you with it if you need it, just let me know | 17:19 |
joshuamckenty | I'm still working on turning NASA Nebula roadmap into blueprints | 17:19 |
joshuamckenty | I know I owe a lot of people that roadmap | 17:19 |
joshuamckenty | will work on it on the plane | 17:19 |
*** jsmith is now known as jsmith-away | 17:19 | |
*** BartVB has joined #openstack | 17:19 | |
joshuamckenty | oh, dendrobates : | 17:19 |
mcgurrin | I'm installing ubunto right now in a vm to set up via the SAIO instructions | 17:19 |
joshuamckenty | A couple more companies want to join openstack, can I introduce them to you, or to Jim? | 17:20 |
*** _jcsmith has joined #openstack | 17:20 | |
joshuamckenty | dendrobates: specifically to work on security integration | 17:20 |
jbryce | joshuamckenty: you can send them to any of us, but i think at the end of the day mark collier is managing those relationships | 17:21 |
joshuamckenty | ah, cool | 17:21 |
*** sirp2 has joined #openstack | 17:22 | |
jbryce | mark.collier@rackspace - he's on vacation this week so rick or jim or i can get things going | 17:22 |
joshuamckenty | I told them it was easier just to fire up colloquoy, but they said something about logos and stuff | 17:22 |
* joshuamckenty laughs at silly business people | 17:22 | |
jbryce | haha…yeah…that sounds like mark's area. = ) | 17:22 |
*** BartVB has quit IRC | 17:24 | |
joshuamckenty | k, I'm out | 17:26 |
*** joshuamckenty has quit IRC | 17:26 | |
_0x44 | jbryce: Email is sent | 17:28 |
jbryce | thanks | 17:28 |
_0x44 | No problem :) | 17:28 |
*** almaisan-away has quit IRC | 17:32 | |
*** jbryce has quit IRC | 17:32 | |
*** mray has joined #openstack | 17:32 | |
mcgurrin | any recommendations on how to include the applicable licenses in my virtual appliance? | 17:33 |
*** almaisan-away has joined #openstack | 17:33 | |
termie | aha, i didn't realize i was filtering out the emails about reviews | 17:34 |
creiht | mcgurrin: applicable licenses, as in like the apache license for swift? | 17:34 |
mcgurrin | yes that and the ubuntu licence | 17:34 |
holoway | mcgurrin: I'm not sure that's even required, beyond having the license available on disk | 17:36 |
holoway | for example, there is no 'ubuntu license' neccessarily | 17:36 |
mcgurrin | I mean the one they release it under | 17:36 |
mcgurrin | but I see your point about being on disk | 17:36 |
creiht | mcgurrin: yeah not entirely sure I understand | 17:37 |
*** jbryce has joined #openstack | 17:37 | |
mcgurrin | Do I need to have a license agreement page that needs to be gone through or anything before people download my pre made SAIO virtual appluance | 17:38 |
creiht | mcgurrin: ahh... I wouldn't think so | 17:38 |
cw | creiht: do you have any rough guesses on how often metadata is read w/o the data? | 17:38 |
mcgurrin | appliance* | 17:38 |
cw | creiht: and how often it's updated? | 17:38 |
creiht | cw: metadata is read on HEAD requests | 17:38 |
mcgurrin | ok, just checking | 17:38 |
holoway | mcgurrin: it's very polite of you to ask | 17:38 |
creiht | cw: it is updated on a POST request, which in our use case doesn't happen very often | 17:39 |
cw | creiht: right, so my thinking is that's probably not much more common than reading the data | 17:39 |
mcgurrin | thanks holoway | 17:39 |
holoway | mcgurrin: in general, there is a NOTICE file in apache licensed projects | 17:39 |
creiht | cw: it depends, on some use cases (mainly backup clients) they can do a lot of HEADS to get info on files currently in swift | 17:39 |
mcgurrin | Where would that go? | 17:39 |
holoway | mcgurrin: in general, you need to make sure that file and the license are in a reasonable location | 17:40 |
holoway | mcgurrin: see section 4 (d) of the ASL | 17:40 |
mcgurrin | so I should probably copy them to the archive that it will be distributed in then, not just have them inside the vm? | 17:40 |
creiht | cw: In a future version, I would imagine that the metadata could also be read to help determine which object server to use to serve data | 17:41 |
notmyname | mtaylor: why are swift devs getting nova bug emails now? | 17:42 |
creiht | cw: for example we have been talking about a change that would get versions from each object server before serving, and ensuring that only the most recent versions get served | 17:42 |
mtaylor | notmyname: uh. because I suck? | 17:42 |
mtaylor | notmyname: I think I was supposed to be investigating that last week and then forgot :) | 17:42 |
notmyname | looks likw swift is subscribed to nova | 17:42 |
mtaylor | it really shouldn't be | 17:42 |
notmyname | i agree :-) | 17:43 |
creiht | holoway: btw... were you waiting on me for anything for your auth fix? | 17:43 |
holoway | creiht: yes! | 17:43 |
redbo | if they subscribe to each other, will the universe explode? | 17:43 |
holoway | creiht: I totally broke python on my laptop, like a moron :) | 17:43 |
creiht | I hadn't seen it come back for review, so I just wanted to make sure I wasn't blocking | 17:43 |
holoway | creiht: so once I get bzr up again, I'll push my changes | 17:43 |
creiht | holoway: ahh cool... just wanted to check in | 17:44 |
holoway | creiht: it's a small change, really just needs t3esting | 17:44 |
holoway | creiht: thank you | 17:44 |
creiht | cool | 17:44 |
mtaylor | actually, neither ~nova nor swift should have been subscribed to nova bugs... that's what ~nova-bugs is for | 17:44 |
mtaylor | notmyname: thanks! fixed | 17:44 |
notmyname | thanks | 17:44 |
creiht | yay | 17:44 |
holoway | creiht: I assumed that a schema change to have unique column constrains to make INSER OR REPLACE work was harder than just having a conditional that does INSERT/UPDATE | 17:45 |
creiht | holoway: agreed | 17:45 |
holoway | creiht: cool - I pushed the change, but it needs tested | 17:46 |
creiht | holoway: cool.. when I get a chance, I will try it out | 17:46 |
holoway | creiht: thanks :) | 17:47 |
mcgurrin | how much configuration is location/cluster size dependent? | 17:49 |
mcgurrin | I was thinking it might be cool to set up a virtual cluster distribution for locations that already have servers doing virtualization | 17:50 |
creiht | mcgurrin: Most of everything that is cluster size dependent is going to be in the ring | 17:50 |
mcgurrin | ring? | 17:50 |
mcgurrin | I mean in the config and such is there any differences | 17:51 |
creiht | mcgurrin: The ring is what holds the mapping of requests to the servers/devices | 17:51 |
creiht | ahh | 17:51 |
mcgurrin | ah, so that would not affect a new installation, just once there are actually files in it? | 17:52 |
mcgurrin | I was thinking that this would also be useful for testing load, especially scalability, you could run up to several VM's per machine and see if to many machines would start to slow it down as well | 17:53 |
creiht | well lets back up a bit, maybe I am misunderstanding your question | 17:53 |
creiht | ok | 17:53 |
creiht | So from an individual server's perspective, it doesn't have to know about the rest of the cluster | 17:53 |
mcgurrin | I was thinking besides just a SAIO virtual appluance I could distribute a virtual cluster set of alppliances | 17:54 |
creiht | cool | 17:54 |
creiht | The ring is what holds the smarts/config for where everything goes | 17:54 |
creiht | So if you get a basic storage vm setup, you could pop as many of those you want, then set up a ring that points to all those servers | 17:55 |
creiht | mcgurrin: btw... have you looked at this? http://swift.openstack.org/development_guidelines.html | 17:55 |
mcgurrin | so there is config needed that is dependant on the number of servers? | 17:55 |
mcgurrin | I will | 17:55 |
creiht | mcgurrin: Not from the servers perspective | 17:55 |
mcgurrin | then from where? | 17:56 |
creiht | All the config values are dependent on what the individual server can handle | 17:56 |
*** pvo has joined #openstack | 17:56 | |
*** pvo has joined #openstack | 17:56 | |
*** ChanServ sets mode: +v pvo | 17:56 | |
creiht | mcgurrin: Pretty much everything that is server away is in the ring | 17:57 |
notmyname | mcgurrin: the ring is the only thing that knows about the whole cluster | 17:57 |
creiht | mcgurrin: doh... I gave you the wrong link | 17:57 |
creiht | http://swift.openstack.org/deployment_guide.html | 17:57 |
creiht | Is what I meant | 17:57 |
creiht | http://swift.openstack.org/overview_architecture.html | 17:58 |
mcgurrin | ah, ok, so I could distribute just 2 images, the ring and the cluster node image and then have some way to detect the number of cluster nodes and set them up in the ring? | 17:58 |
creiht | Also has some overview of the ring that might help | 17:58 |
creiht | mcgurrin: I think you could do it with one image | 17:58 |
mcgurrin | that would then give me a fully scalable cluster VM set that could be deployed to any number of VM's? | 17:58 |
jonesy | creiht: my first look at the deployment guide -- nice work! | 17:58 |
mcgurrin | oh | 17:58 |
JordanRinke_ | any bzr gurus here have a second to chat? | 17:59 |
creiht | Include the ring stuff in the image, the user chooses one of them to do admin from, builds the ring, and then copies the ring to all of the images | 17:59 |
mtaylor | JordanRinke_: what' sup? | 17:59 |
JordanRinke_ | need to figure out how to review a merge request and commit / deny it | 17:59 |
creiht | jonesy: thanks! It is a first stab, but let me know if there are any areas where you find questions | 18:00 |
*** _jcsmith has quit IRC | 18:00 | |
*** croteb has joined #openstack | 18:00 | |
mtaylor | JordanRinke_: so, there's two ways - you can do it all via email, or you can do it via the web interface | 18:00 |
JordanRinke_ | mind if I send you a dm mtaylor? | 18:00 |
*** jsmith-away is now known as jsmith | 18:01 | |
mtaylor | JordanRinke_: go for it | 18:01 |
*** binaryWarrior has joined #openstack | 18:04 | |
*** jtdowney has joined #openstack | 18:06 | |
mcgurrin | Is there any sensitive data in the ring files? | 18:07 |
notmyname | no | 18:07 |
mcgurrin | ok, I am thinking that it might be possible to make a eplication service then | 18:08 |
notmyname | well, it has all of you servers in it, so don't give it to someone who you don't want to know your network topology | 18:08 |
mcgurrin | yeah but that should not be to much of an issue for someone using my virtual apliances | 18:09 |
notmyname | that is, it details each drive on each machine and has the IP addresses and a weight for the drive | 18:09 |
mcgurrin | yeah, I got that but this system is mostly for testing and small environments, a major production environment will probably not use my applances | 18:10 |
mcgurrin | the service would be only for my appliance users | 18:10 |
notmyname | so yeah, no problems distributing the ring files then | 18:10 |
*** chromakode has joined #openstack | 18:11 | |
*** jsmith is now known as jsmith-away | 18:12 | |
*** jeevan_ullas has joined #openstack | 18:12 | |
*** eth00 has quit IRC | 18:16 | |
*** jsmith-away is now known as jsmith | 18:16 | |
*** greenisus has joined #openstack | 18:18 | |
*** karmabot has joined #openstack | 18:25 | |
chromakode | karmabot, logs | 18:25 |
karmabot | logs(0): http://174.143.201.82/logs/ | 18:25 |
*** almaisan-away is now known as almaisan | 18:25 | |
* mtaylor really needs to put a hostname on that | 18:26 | |
chromakode | :) | 18:27 |
mtaylor | chromakode: ok. now we need to write a nice pretty web front end for those logs... | 18:27 |
*** ttx_ has joined #openstack | 18:27 | |
mtaylor | chromakode: or convince someone else to | 18:27 |
*** ttx_ has quit IRC | 18:28 | |
*** mray has quit IRC | 18:29 | |
*** dendrobates is now known as dendro-afk | 18:33 | |
*** _buck has joined #openstack | 18:35 | |
*** rdw has quit IRC | 18:36 | |
jbryce | mtaylor: i'm doing dns stuff right now...do you have a hostname in mind? | 18:37 |
mtaylor | jbryce: uh... how about irclogs.openstack.org ? | 18:38 |
jbryce | ok...it's set up | 18:40 |
mtaylor | jbryce: know anybody who wants to write a quick web frontend on top of a pile of irc logs? | 18:40 |
*** ChanServ sets mode: +o mtaylor | 18:40 | |
*** mtaylor changes topic to "Find out more and get involved here: http://www.openstack.org | Nova Docs: nova.openstack.org | Swift Docs: swift.openstack.org | Wiki: wiki.openstack.org | http://wiki.openstack.org/LifeWithBzrAndLaunchpad | http://irclogs.openstack.org/logs" | 18:40 | |
jbryce | the dns hasn't propagated yet so it may take a little while longer for that link to work | 18:41 |
mtaylor | good point | 18:41 |
MaxR_AWY | is iSCSI support planned or just being considered? | 18:42 |
*** MaxR_AWY is now known as MaxR_ | 18:42 | |
notmyname | MaxR_: for swift? | 18:42 |
notmyname | or nova? | 18:43 |
MaxR_ | I guess both? | 18:43 |
MaxR_ | The mention I saw just said "OpenStack" | 18:43 |
notmyname | ah. | 18:43 |
MaxR_ | article on securecloudreview.com | 18:43 |
notmyname | I would guess nova (compute) then. there is no plan (that I am aware of) for iSCSI support in swift | 18:44 |
*** jfontan has joined #openstack | 18:45 | |
jbryce | MaxR_: that article was referring to compute | 18:45 |
jbryce | MaxR_: it's a little confusing because they mentioned the storage project, but where it talks about supported storage mechanisms in that article (AoE,pNFS,iSCSI), it's referring to storage for the virtual compute | 18:47 |
MaxR_ | alright, thank you for the explanation | 18:47 |
*** mattf has quit IRC | 18:48 | |
jbryce | mtaylor: have you seen http://curtis.lassam.net/software/lumberjack/ ? | 18:51 |
mtaylor | jbryce: I have not | 18:52 |
mtaylor | jbryce: ooh. that web front end is nice | 18:52 |
mtaylor | jbryce: well, I say nice - it exists - there doesn't seem to be much data in his example :) | 18:53 |
jbryce | yeah | 18:53 |
jbryce | mtaylor: i just googled and found it, so i know nothing else about it other than it seems to try to solve a similar problem (and it's bsd licensed) | 18:53 |
mtaylor | jbryce: http://github.com/chromakode/karmabot/ is the one chromakode just stuck in here | 18:55 |
*** ianw has joined #openstack | 18:58 | |
*** jhalogen has quit IRC | 19:01 | |
*** aliguori has joined #openstack | 19:03 | |
*** adrian_otto has quit IRC | 19:08 | |
*** henriquetft has joined #openstack | 19:08 | |
MaxR_ | It would be great if Nova and Swift were combined in such a way that would allow Swift to store the VMs that Nova runs. :P | 19:09 |
creiht | MaxR_: that is the plan :) | 19:09 |
MaxR_ | I found a project called Nimbus that appears to have done similar already.. however their backend storage component Cumulus is nowhere near as good as Swift. No idea on their compute portion | 19:10 |
MaxR_ | creiht, I know this is a bad question to even ask.. but is there any idae at this point how far out that idea is from beign implemented? | 19:11 |
* MaxR_ watches as his spelling skills go out the window as the day progresses... | 19:11 | |
creiht | MaxR_: That is more on the nova side of things, but last I looked, I thought it was on the roadmap for the initial release | 19:12 |
redbo | Well, we'll store the images, not the actual vms. You need block storage for that. | 19:12 |
creiht | yeah what redbo said | 19:12 |
*** jfontan has left #openstack | 19:12 | |
MaxR_ | I see that Amazon has cloud block storage...are there any other projects out there doing the same? | 19:12 |
MaxR_ | EBS.. | 19:13 |
creiht | cumulus nimbus is very difficult to search for | 19:13 |
*** mattf has joined #openstack | 19:13 | |
MaxR_ | lol creiht.. yeah I found it the same http://www.nimbusproject.org (check their changelog for info on cumulus) | 19:14 |
creiht | k | 19:14 |
MaxR_ | not very well thought out name :P | 19:14 |
anm_ | MaxR_: I'd think if the distro running swift has an iscsi initiator, you can mount an iSCSI server and be off an running on iSCSI with swift. it shouldn;t care | 19:15 |
creiht | anm_: swift doesn't have support for iSCSI (unless you are talking about running them side by side) | 19:16 |
MaxR_ | My reading on Nova is not as thorough as I've been on Swift, so I apologize ahead of time if this question is answered somewhere on the site. But how does Nova do the user scheduling of CPU time for a VM? Is it per-core? | 19:17 |
g0rdy | Hi folks, is there anyone in St Louis that can evangelize OpenStack for us, in person? | 19:18 |
MaxR_ | anm_: From what I've gathered today, swift to use iSCSI you'd need a block-level storage. swift is object-level storage. | 19:18 |
anm_ | creiht: I was referring to side by side, so swift's node just mounts an iscsi target and mounts/maps it into the local filesystem - swift wouldn;t necessarily know that /srv/node/1-999 is an iSCSI target, right? | 19:19 |
MaxR_ | err that first swift shouldn't be in my comment :P | 19:19 |
MaxR_ | evangelize.. lol | 19:19 |
anm_ | I may be thinking about it all wrong too | 19:19 |
creiht | anm_: in theory yes, but not sure you would want to do that :) | 19:19 |
anm_ | creiht: probably not, but I started building my cloud with xen, so I have this 28TB iSCSI server sitting here :( | 19:20 |
creiht | ahh | 19:20 |
MaxR_ | anm_: All data in swift is stored and retrieved via an HTTP API.. to utilize the data in a live environment would require an intermediary piece of software, like an FTP-type client | 19:20 |
anm_ | so, I can mount each drive as jbod on my swift box | 19:20 |
creiht | right | 19:21 |
MaxR_ | now that makes sense... | 19:21 |
creiht | that would work, but I would be concerned about performance | 19:21 |
anm_ | creiht: yes, performanc is an issue | 19:21 |
creiht | Now a better use for that server may be to use with nova for block storage | 19:22 |
anm_ | I have dual gb nics in a separate iSCSI network, lagged together, and MPIO'd together on my stoareg server | 19:22 |
anm_ | creiht: That's actaully more in line with how I'm goig to use it | 19:22 |
creiht | Then you could expose that to nova through whatever block storage method they provide | 19:24 |
creiht | Take all that with a grain of salt of course, since I work mainly on swift :) | 19:24 |
creiht | so looking at cumulus docs, it looks like it is a very simple implementation of the S3 api on top of a file system | 19:26 |
creiht | It isn't distributed | 19:27 |
MaxR_ | yeah.. :( | 19:27 |
MaxR_ | inferior for sure | 19:27 |
*** dendro-afk is now known as dendrobates | 19:28 | |
dubs | g0rdy: you're in luck. Rick Clark (dendrobates) is from St Louis | 19:29 |
dubs | and there he is now :) | 19:29 |
*** kashyapc has quit IRC | 19:29 | |
g0rdy | dubs: thanks I just found the wiki people page :) | 19:30 |
MaxR_ | I see mention in the Nova documentation: "Nova Storage manages creating, attaching, detaching, and destroying persistent storage volumes, ala EBS. Currently uses ATA-over-Ethernet." -- Where can I read more information about the Nova Storage EBS specifically? | 19:35 |
*** cw has quit IRC | 19:35 | |
*** rdw has joined #openstack | 19:40 | |
*** BartVB has joined #openstack | 19:42 | |
*** lulinha has joined #openstack | 19:44 | |
*** lulinha has left #openstack | 19:44 | |
MaxR_ | Are there any other locations for information about Nova other than the wiki site? (In terms of documentation) | 19:49 |
creiht | MaxR_: nova.openstack.org | 19:50 |
chromakode | lumberjack looks cool | 19:50 |
chromakode | one of the core tenets of karmabot is that it's ridiculously easy to extend | 19:50 |
MaxR_ | That's where I'm at already lol Perhaps I'm just looking in the wrong place, but there seems to be very little information about the Nova Storage EBS... | 19:51 |
MaxR_ | other than just a mention on one page that it even has an EBS | 19:51 |
creiht | MaxR_: I have to defer to the nova devs | 19:52 |
MaxR_ | Fair nuff.. guess I'll just be patient and wait for one of them to poke their head in here (= | 19:52 |
*** bithooki has left #openstack | 19:57 | |
*** _jcsmith has joined #openstack | 19:57 | |
justinsb | MaxR_: There's little documentation for the Nova EBS. It's called the volume store. Currently it uses ATA over Ethernet. I'm working on getting iSCSI working as well (works great in a branch, but getting anything merged is a battle with Launchpad) | 19:59 |
MaxR_ | Is it redundant or distributed? | 19:59 |
MaxR_ | Not sure how the AOE is implemented | 20:00 |
justinsb | ATA over Ethernet is just a transport | 20:00 |
justinsb | At the most basic level, you're just exposing a disk for use by a remote machine | 20:00 |
justinsb | But that disk could itself be RAIDed or distributed | 20:00 |
MaxR_ | okay.. so if I want redundancy then I need to do it via RAID or some other method before exposing it | 20:01 |
justinsb | But that's currently not part of Nova | 20:01 |
justinsb | MaxR_: Exactly | 20:01 |
*** justinsheehy has quit IRC | 20:01 | |
justinsb | Currently, it's just a way of "attaching a remote disk" | 20:01 |
justinsb | iSCSI is much the same at the moment | 20:01 |
*** sirp2 has quit IRC | 20:01 | |
justinsb | Bret and some others have pointed at a project called SheepDog as looking interesting long term | 20:01 |
redbo | step 1: buy a SAN | 20:02 |
mtaylor | justinsb: what problems are you having getting a branch merged? | 20:02 |
creiht | step 2: buy another SAN to back it up :) | 20:02 |
redbo | haha | 20:02 |
MaxR_ | yeah... :/ Which adds up to a LOT of money | 20:02 |
mtaylor | step 3: hire many many SAN consultants to turn either SAN on | 20:02 |
justinsb | mtaylor: Just feels that launchpad is more work than writing the actual code! | 20:02 |
redbo | it's hard to tell from the sheepdog page if it's block storage or blob storage. | 20:03 |
mtaylor | justinsb: heh. well, certainly interested in working on improving that. keep poking me in the eye as you have pain | 20:03 |
MaxR_ | What I'd love to see is a component for the cloud platform that has the redundancy and configurability of swift's object store, with the convenience of block-level access.. and the ability to just add commodity servers and storage to the cloud to expand the overall available space on the cloud :P | 20:04 |
MaxR_ | I get the feeling like my hope is still way out on the horizon | 20:04 |
redbo | and a cookie? | 20:04 |
MaxR_ | lol I'll settle for cake :) | 20:04 |
MaxR_ | as long as I get to eat it too | 20:04 |
mtaylor | let them eat cake! | 20:04 |
justinsb | mtaylor: Well I think you've set it up well (I presume you've set it up). It's just annoying that e.g. bug reports are separate from merge requests, and the blueprints don't have a good discussion platform built in... Just feels like I have to repeat everything two or three times to keep launchpad happy... | 20:05 |
MaxR_ | But seriously... if there way some way to have block-level storage even with just easy expandability.. that seems like an important piece towards improving virtualization technologies | 20:05 |
redbo | there's ceph, but it's not perfect | 20:05 |
MaxR_ | Step at a time though (= | 20:05 |
justinsb | MaxR_: I think that's what SheepDog is aiming for | 20:05 |
creiht | MaxR_: I think we all want that, a good solution hasn't been identified/built yet | 20:06 |
creiht | at least that I am aware of yet | 20:06 |
mtaylor | justinsb: the blueprints definitely have much room for improvement - I've got a couple of todo-list items for them - specifically hooking them into a discussion thread similar to merge requests would be good | 20:06 |
MaxR_ | Alright, glad to hear I'm not out in left field, or missing some crucial point hehe | 20:06 |
_jcsmith | Is the cloud files NTT deployment annouced today based on swift? does anybody know? | 20:06 |
mtaylor | justinsb: do you have any thoughts on how bug reports and merge requests might be better integrated? | 20:06 |
MaxR_ | I'll check out Ceph and SheepDog though. I appreciate the suggestions | 20:06 |
creiht | _jcsmith: We are curious as well :) | 20:07 |
mtaylor | justinsb: I had a thought the other day on auto-populating a merge request from a blueprint or vice-versa ... but then I didn't write it down and have forgotten what I came up with :( | 20:08 |
_jcsmith | it talks about mounting the filesystem from inside a server, which is interesting | 20:08 |
justinsb | mtaylor: I think the way JIRA does it seems to work. Patches are proposed and discussed within the bug report | 20:08 |
_jcsmith | I saw redbo has a fuse plugin | 20:08 |
MaxR_ | it appears to have some serious limitations though :/ | 20:09 |
justinsb | mtaylor: I don't really grok blueprints yet | 20:09 |
justinsb | mtaylor: But I really don't get why bug reports and merge requests are separate | 20:09 |
justinsb | mtaylor: Seems like the sort of thing which makes sense to a programmer, but no sense in practice | 20:09 |
eday | mtaylor: perhaps a Bugs state that automatically proposes for merge if a branch is linked? | 20:09 |
justinsb | mtaylor: And I say that because it feels like the sort of mistake I usually make as a programmer :-) | 20:10 |
mtaylor | hehe | 20:10 |
mtaylor | justinsb: I think the two are meant to be two different discussions - although I certainly understand what you're saying... | 20:10 |
MaxR_ | I love how on the press release of NTT Cloud Files it says that enterprises can eliminate the capital invetment into teh purchase of a SAN... | 20:10 |
mtaylor | I think the bug discussion is meant to be a discussion of the bug itself, whereas the merge request is meant to be a code review of the solution... combining is an interesting idea I'll have to think about though | 20:11 |
mtaylor | eday: problem is, I often push branches that aren't ready for review... triggering review based on having declared that I fixed a bug would be odd | 20:12 |
mtaylor | justinsb: I'm going to go chat about bugs/merge requests with some folks... | 20:13 |
eday | mtaylor: it may not be automatic when a branch is linked, could still be manual, just on the bugs page | 20:13 |
redbo | _jcsmith: yeah, it does some not great stuff, but it works :) | 20:13 |
eday | mtaylor: or just a button on the bugs to propose for merge directly | 20:13 |
*** ianw_ has joined #openstack | 20:14 | |
*** ianw has quit IRC | 20:14 | |
*** ianw_ is now known as ianw | 20:14 | |
MaxR_ | This Ceph programs looks very cool..and the fact that it made its way into the Linux kernel certainly gives it some clout.. ashame it's only at 0.20.2 :) | 20:15 |
*** _buck has quit IRC | 20:15 | |
eday | mtaylor: I'm actually fine with the current process, but just looking for other shortcuts :) | 20:15 |
mcgurrin | I think some organizations are looking at ceph as well, I know dreamhost is | 20:16 |
*** jfontan has joined #openstack | 20:16 | |
*** thinkscientist has joined #openstack | 20:16 | |
mtaylor | eday: well yeah, I actually really like hearing things that irk people... it will at least get me to reconsider things that I may not have bothered thinking about | 20:16 |
*** mrayzenoss has joined #openstack | 20:16 | |
*** mrayzenoss is now known as mray | 20:16 | |
*** mray has joined #openstack | 20:16 | |
thinkscientist | Hello all | 20:16 |
creiht | howdy | 20:16 |
thinkscientist | Not only do I have irc working on Android | 20:17 |
justinsb | mtaylor: Maybe it's also my fault... maybe I shouldn't be opening bugs, but just pushing the fixes? | 20:17 |
*** theron has joined #openstack | 20:17 | |
thinkscientist | I am on the open stack channel :) | 20:17 |
MaxR_ | hehe <3 Android | 20:17 |
redbo | Ceph looks pretty nice, but there are a few things that scare me. Like it has no topological awareness, so it can assign every copy of a block to the same cabinet. | 20:18 |
mtaylor | justinsb: funny, I was _just_ discussion something similar but in the other direction | 20:19 |
thinkscientist | Soooo guys new to open stack looked at nova site but not too informative | 20:19 |
thinkscientist | Didn't realise there was a wiki gonna go look there | 20:19 |
MaxR_ | ouch, thank you for the heads up on that redbo | 20:19 |
mtaylor | justinsb: I was thinking a little while ago that all branches to be merged should really be related to a bug or a blueprint, but often the overhead there is too much work | 20:19 |
mtaylor | justinsb: so I was wondering if we couldn't auto-populate some intormation and some linkage in one direction or the other | 20:19 |
thinkscientist | Any advice of where to look? | 20:21 |
MaxR_ | thinkscientist: nova.openstack.org is what I was given... | 20:21 |
MaxR_ | But like you said, there's only a limited about of information there | 20:21 |
justinsb | mtaylor: I like the idea that all work is related to a bug or a blueprint (feature) | 20:21 |
MaxR_ | and it appears to still be so new, that there is no external documentation or information available either | 20:21 |
mtaylor | justinsb: yeah - the trick is how to do that without making is _way_ more work for everyone | 20:22 |
MaxR_ | "wild wild west" ;) | 20:22 |
thinkscientist | Lol | 20:22 |
thinkscientist | Has anyone installed | 20:22 |
thinkscientist | Tried house.f | 20:22 |
justinsb | mtaylor: Hmmm... does launchpad have an API? | 20:22 |
thinkscientist | Going hands on ignore last msg | 20:23 |
mtaylor | yup | 20:23 |
mtaylor | justinsb: sorry, that was for you - yup | 20:23 |
justinsb | mtaylor: I'm just wondering if it would be possible to write a quick script that would do things like 'open bug' | 20:23 |
justinsb | mtaylor: Which would do the branch as well | 20:23 |
mtaylor | justinsb: absolutely. actually, there are already bzr plugins which do launchpadlib operations | 20:23 |
justinsb | mtaylor: And then make the merge proposal super-simple (templated), and auto-linked to the bug | 20:23 |
justinsb | mtaylor: Ah ... so it's a case of me being an incompetent workman and blaming my tools! | 20:24 |
mtaylor | justinsb: well, there is a new command in bzr 2.2 (which originally came from the pipeline plugin) | 20:24 |
mtaylor | justinsb: which is bzr lp-submit - which will make the merge prop for you from your current branch | 20:24 |
mtaylor | justinsb: making one similar that links in bug meta info or whatnot shouldn't be too hard | 20:24 |
MaxR_ | Is there a list anywhere that shows the expected differences between the swift code that is available now as a developer preview, versus what will be available on the first non-developer release? | 20:24 |
justinsb | mtaylor: I'll check it out... thanks! | 20:25 |
mtaylor | justinsb: sure thing! | 20:26 |
creiht | btw... there is also wiki.openstack.org thathas some info as well | 20:26 |
creiht | MaxR_: We still need to work on that | 20:26 |
mtaylor | justinsb: and definitely keep poking me about workflow improvements | 20:26 |
creiht | MaxR_: I expect the main differences will be packaging, documentation, and whatever features nova needs for their first release :) | 20:27 |
MaxR_ | Nova used to be called Nebula... is OpenNebula the same thing? | 20:28 |
creiht | nice: hah... the nova readme is quite awesome | 20:28 |
creiht | MaxR_: Different things | 20:28 |
thinkscientist | Hmmm what platforms are you guys installing nova under | 20:28 |
*** cw has joined #openstack | 20:29 | |
thinkscientist | Brb | 20:31 |
dendrobates | thinkscientist: debian and ubuntu for now, but the fedora team is working in it | 20:31 |
*** thinkscientist has quit IRC | 20:31 | |
*** miclorb has joined #openstack | 20:31 | |
*** justinsheehy has joined #openstack | 20:33 | |
mray | anyone else notice that when you hit the Search box on wiki.openstack.org it redirects automatically? | 20:33 |
MaxR_ | lol I thought IW as going crazy | 20:34 |
MaxR_ | I kept clicking in it and hitting A and the page would submit | 20:34 |
creiht | mtaylor: Who is in charge of the wiki? | 20:34 |
creiht | mray: Yeah it has been reported several times :/ | 20:34 |
MaxR_ | I just chalked it up to browser issue as I eventually was able to do a search | 20:34 |
mray | creiht: thanks | 20:34 |
dendrobates | hmm, soren or I will look into it. | 20:35 |
dendrobates | we need a wikimaster\ | 20:35 |
*** PiotrSikora has quit IRC | 20:36 | |
*** thinkscientist has joined #openstack | 20:39 | |
mtaylor | dendrobates: feel free to dump that on my plate ... although I don't have root on it right now | 20:40 |
thinkscientist | Back! | 20:40 |
thinkscientist | Just couldn't hack the irc on such a small screen | 20:41 |
lbieber | dendrobates: Todd Morey has been maintaining the wiki but as Monty noted he can help with that also | 20:41 |
thinkscientist | Lol one of those f | 20:41 |
thinkscientist | Great ideas that sucked in implementation | 20:41 |
dendrobates | lbieber: I don't think todd has access to the box | 20:41 |
thinkscientist | Anyone feel that way about open stack so far | 20:41 |
lbieber | dendrobates: hmmm, I could be wrong then | 20:42 |
*** pvo has quit IRC | 20:42 | |
dendrobates | soren, ant, pvo and me maintain it. | 20:42 |
MaxR_ | thinkscientist: I think OpenStack is the future of cloud computing, personally. :) | 20:42 |
dendrobates | I am far from a moin expert though | 20:42 |
_jcsmith | any thoughts on some methods to performance benchmark a swift cluster? API calls, uploads/downloads? I figure ab or siege for the actual objects, what about the api? I could write something, but maybe there is something already out there? | 20:43 |
*** littleidea_ has quit IRC | 20:43 | |
creiht | _jcsmith: I have a tool that can help with that... I need to refactor it a bit and release in swift | 20:44 |
MaxR_ | throughput of upload/download | 20:44 |
creiht | apache ab can be difficult to get auth to work with it | 20:44 |
creiht | Though you could disable auth | 20:44 |
thinkscientist | I don't know eucalyptus has a headstart | 20:44 |
_jcsmith | that reminds me, auth is another part to benchmark | 20:45 |
creiht | _jcsmith: just a reminder that the auth in swift is for dev purposes only :) | 20:45 |
MaxR_ | thinkscientist: If you want quota management in Eucalyptus, you'll need to pay for the software.. heh | 20:45 |
*** jonesy has quit IRC | 20:45 | |
*** jonesy_ is now known as jonesy | 20:45 | |
_jcsmith | creiht: I've read that, is that because it's sqllite? I've added a few users. I'm sure that scaling to 1000s of users and handing high auth volume would break at some point. Doesn't the memcache layer help the auth stuff too? what do you think would break first in the current implementation? | 20:46 |
creiht | _jcsmith: Well the passwords are in clear text :) | 20:47 |
MaxR_ | heh.. | 20:48 |
creiht | _jcsmith: Under benchmarking, it should be fine since the auth is cached in memcache | 20:48 |
_jcsmith | true, I did notice that, but I see it grabbing the key and comparing it, obviously not much extra code to hash and compare hashes | 20:48 |
*** avsm has joined #openstack | 20:48 | |
creiht | There also isn't any security on things like adding accounts | 20:49 |
creiht | It is the bare minimum needed for the system to still work | 20:50 |
*** littleidea_ has joined #openstack | 20:50 | |
thinkscientist | anyone tried authenticating against ad | 20:50 |
creiht | It also mimics the auth that the rackspace cloud uses | 20:50 |
_jcsmith | I see, that makes sense. | 20:50 |
*** jfontan has left #openstack | 20:50 | |
_jcsmith | thats a good question, is anybody working on an ldap plugin? | 20:50 |
thinkscientist | For any large organization that will need to be possible even if it piggybacks on ldap | 20:50 |
creiht | We figured that most people would already have an auth system (like AD), and would prefer to plug into it | 20:50 |
_jcsmith | I hate LDAP but customers always seem to use it | 20:50 |
creiht | In swift auth is wsgi middleware, so it would just require writing the middleware piece | 20:51 |
creiht | and then configuring swift to use the middleware | 20:51 |
*** ianw_ has joined #openstack | 20:51 | |
thinkscientist | Ahhh | 20:51 |
thinkscientist | Need to go understand how swift fits in | 20:52 |
creiht | thinkscientist: Swift is an object store, similar in some respects to amazon's S3 | 20:52 |
*** ianw has quit IRC | 20:53 | |
*** ianw_ is now known as ianw | 20:53 | |
thinkscientist | O.O so that won't help too much with linking into enterprise | 20:53 |
MaxR_ | I still find it amazingly ambigous speaking about storing things "on the cloud"... that can be interpreted so many ways | 20:53 |
* jaypipes wonders how married folks are to gflags... | 20:53 | |
*** vish1 has joined #openstack | 20:54 | |
vish1 | whoot | 20:54 |
thinkscientist | MaxR_: Agreed | 20:54 |
_jcsmith | im a little ignorant of WSGI, is any auth WSGI middleware going to work? or do I have to modify swift code to callback to the auth WSGI module? | 20:54 |
*** littleidea_ has quit IRC | 20:54 | |
thinkscientist | I think I need to go do the install in a vm and then I will come back | 20:55 |
creiht | _jcsmith: WSGI is a specification of an interface for how python web apps should interact with each other | 20:56 |
soren | dendrobates: It's a by-product to the way... err.. Chris, I think, made the logo link to the homepage. | 20:56 |
creiht | An oversimplified way of thinking of it is that it is kinda like unix pipes | 20:56 |
*** gnrfan has quit IRC | 20:56 | |
_0x44 | soren: I didn't make anything link to the homepage | 20:57 |
soren | _0x44: The other Chris. | 20:57 |
_0x44 | Too many Chrises. | 20:57 |
_0x44 | :) | 20:57 |
vish1 | ...on the dance floor | 20:57 |
dubs | vish1: nice | 20:58 |
*** allsystemsarego has quit IRC | 20:58 | |
creiht | so the request gets piped to auth which pipes to the swift app | 20:59 |
*** almaisan is now known as almaisan-away | 20:59 | |
*** gnrfan has joined #openstack | 20:59 | |
*** rnewson has joined #openstack | 20:59 | |
*** terje has quit IRC | 21:00 | |
_0x44 | vish1: So you saw the video from the wedding I attended this weekend? | 21:00 |
_0x44 | ;) | 21:00 |
_jcsmith | creiht: thanks, give me a good place to start | 21:00 |
*** terje has joined #openstack | 21:00 | |
_jcsmith | give/gave | 21:00 |
*** jtimberman has quit IRC | 21:00 | |
creiht | In theory, other wsgi auth middleware should be easy to integrate with swift, but I haven't tried personally | 21:00 |
soren | \o/ | 21:00 |
* soren has unbroken Nova :) | 21:01 | |
*** jtimberman has joined #openstack | 21:01 | |
*** jtimberman has joined #openstack | 21:01 | |
* soren cleans up and commits | 21:01 | |
*** justinsheehy has quit IRC | 21:01 | |
*** justinsheehy has joined #openstack | 21:02 | |
creiht | _jcsmith: If you are going to do some perf testing, let me know so that we can go through some things to make sure swift is set up optimally for the cluster | 21:02 |
creiht | http://swift.openstack.org/deployment_guide.html | 21:02 |
creiht | covers a lot of it actually | 21:03 |
_jcsmith | we will be setting up a virtual environment, then onto bare metal, measuring along at different phases of the project | 21:03 |
creiht | cool | 21:03 |
_jcsmith | I saw that guide, very helpful | 21:03 |
*** gnrfan has quit IRC | 21:03 | |
*** sirp2 has joined #openstack | 21:03 | |
creiht | _jcsmith: Well if you run into isssues, let me know | 21:03 |
_jcsmith | do you have a preference for load balancers in front of the proxy, haproxy, zeus, lvs? | 21:04 |
creiht | _jcsmith: Are you going to be using 10g or 1g networking? | 21:04 |
creiht | and do you require https? | 21:05 |
_jcsmith | starting with 1g for alpha, possibly 10g on release | 21:05 |
_jcsmith | yeah, https | 21:05 |
redbo | mtaylor: are the other swift packages not being built and added to the PPA? | 21:06 |
creiht | We terminate ssl at the proxy layer, and most of the load balancers that support ssl can saturate 1g pretty well | 21:06 |
creiht | I think pound did pretty good | 21:06 |
creiht | We like haproxy a lot, but doesn't have ssl support | 21:06 |
creiht | Once you bump up to 10g, zeus is much more efficient (at least in our testing so far) at utilizing 10g | 21:07 |
*** MaxR_ has quit IRC | 21:07 | |
creiht | We are still doing more testing on that side of things though, trying to find a better solution | 21:07 |
g0rdy | +creiht: F5 maybe ? | 21:07 |
mcgurrin | I know it is not mainly for load balencing but has anyone tried the load balencing built into astaro's ASG? | 21:08 |
* creiht is looking forward to seeing openssl performance on intel chips with AES accel. | 21:08 | |
_jcsmith | when I worked on CDN stuff we did a lot of stuff with foundry and lvs, but we werent doing SSL | 21:08 |
_jcsmith | goo points | 21:08 |
_jcsmith | good | 21:08 |
creiht | g0rdy: We didn't test F5's mainly due to price, but I imagine they would do :) | 21:08 |
creiht | ssl really throws a wrench into everything | 21:09 |
*** ubuntu has joined #openstack | 21:09 | |
*** ubuntu is now known as vishvananda | 21:09 | |
creiht | and I imagine most hardware based LB solutions could work well, as long as the cost doesn't bother you | 21:09 |
mcgurrin | I think astaro can do SSL or work on the TCP level and have the application do the SSL, would that help? | 21:09 |
vishvananda | ircfail :( | 21:10 |
* creiht looks up astaro | 21:10 | |
creiht | mcgurrin: not sure | 21:11 |
creiht | mcgurrin: they seem to be more focused on firewalls | 21:11 |
redbo | haproxy can do tcp-only forwarding, but then you have to deal with the fact that you don't get x-forwarded-for headers. | 21:11 |
*** armmer has quit IRC | 21:11 | |
*** armmer has joined #openstack | 21:12 | |
mcgurrin | I iuse one and they do have a server load balancing tool built in | 21:12 |
mcgurrin | if you want to try it get the free home licence or a free trial | 21:12 |
mcgurrin | they have VM based solutions or phisical or special hardware | 21:12 |
creiht | mcgurrin: My initial guess is that at 1g networking, it might work, but I'm initially sceptical of 10g | 21:12 |
mcgurrin | I'm not sure that they have 10g support in the hardware models but they do use link aggressages or your own hardware so I'm not sure on that | 21:13 |
creiht | Of course if someone wants to try it out, I'm always curious to hear how it goes :) | 21:13 |
creiht | turns out that saturating 10g with SSL traffic is difficult | 21:14 |
mcgurrin | anyone know what a Gigabit SFP Ports is? | 21:14 |
creiht | And if you want to do it in hardware, it gets pretty expensive | 21:14 |
mcgurrin | yeah, you would want the software appliance I assume | 21:15 |
mcgurrin | one cool thing with the astaro's though is that they themselves can be clustered to handle more load | 21:16 |
redbo | unless the new westmere chips turn out to be awesome | 21:17 |
creiht | :) | 21:17 |
*** justinsheehy has quit IRC | 21:17 | |
*** jaypipes has quit IRC | 21:17 | |
redbo | "clustered to handle more load"? That would take some explanation. | 21:18 |
*** jbryce has left #openstack | 21:18 | |
mcgurrin | you can cluster ASGs to split the processing and such so if you were dealing with decrypting SSL at the load balancer you would have more processing power | 21:19 |
pandemicsyn | mcgurrin: Gigabit SFP is probably a 1G fiber port | 21:19 |
mcgurrin | ah | 21:19 |
mcgurrin | the clustering also provides High availability | 21:19 |
mcgurrin | so you have less chance of going down | 21:20 |
*** vish1 has quit IRC | 21:20 | |
thinkscientist | It's a small form factor pluggable port | 21:21 |
thinkscientist | Just a kind of connector for connectivity (physical) | 21:21 |
thinkscientist | I don't agree with the whole hierarchical architecture | 21:22 |
thinkscientist | I think we need to just get away from hierarchies | 21:22 |
creiht | thinkscientist: hierarchical architecture? | 21:22 |
thinkscientist | I think peer to peer is the way forward for a distributed architecture | 21:22 |
*** sirp2 has quit IRC | 21:23 | |
creiht | _jcsmith: btw, the proxies can be configured to terminate ssl, if you want to do that, but that will of course slow the proxy down a bit | 21:23 |
*** theron has quit IRC | 21:24 | |
pandemicsyn | g0rdy: F5's are cool put expensive - a BigIP 8800 still only does 6Gbps of SSL throughput but for *alot* more $$$ (was just scrolling back catching up channel) | 21:24 |
creiht | pandemicsyn++ | 21:25 |
*** theron has joined #openstack | 21:25 | |
thinkscientist | creiht: since we're talking about the clusters I thought I'd add my 2c about the arch | 21:25 |
creiht | :) | 21:25 |
creiht | thinkscientist: When we began the design of swift, we gave some thought to some of the p2p architectures | 21:26 |
creiht | the problem is that it adds a lot of complexity to the system | 21:26 |
thinkscientist | It does but in the end...it's the way forward | 21:26 |
*** sirp2 has joined #openstack | 21:26 | |
creiht | We wanted to see if we could solve it using simpler concepts | 21:26 |
thinkscientist | Recommend you visit itunes and watch a lecture in the computer systems colloquium series | 21:26 |
thinkscientist | It's called rethinking time | 21:27 |
creiht | thinkscientist: that may very well be, and if some of those concepts make it into swift that's great... patches welcome :) | 21:27 |
*** miclorb has quit IRC | 21:27 | |
thinkscientist | That's a rewrite | 21:27 |
thinkscientist | Lol | 21:27 |
* creiht was joking a bit | 21:27 | |
thinkscientist | Me too | 21:28 |
thinkscientist | I think my thoughts are more applicable to nova than to swift itself | 21:28 |
eday | thinkscientist: we've been thinking about this quite a bit, the are pros and cons to each arch | 21:29 |
eday | (in nova context) | 21:29 |
redbo | There are some problems we could solve with a mesh network or gossip. Mostly sharing failures around. But swift is really pretty simple. | 21:29 |
thinkscientist | Cons for | 21:29 |
thinkscientist | Peer to peer | 21:29 |
creiht | #1: complexity :) | 21:29 |
thinkscientist | I mean even in respect of swift peer to peer is still great, definitely looking at global scaling it makes sense | 21:30 |
*** ianw has quit IRC | 21:30 | |
thinkscientist | We gotta find a way to overcome the complexities, what things do you think make it complex? | 21:31 |
redbo | man.. there are peers, they talk to each other. | 21:31 |
antonym | dendrobates: jk0 fixed the header on the wiki, search should work now | 21:31 |
*** jaypipes has joined #openstack | 21:31 | |
soren | vishvananda: In http://bazaar.launchpad.net/~hudson-openstack/nova/trunk/revision/150 you added a nova-dhcpbridge.conf to debian/nova-api.install, but you forgot to add the actual nova-dhcpbridge.conf. Can you fix that please? My packages are very unhappy with me :) | 21:32 |
thinkscientist | soren: Lol | 21:32 |
eday | thinkscientist: complexities of keeping things in sync, advertising resource availability to it's peers | 21:33 |
eday | thinkscientist: also most likely different types of peering relationships (read-only, provision, ...). with a hierarchical nature, you're looking at one logical path and events get bubbled up as needed. | 21:34 |
eday | thinkscientist: having said all that, it may still be worth it :) | 21:35 |
*** jsmith is now known as jsmith-away | 21:35 | |
soren | eday: Did you ever work out what was wrong with that echoserver test thing? | 21:36 |
vishvananda | soren: sure | 21:36 |
eday | soren: my tornado vs twisted vs c++ load test? not yet, was poking at it this weekend for a few minutes | 21:36 |
eday | soren: i should fix that and get some #s today | 21:37 |
*** sirp2 has quit IRC | 21:38 | |
soren | eday: Excellent. | 21:38 |
*** malthe has quit IRC | 21:38 | |
soren | vishvananda: Ta. | 21:38 |
thinkscientist | But in an d | 21:38 |
thinkscientist | Ideal world the babdwidth won't be a problem | 21:38 |
thinkscientist | and it would mean a proper cloud, | 21:38 |
thinkscientist | Cloud and hierarchy just doesn't fit with me | 21:38 |
thinkscientist | Shouldn't everything be everywhere? | 21:39 |
thinkscientist | From a dr perspective it would be great | 21:39 |
mtaylor | eday: is the c++ part of your load test using scalestack? | 21:39 |
eday | mtaylor: yeah, the client and server reference are scalestack | 21:39 |
mtaylor | cool | 21:39 |
thinkscientist | Definitely make sense if you were a large org | 21:39 |
vishvananda | nnnn | 21:39 |
thinkscientist | You could then have a RAIC | 21:39 |
vishvananda | s | 21:40 |
thinkscientist | Redundant Array of Independent Clouds | 21:40 |
eday | vishvananda: in your daemon-refactor branch, any reason you choose nova/compute/node.py => nova/compute/computeservice.py instead of compute/service.py? compute/computeservice seems redundant :) | 21:40 |
thinkscientist | And since it's peer to peer downtime of one cloud shouldn't cause any issues | 21:40 |
eday | vishvananda: rework looks really good otherwise though :) | 21:40 |
eday | thinkscientist: you can accomplish the same reliability with a properly distributed hierarchical service as well. ie: DNS | 21:42 |
thinkscientist | so when you make dns changes depending on where you are you don't get hit with propagation times? | 21:42 |
soren | mtaylor: You mentioned there was a command line tool to request merges, didn't you? | 21:42 |
vishvananda | it | 21:43 |
gustavomzw | eday: there is a p2p arch proposal in http://wiki.openstack.org/Overview, is it in the plans? | 21:43 |
eday | soren: bzr pipelines have that feature | 21:43 |
vishvananda | soren: connection problems, the file is there but named nova-dhcp.conf insteadd | 21:43 |
soren | vishvananda: Oh, yes, there it is. | 21:44 |
eday | gustavomzw: there are both hier. and p2p proposals, neither has really been decided for the long term yet, but leaning towards the hier. one. I'm going to be putting some things up on the mailing list to discuss this soon | 21:45 |
soren | vishvananda: No rush, just push a fix whenever you've got your intertubes fixed. | 21:45 |
mtaylor | soren: yes. | 21:45 |
gustavomzw | eday: great | 21:45 |
*** greenisus has quit IRC | 21:45 | |
mtaylor | soren: also, it's in mainline if you upgrade to 2.2beta | 21:45 |
soren | eday: bzr pipelines sounds like something I should know about already. | 21:45 |
mtaylor | soren: bzr lp-submit | 21:45 |
mtaylor | soren: but also, you should learn all about pipelines anyway | 21:46 |
*** littleidea has joined #openstack | 21:46 | |
* mtaylor needs to start sending out bzr tips and tricks | 21:46 | |
soren | mtaylor: It looks remarkably similar to looms? | 21:46 |
eday | thinkscientist: we have the luxury of choosing our cache eviction policies, and choosing to do active eviction/updates, not TTL based | 21:46 |
mtaylor | soren: yes. except it works :) | 21:47 |
soren | mtaylor: Wicked! | 21:47 |
mtaylor | soren: stewart uses it for everything he does in drizzle | 21:47 |
thinkscientist | eday: Is that documented anywhere? | 21:47 |
eday | thinkscientist: nope, just in discussion, like now :) | 21:47 |
stewart | soren, it has downsides as well. useful for constructing a series of things that cannot be pulled apart. | 21:48 |
thinkscientist | Do these chats get logged? | 21:48 |
soren | mtaylor: I think pipelines would please the git crowd. | 21:48 |
mtaylor | soren: also, if you use it and lp-submit, it will understand that one branch depends on the other and fill that out in the merge request... so each merge request will only just now have the relevant diff | 21:48 |
eday | soren: yeah, pipelines are pretty awesome to deal with your dependencies | 21:48 |
mtaylor | soren: indeed. | 21:48 |
stewart | soren, but if a patch is independent, it's a bad idea to have it in a pipeline. | 21:48 |
thinkscientist | eday: Also looking for mailing list | 21:48 |
eday | thinkscientist: this IRC channel is now | 21:48 |
soren | mtaylor: They're very particular about their disk usage, apparantly. Not having a separate working tree for stuff seems like an easy sell. | 21:48 |
mtaylor | soren: yes. listen to the wisdom of the stewart | 21:48 |
stewart | soren, mainly because you cannot pull anything out but the first pipe. | 21:48 |
stewart | soren, i.e. it does not match the feature set of quilt. but then again, few things do. | 21:49 |
mtaylor | stewart: have you looked at bzr-colocated? | 21:49 |
stewart | mtaylor, no. | 21:49 |
soren | mtaylor: The dependency handling sounds awesome. | 21:49 |
soren | mtaylor: I haven't really needed it yet, but I'm sure it'll be very handy at some point. | 21:49 |
stewart | mtaylor, partly because what i do not need more of is bzr plugins | 21:49 |
mtaylor | soren: yes. I love it some times | 21:50 |
mtaylor | I do not use it as heavily as stewart | 21:50 |
mtaylor | but then, I don't think anyone does :) | 21:50 |
soren | mtaylor: At the rate people are submitting stuff to Nova, I suspect "at some point" may be very soon. :) | 21:50 |
mtaylor | soren: :) | 21:51 |
thinkscientist | Do I need to sign a cla to be on the mailing list? | 21:51 |
eday | thinkscientist: no | 21:52 |
thinkscientist | Where can I sign up | 21:52 |
thinkscientist | Looking but can't see it? | 21:52 |
soren | thinkscientist: Join the team on Launchpad and subscribe to its mailing list. | 21:52 |
eday | mtaylor: whats the official mailing list now? are we using both swift and nova? or openstack? or openstack-discuss? :) | 21:53 |
soren | I thought we got rid of .*-discuss$. | 21:53 |
vishvananda | soren: https://code.launchpad.net/~vishvananda/nova/dhcplease-flags/+merge/30997 | 21:54 |
eday | soren: I'm still subscribed to it, hmm | 21:55 |
mtaylor | eday: yes | 21:56 |
mtaylor | eday: well, not openstack-discuss | 21:56 |
thinkscientist | Is there just a bug mailing list | 21:56 |
*** metoikos has quit IRC | 21:56 | |
eday | mtaylor: what are the -core lists for? | 21:57 |
mtaylor | eday: use openstack@ or nova@ - we found/caused a bug in launchpad that is preventing swift@ from being created right now - it's in work | 21:57 |
mtaylor | eday: they're just there, really - in case core team wants to talk about anything - I doubt they'll be useful for much | 21:57 |
mtaylor | thinkscientist: https://edge.launchpad.net/nova/+subscribe | 21:57 |
mtaylor | thinkscientist: and https://edge.launchpad.net/swift/+subscribe | 21:57 |
thinkscientist | Sorry Monty :( | 21:58 |
*** phildoc has quit IRC | 21:58 | |
mtaylor | thinkscientist: it's all good | 21:58 |
eday | mtaylor: ok. would you be opposed to removing -core lists? I'm thinking it will just create confusion :) can always recreate if we need them | 21:59 |
mtaylor | eday: I'm not opposed to it, no | 21:59 |
*** littleidea has quit IRC | 22:00 | |
*** viewlogic has joined #openstack | 22:01 | |
mtaylor | eday: done | 22:01 |
thinkscientist | Good night folk :) | 22:01 |
eday | thinkscientist: gnight! | 22:02 |
eday | mtaylor: awesome | 22:02 |
*** thinkscientist has quit IRC | 22:02 | |
eday | mtaylor: so is openstack-discuss dead for all uses now? might want to blow that away too | 22:02 |
*** henriquetft has quit IRC | 22:03 | |
mtaylor | eday: yup. I'm sending an email out about that right now | 22:03 |
mtaylor | eday: although it's private, so nobody will really seeit | 22:03 |
soren | justinsb: Sorry about the delay. I've just pushed a branch (and a merge proposal) that should fix the S3 handler, making image fetching actually work, check authorization headers properly, and makes the compute nodes pass proper auth headers. | 22:04 |
justinsb | soren: Awesome! I'll have to look & learn from how you did it all... | 22:08 |
eday | mtaylor: ~openstack-planet, ~*-core, and openstack-devel is owned by you, not openstack-admins, btw :) | 22:08 |
mtaylor | eday: thanks. I will fix | 22:08 |
mtaylor | someone should review lp:~vishvananda/nova/clean-auth ... https://code.edge.launchpad.net/~vishvananda/nova/clean-auth/+merge/30431 | 22:09 |
eday | mtaylor: (was just updating the branding and found I could not on those) | 22:09 |
*** mattf has quit IRC | 22:09 | |
soren | justinsb: I've also worked out how to do unit tests for the S3 interface as well as where we can intercept exceptions so that we're sure to catch them no matter where they arise. | 22:10 |
mtaylor | soren, justinsb: I tried running run_test.py through coverage and it breaks things | 22:10 |
mtaylor | soren: just try "coverage run run_tests.py" and you'll see what I mean... | 22:11 |
soren | mtaylor: That's near the top of my list of things to ponder. | 22:11 |
mtaylor | col | 22:11 |
mtaylor | cool | 22:11 |
mtaylor | soren: when it's fixed, I've got the hudson job for it added | 22:11 |
soren | mtaylor: coverage: command not found | 22:11 |
soren | mtaylor: Yeah, that doesn't work at all! :) | 22:11 |
mtaylor | soren: hehe | 22:12 |
mtaylor | soren: I seem to have easy_installed it on the hudson box | 22:12 |
*** tjyang has joined #openstack | 22:12 | |
mtaylor | soren: we should package that | 22:12 |
soren | mtaylor: python-coverage. It's already there. | 22:13 |
termie | so, i know this thing says" to merge this branch..." but i thought the hudson was in charge? | 22:15 |
soren | termie: It is. | 22:15 |
soren | termie: So don't worry about it. | 22:15 |
soren | :) | 22:15 |
soren | Magic will happen. | 22:15 |
*** theron has quit IRC | 22:16 | |
termie | but, for example, prior to today this branch was already in an 'approved' state but monty had said needs fixins (GRAVY) | 22:16 |
termie | and i worry that it could have somehow been merged | 22:16 |
termie | before i got a change to address the fixins (BACON) | 22:16 |
soren | mtaylor: The failures seem limited to stuff that depends on flags. | 22:16 |
*** z0 has quit IRC | 22:16 | |
soren | termie: Which merge proposal is this? | 22:16 |
termie | https://code.launchpad.net/~termie/nova/testing_dependencies/+merge/30738 | 22:17 |
termie | it had 2 approves and one needs fixins (SOUR CREAM) | 22:17 |
termie | and now mtaylor has also approved it | 22:17 |
termie | but prior to that it claimed its state was Approved | 22:17 |
mtaylor | it was | 22:17 |
termie | and so i worry that perhaps it might have been merged without me ever noticing | 22:17 |
soren | termie: It says: Approved revision: 155. | 22:17 |
soren | termie: ...so that last revision is still outstanding. | 22:18 |
mtaylor | ah. bleh. sorry - tarmac thing we were working on earlier... | 22:18 |
termie | soren: but the top part says "status approved" | 22:18 |
mtaylor | somebody pop its status to work in progress and then back to approved | 22:18 |
termie | ah okay, so this isn't the usual way? | 22:18 |
mtaylor | yeah - there's a long argument going on about how to deal with an element of this ... long story short is: | 22:18 |
soren | termie: Yes. and right below it, it says Approved revision: 155. :) | 22:18 |
mtaylor | tarmac merges branches as of the revision when the state was changed to approved | 22:19 |
termie | quantum approval state! | 22:19 |
mtaylor | BUT - lp doesn't consider a prop merged until it's merged all the way | 22:19 |
termie | it seems silly to merge in the bad revision :/ | 22:19 |
termie | i guess i can't have cake and eat the batter also | 22:20 |
mtaylor | so we're adding code to tarmac to write a note back to the merge prop in this case and say "hey, there's new revs that didn't get approved, somebody do something, mkay?" | 22:20 |
mtaylor | and then we're going to continue arguing about how to best handle this in general :) | 22:20 |
soren | mtaylor: You can by the way link people to http://launchpad.net/people/+me/ to get them to go to their own page. | 22:20 |
*** avsm has quit IRC | 22:20 | |
termie | i'm a big fan of the "squash" option in git, ftr | 22:20 |
soren | (re your recent e-mail about lists) | 22:20 |
termie | i think having a single commit that represents the change being reviewed is ideal | 22:21 |
mtaylor | soren: oh, excellent | 22:21 |
soren | termie: That's what the merge proposal is, really. | 22:21 |
mtaylor | termie: yeah - when it hits trunk, it will be a single commit | 22:21 |
* termie blinks | 22:21 | |
soren | termie: When I submit a merge proposal, I don't expect anyone to review the individual revisions. They are just my work-in-progress. I expect them to review the final diff. | 22:21 |
termie | soren: agreed, it just doesn't look like that is what would happen from here | 22:22 |
soren | termie: I don't mind people seeing how I did it wrong 20 times before getting things right :) | 22:22 |
soren | termie: How so? You just see the big diff at the bottom, not the indivial revisions. | 22:22 |
termie | soren: also not a probelm, i am more interested in a clean mainline history, which this sounds like it will do | 22:22 |
mtaylor | termie: oh - so the reason your prop didn't get merged when it was approved before is that it didn't have a commit message associated with it... currently tarmac requires a prop be both Approved and have a commit message | 22:23 |
termie | soren: but does not look like it since it shows multiple commits along with the mainline ones | 22:23 |
termie | although this one is now all confused looking since it is half-merged | 22:24 |
termie | this conversation has goten very scattered, lemme regroup | 22:24 |
termie | my usual workflow is: branch from trunk, work on a feature in a branch, get that branch reviewed, when it is finished merge that into trunk as a single commit | 22:25 |
soren | termie: Your individual commits are still available in the history. They just aren't shown by default. bzr is very particular about history. It never discards any. | 22:25 |
termie | is that what will occur here? | 22:25 |
mtaylor | eday: done | 22:26 |
mtaylor | termie: yes | 22:26 |
soren | termie: Yes and no. | 22:26 |
soren | termie: If you just look at "bzr log", it'll just show the single commit. | 22:26 |
soren | termie: i.e. the merge commit. | 22:27 |
soren | termie: ..but "bzr log -n0" will show every single commit. | 22:27 |
termie | (what would the commit message for that commit be?) | 22:27 |
soren | termie: The one set in the merge proposal. | 22:27 |
termie | soren: great, that explains it perfectly | 22:27 |
mtaylor | termie: the commit message will be whatever commit message you set on the merge prop | 22:27 |
mtaylor | soren types faster than me | 22:27 |
soren | termie: If you have bzr-gtk installed, try going into one of the nova working trees you have and do a "bzr visualize". | 22:27 |
*** justinsheehy has joined #openstack | 22:28 | |
soren | termie: You can see the whole ancestry graph in all its glory. | 22:28 |
* mtaylor uses almost none of bzr-gtk - although I really like bzr gannotate when I'm trying to trackdown where the heck something came from | 22:28 | |
termie | righto | 22:28 |
soren | mtaylor: I only use visualize and only for giggles :) | 22:28 |
chilts | yay, there are Git repos on GitHub :) | 22:29 |
termie | familiar with those bits from the git world as well | 22:29 |
mtaylor | sigh | 22:29 |
termie | so, should i be changing my 'commit message' to be more useful in this merge, i assume the description is only for the reviewers? | 22:29 |
mtaylor | yes | 22:29 |
soren | Precisely. | 22:29 |
termie | also, i just this second realzied that little icon is a pencil for 'edit' not an exclamation mark for "there's a problem here" | 22:30 |
mtaylor | termie: and if you use bzr lp-submit ... you can do it all in one go: bzr lp-submit -m"Commit Message" ... and then it'll pop up and editor where you can put in the description | 22:30 |
soren | In case stuff gets approved straight away, they're often very similar or exactly the same. If it goes through multiple revisions, what was in the description may no longer be descriptive for the changes at all. hence the seperation. | 22:30 |
* mtaylor has disagreed with the edit icon for quite some time now | 22:30 | |
termie | mtaylor: i'll consider adding that to git-bzr-ng ;) | 22:30 |
mtaylor | termie: hehe | 22:31 |
* mtaylor is tempted to say he doesn't grok why you'd want such a thing - but since he uses bzr to access other people's non-bzr code, he guesses he actually _does_ understand :) | 22:31 | |
* soren calls it a day | 22:32 | |
termie | mtaylor: i think for git people once you get used to working with the index you feel like your hands are tied anywhere else | 22:32 |
termie | soren: gnight timezone friend | 22:33 |
soren | See you tomorrow, guys! | 22:33 |
soren | termie: :) | 22:33 |
mtaylor | night soren ! | 22:33 |
termie | mtaylor: not that i can actually work with the index much with git-bzr-ng stuff, but i can atleast checkout differnet branches in the same directory | 22:33 |
termie | mtaylor: that's really the pain that made me write the thing | 22:33 |
termie | i have git setup in my prompt so that it shows which branch i am in, and that effectively adds a third dimension to the file system | 22:34 |
mtaylor | termie: yeah - it's funny, because I keep trying to do git clones into different dirs and there just isn't a useful workflow fo rit | 22:34 |
mtaylor | termie: so that's consistently one of my least favorite things about git, and yet one of the features that people from git miss the most :) | 22:35 |
termie | it ends up being "git checkout branch" vs cd ../../../branch/path/to/dir | 22:35 |
mtaylor | just goes to show - different strokes for different folks | 22:35 |
creiht | mtaylor: I never thought I would like everything in the same dir, until I began using it | 22:35 |
mtaylor | I keep meaning to poke at the bzr-colo plugin for doing colocated branches | 22:35 |
termie | which leads me to believe bzr natives have scripts that dig back to the root repo and then back up a new branch | 22:35 |
mtaylor | but I do suggest you guys check out bzr-pipelines at some point | 22:36 |
mtaylor | for those who haven't jumped ship to termie's git-bzr-ng yet | 22:36 |
termie | mtaylor: it works really well! | 22:36 |
termie | mtaylor: :D just don't use rebase to delete commits :-O | 22:37 |
termie | but that's common courtesy anyway | 22:37 |
* termie reads about bzr-pipelines | 22:37 | |
mtaylor | termie: hehehe. yeah. that will... um... be quite evil :) | 22:37 |
mtaylor | http://doc.bazaar.canonical.com/plugins/en/pipeline-plugin.html | 22:38 |
termie | so it seems like, using non-dvcs-specific terms, a way of stacking diffs into a group? | 22:38 |
mtaylor | yeah | 22:38 |
mtaylor | it's similar to the quilt approach of stacked branches | 22:39 |
mtaylor | but it's all colocated - so if you're working on related and interdependent things, it's pretty stellar | 22:39 |
mtaylor | you know, if you're in to that sort of thing | 22:39 |
termie | ;) | 22:39 |
termie | yeah it seems pretty similar to my workflow with git | 22:40 |
_0x44 | mtaylor: Can you shelve stuff in one pipeline and unshelve in another one? | 22:40 |
termie | it looks like it helps in a couple nice ways though | 22:40 |
mtaylor | _0x44: yes | 22:40 |
*** littleidea_ has joined #openstack | 22:40 | |
mtaylor | shelve and unshelve make me a happy hacker | 22:40 |
termie | for example, in my git workflow i will branch... do some work, send it for review, branch from that branch do more dependent work, later on update the first branch per review then i have to merge those changes into the dependent branch | 22:41 |
_0x44 | mtaylor: Why don't they pipelines the default then? You can't shelve/unshelve bzr branches and that makes me unhappy. | 22:41 |
termie | this sounds like it keeps things stacked so that branch 2 knows it needs to incorporate branch 1 also | 22:41 |
mtaylor | _0x44: well, pipelines are new | 22:41 |
mtaylor | termie: yes | 22:41 |
mtaylor | _0x44: it's a plugin - they typically poke at new features like this in a plugin for a while, and then as various plugins shake out to be pretty standard for folks, they'll incorporate into mainline | 22:42 |
termie | not the worlds most commonly difficult problem but there are a few times where i've gotten backed up enough that it got annoying | 22:42 |
*** algorist has quit IRC | 22:42 | |
_0x44 | mtaylor: Yeah, soren keeps telling me that's how things should be, but I feel second-class with my workflow only being accessible through plugins. :) | 22:43 |
*** algorist has joined #openstack | 22:43 | |
termie | _0x44: that's how they want you to feel | 22:43 |
_0x44 | termie: Did you see http://github.com/kfish/git-bzr ? I think I like yours better | 22:43 |
termie | _0x44: git-bzr-ng will love you | 22:43 |
_0x44 | termie: Yeah, that's why I'm using git-bzr now, and switching to git-bzr-ng soon. | 22:43 |
mtaylor | _0x44: well, I think we just assume people will install plugins | 22:43 |
termie | _0x44: i saw it yes, i link to it at the bottom of the readme | 22:43 |
termie | i think mine works significantly more directly | 22:43 |
termie | i also couldn't get git-bzr to work when i tried it | 22:44 |
_0x44 | termie: I can't read, so maybe that's why I didn't see ;) | 22:44 |
mtaylor | termie: I'll be interested to see if you get annoyed enough to write an lp-submit command for git-bzr-ng :) | 22:44 |
termie | mtaylor: i like writing merge proposal descriptions separately from commits | 22:44 |
termie | though i suppose i am used to writing that on the commandline with rietveld tools | 22:45 |
mtaylor | yay! tarmac made itself useful | 22:45 |
eday | mtaylor: yeah, thats pretty cool | 22:45 |
*** adrian_otto has joined #openstack | 22:45 | |
mtaylor | it's sort of funny, since we put it in, I've been waiting for the first time that happened :) | 22:45 |
termie | mtaylor: also, with the basics complete as they are now, adding other bzr commands is about 10 minutes worth of work | 22:45 |
termie | mtaylor: oh oh oh oh per-line codereviews | 22:46 |
termie | mtaylor: want want will send booze | 22:46 |
mtaylor | termie: yeah - I filed a bug | 22:46 |
mtaylor | termie: https://bugs.launchpad.net/bugs/609297 | 22:47 |
uvirtbot | Launchpad bug 609297 in launchpad-code "need ability to do in-line reviews" [Undecided,New] | 22:47 |
*** miclorb_ has joined #openstack | 22:47 | |
*** JordanRinke_ has quit IRC | 22:52 | |
*** metoikos has joined #openstack | 22:56 | |
*** matiu has joined #openstack | 23:00 | |
*** gnrfan has joined #openstack | 23:00 | |
*** jsgotangco has joined #openstack | 23:01 | |
*** _jcsmith has quit IRC | 23:06 | |
*** mray has quit IRC | 23:11 | |
*** jonesy_ has joined #openstack | 23:18 | |
*** rnewson has quit IRC | 23:21 | |
uvirtbot | New bug: #610274 in swift "obj replicator has hard coded replica count" [Medium,New] https://launchpad.net/bugs/610274 | 23:26 |
*** binaryWarrior has quit IRC | 23:27 | |
*** gasbakid has joined #openstack | 23:29 | |
*** sirp1 has joined #openstack | 23:45 | |
*** jsmith-away is now known as jsmith | 23:45 | |
*** mtaylor has quit IRC | 23:54 | |
*** tanto has quit IRC | 23:55 | |
*** jhalogen has joined #openstack | 23:58 | |
*** sirp1 has quit IRC | 00:01 | |
termie | i have a client who apparently uses rackspace cloud | 00:02 |
jsgotangco | hey | 00:02 |
termie | so this is my first time using that stuff | 00:02 |
termie | but anyway | 00:02 |
termie | i think my server needs more randomness | 00:02 |
termie | webcam + lavalamp anybody? | 00:03 |
*** mtaylor has joined #openstack | 00:04 | |
*** ChanServ sets mode: +v mtaylor | 00:04 | |
*** lbieber has quit IRC | 00:08 | |
*** mtaylor has quit IRC | 00:12 | |
*** mattf has joined #openstack | 00:16 | |
*** murkk has quit IRC | 00:22 | |
eday | soren: initial tests that I have working (1k concurrent connections, each pumping about 40mb) show twisted at ~116 seconds, c++ at ~41 seconds. eventlet still not making it through, but I've not tried to fix it much yet | 00:25 |
gustavomzw | eday: what are you using for testing? | 00:26 |
eday | gustavomzw: a c++ echo client I wrote, based on libevent | 00:26 |
*** lbieber has joined #openstack | 00:31 | |
gustavomzw | eday: how is the test setup? are you running from your host? | 00:34 |
*** lbieber has quit IRC | 00:36 | |
eday | gustavomzw: just on my laptop, so not hitting real network. testing just the cpu overhead really | 00:38 |
*** jdmaturen has quit IRC | 00:39 | |
*** jsmith is now known as jsmith-away | 00:57 | |
*** pvo has joined #openstack | 00:59 | |
*** ChanServ sets mode: +v pvo | 00:59 | |
*** lbieber has joined #openstack | 01:03 | |
*** lbieber has quit IRC | 01:06 | |
*** mtaylor has joined #openstack | 01:08 | |
*** ChanServ sets mode: +v mtaylor | 01:08 | |
*** jtdowney has quit IRC | 01:11 | |
*** jtdowney has joined #openstack | 01:11 | |
*** gustavomzw has quit IRC | 01:11 | |
*** jtdowney has quit IRC | 01:15 | |
*** binaryWarrior has joined #openstack | 01:16 | |
*** binaryWarrior has quit IRC | 01:21 | |
*** binaryWarrior has joined #openstack | 01:21 | |
*** binaryWarrior has joined #openstack | 01:22 | |
pvo | where is the most current nova install docs? | 01:22 |
pvo | I should specify ones that work. .. | 01:22 |
*** binaryWarrior has quit IRC | 01:22 | |
*** binaryWarrior has joined #openstack | 01:23 | |
*** justinsheehy has quit IRC | 01:23 | |
*** justinsheehy has joined #openstack | 01:28 | |
*** silassewell has quit IRC | 01:37 | |
*** gasbakid has quit IRC | 01:37 | |
eday | soren: found the issue, needed to crank up both /proc/sys/net/core/netdev_max_backlog and /proc/sys/net/core/somaxconn | 01:37 |
*** lbieber has joined #openstack | 01:37 | |
*** algorist has quit IRC | 01:38 | |
*** rdw has quit IRC | 01:41 | |
pvo | eday: found some answers re: api. | 01:47 |
pvo | endpoints should be ignorant | 01:47 |
pvo | makes the design simpler | 01:47 |
*** binaryWarrior has quit IRC | 01:49 | |
eday | pvo: hm? as in endpoint speaking HTTP should be a http->queue proxy pretty much? | 01:51 |
pvo | pretty much. First pass at api should be simply implementing rackspace api, sans auth | 01:52 |
pvo | since nas and rackspace will do auth differently | 01:52 |
pvo | nas/nasa | 01:52 |
eday | pvo: yeah, thats one of the things we talked about in Austin, need to pull nova-auth out into it's own thing and abstract both ends | 01:53 |
*** lbieber has quit IRC | 01:54 | |
*** _anm has joined #openstack | 01:54 | |
*** PiotrSikora has joined #openstack | 01:57 | |
*** littleidea_ has quit IRC | 02:06 | |
*** ianw has joined #openstack | 02:08 | |
creiht | eday: let me know if you need any help with the eventlet stuff | 02:08 |
*** littleidea_ has joined #openstack | 02:09 | |
termie | night all | 02:42 |
*** ianw has quit IRC | 02:47 | |
*** mrayzenoss has joined #openstack | 02:50 | |
*** mrayzenoss is now known as mray | 02:50 | |
*** mray has joined #openstack | 02:50 | |
*** jdmaturen has joined #openstack | 02:50 | |
*** rdw has joined #openstack | 02:51 | |
jaypipes | woot. datastore cleanly abstracted and passing all local tests... | 02:57 |
jaypipes | \o/ | 02:57 |
*** pvo_ has joined #openstack | 02:57 | |
*** ChanServ sets mode: +v pvo_ | 02:58 | |
*** littleidea_ has quit IRC | 02:58 | |
*** pvo has quit IRC | 03:00 | |
*** pvo_ is now known as pvo | 03:00 | |
_anm | other than euca-tools, what other ec2 and s3 tools can be used to manage instances running under nova? amazon's tools TOS forbids using them anywhere bu amazon | 03:01 |
_anm | im getting lots of errors managing images with euca-tools | 03:01 |
jaypipes | vishvananda: yo. :) re: https://code.launchpad.net/~vishvananda/nova/clean-auth/+merge/30431, I'd love to get that merged in asap. mtaylor, think you could review it? | 03:03 |
jaypipes | vishvananda: I've got the datastore abstracted pretty neatly and would like to push up some code that already has your code integrated... | 03:04 |
jaypipes | justinsb: around? | 03:04 |
justinsb | jaypipes: Hi! | 03:05 |
justinsb | jaypipes: Just saw your note on the datastore - cool stuff! | 03:06 |
jaypipes | justinsb: heya! :) so, I've got a ton of refactoring done...should be pretty close to integrating all your sqlite stuff :) | 03:06 |
justinsb | jaypipes: Awesome - anything you need me to do? | 03:06 |
jaypipes | justinsb: shall I push a branch for you to peruse? | 03:06 |
justinsb | jaypipes: Sure | 03:06 |
jaypipes | no, not really...just input would be great :) | 03:06 |
*** littleidea has joined #openstack | 03:07 | |
justinsb | jaypipes: Would you rather just have write access to my branch? Can I do that in Launchpad? | 03:07 |
*** kashyapc has joined #openstack | 03:07 | |
jaypipes | justinsb: nope, I started from a fresh trunk since there were a few things that conflicted...should be trivial to pull the Sqlite adapter code into the new stuff. | 03:07 |
justinsb | jaypipes: OK! | 03:08 |
*** RandomBrian has quit IRC | 03:24 | |
*** nettok has joined #openstack | 03:24 | |
*** lbieber has joined #openstack | 03:27 | |
*** mray has quit IRC | 03:31 | |
mtaylor | justinsb: you can push a branch to a team that you're both members of - like you can just push to lp;~nova/nova/whatever -and then either of you could push/pull to/from it | 03:32 |
mtaylor | jaypipes: looking over it now | 03:32 |
jaypipes | mtaylor: cheers | 03:32 |
*** kashyapc has quit IRC | 03:35 | |
mtaylor | jaypipes: ok. I did yours - could you give a quick look at https://code.edge.launchpad.net/~vishvananda/nova/virtualenv-fix/+merge/31005 | 03:40 |
jaypipes | mtaylor: yup | 03:41 |
*** mrayzenoss has joined #openstack | 03:45 | |
*** mrayzenoss is now known as mray | 03:45 | |
*** mray has joined #openstack | 03:45 | |
*** rajijoom has joined #openstack | 03:52 | |
justinsb | jaypipes: So the scope of the initial patch is purely refactoring out the data store with a clean abstraction layer? | 03:53 |
jaypipes | justinsb: yep :) | 03:54 |
*** gnrfan has quit IRC | 03:54 | |
justinsb | jaypipes: Cool | 03:54 |
justinsb | jaypipes: what did you do with pools? | 03:55 |
jaypipes | justinsb: haven't done anything yet. | 03:55 |
justinsb | jaypipes: So it still uses Redis directly? | 03:55 |
jaypipes | justinsb: nope, gimme a few minutes and I should have code to show you :) | 03:56 |
justinsb | jaypipes: OK :-) | 03:56 |
*** mray has quit IRC | 03:59 | |
*** rdw has quit IRC | 03:59 | |
*** copilot_ has quit IRC | 03:59 | |
*** doublerr has quit IRC | 03:59 | |
*** Daviey has quit IRC | 03:59 | |
*** rbergeron has quit IRC | 03:59 | |
*** creiht has quit IRC | 03:59 | |
*** letterj has quit IRC | 03:59 | |
*** larissa` has joined #openstack | 04:00 | |
*** mray has quit IRC | 04:01 | |
*** rdw has quit IRC | 04:01 | |
*** copilot_ has quit IRC | 04:01 | |
*** doublerr has quit IRC | 04:01 | |
*** Daviey has quit IRC | 04:01 | |
*** rbergeron has quit IRC | 04:01 | |
*** creiht has quit IRC | 04:01 | |
*** letterj has quit IRC | 04:01 | |
*** jdmaturen has quit IRC | 04:02 | |
*** gdusbabek has quit IRC | 04:02 | |
*** larissa has quit IRC | 04:02 | |
*** mjmac has quit IRC | 04:02 | |
*** pandemicsyn has quit IRC | 04:02 | |
*** pandemicsyn has joined #openstack | 04:02 | |
*** tjyang has quit IRC | 04:02 | |
*** polvi has quit IRC | 04:02 | |
*** gdusbabek has joined #openstack | 04:02 | |
*** matteo74 has quit IRC | 04:02 | |
*** jtimberman has quit IRC | 04:02 | |
*** driftx has quit IRC | 04:02 | |
*** driftx has joined #openstack | 04:02 | |
*** mray has joined #openstack | 04:02 | |
*** rdw has joined #openstack | 04:02 | |
*** copilot_ has joined #openstack | 04:02 | |
*** rbergeron has joined #openstack | 04:02 | |
*** doublerr has joined #openstack | 04:02 | |
*** Daviey has joined #openstack | 04:02 | |
*** creiht has joined #openstack | 04:02 | |
*** letterj has joined #openstack | 04:02 | |
*** brown.freenode.net sets mode: +vv creiht letterj | 04:02 | |
*** matteo_ has joined #openstack | 04:02 | |
*** polvi has joined #openstack | 04:03 | |
*** tjyang has joined #openstack | 04:03 | |
*** mjmac has joined #openstack | 04:03 | |
*** jtimberman has joined #openstack | 04:03 | |
*** larissa` is now known as larissa | 04:03 | |
*** _cerberus_ has quit IRC | 04:03 | |
*** termie has quit IRC | 04:03 | |
*** ralph has quit IRC | 04:03 | |
*** jtimberman has quit IRC | 04:03 | |
*** lbieber has quit IRC | 04:03 | |
*** PiotrSikora has quit IRC | 04:03 | |
*** miclorb_ has quit IRC | 04:03 | |
*** zul has quit IRC | 04:03 | |
*** londo has quit IRC | 04:03 | |
*** cory_ has quit IRC | 04:03 | |
*** vishvananda has quit IRC | 04:03 | |
*** rackerhacker has quit IRC | 04:04 | |
*** gholt has quit IRC | 04:04 | |
*** thom has quit IRC | 04:04 | |
*** samppah_ has quit IRC | 04:04 | |
*** mjmac has quit IRC | 04:04 | |
*** _anm has quit IRC | 04:04 | |
*** justinsheehy has quit IRC | 04:04 | |
*** mtaylor has quit IRC | 04:04 | |
*** jonesy_ has quit IRC | 04:04 | |
*** viewlogic has quit IRC | 04:04 | |
*** matiu has quit IRC | 04:04 | |
*** croteb has quit IRC | 04:04 | |
*** almaisan-away has quit IRC | 04:04 | |
*** Phil_FL has quit IRC | 04:04 | |
*** jonesy has quit IRC | 04:04 | |
*** mcgurrin has quit IRC | 04:04 | |
*** anticw has quit IRC | 04:04 | |
*** g0rdy has quit IRC | 04:04 | |
*** gein has quit IRC | 04:04 | |
*** e-vad has quit IRC | 04:04 | |
*** tjyang has quit IRC | 04:04 | |
*** rajijoom has quit IRC | 04:04 | |
*** _0x44 has quit IRC | 04:04 | |
*** filler has quit IRC | 04:04 | |
*** polvi has quit IRC | 04:04 | |
*** poli has quit IRC | 04:04 | |
*** soren has quit IRC | 04:04 | |
*** CnC has quit IRC | 04:04 | |
*** eday has quit IRC | 04:04 | |
*** scotticus has quit IRC | 04:04 | |
*** uvirtbot has quit IRC | 04:04 | |
*** jeremydei has quit IRC | 04:04 | |
*** bytee has quit IRC | 04:04 | |
*** snowboarder04 has quit IRC | 04:04 | |
*** sander__ has quit IRC | 04:04 | |
*** stefano_s has quit IRC | 04:04 | |
*** cparedes has quit IRC | 04:04 | |
*** btorch has quit IRC | 04:04 | |
*** jaypipes has quit IRC | 04:04 | |
*** ttx has quit IRC | 04:05 | |
*** cloud0 has quit IRC | 04:05 | |
*** jsmith-away has quit IRC | 04:05 | |
*** ghchinoy has quit IRC | 04:05 | |
*** littleidea has quit IRC | 04:05 | |
*** armmer has quit IRC | 04:05 | |
*** jeevan_ullas has quit IRC | 04:05 | |
*** clayg has quit IRC | 04:05 | |
*** wreese has quit IRC | 04:05 | |
*** justinsb has quit IRC | 04:05 | |
*** benoitc has quit IRC | 04:05 | |
*** anm_ has quit IRC | 04:05 | |
*** larsemil has quit IRC | 04:05 | |
*** ikk has quit IRC | 04:05 | |
*** brainproxy has quit IRC | 04:05 | |
*** admboom has quit IRC | 04:05 | |
*** chilts has quit IRC | 04:05 | |
*** chmouel has quit IRC | 04:05 | |
*** mattf has quit IRC | 04:05 | |
*** BartVB has quit IRC | 04:05 | |
*** omar has quit IRC | 04:05 | |
*** dubs has quit IRC | 04:05 | |
*** Garo_ has quit IRC | 04:05 | |
*** ChanServ sets mode: +v pandemicsyn | 04:05 | |
*** soren has joined #openstack | 04:05 | |
*** jtimberman has joined #openstack | 04:05 | |
*** mjmac has joined #openstack | 04:05 | |
*** tjyang has joined #openstack | 04:05 | |
*** polvi has joined #openstack | 04:05 | |
*** rajijoom has joined #openstack | 04:05 | |
*** lbieber has joined #openstack | 04:05 | |
*** littleidea has joined #openstack | 04:05 | |
*** PiotrSikora has joined #openstack | 04:05 | |
*** _anm has joined #openstack | 04:05 | |
*** justinsheehy has joined #openstack | 04:05 | |
*** mtaylor has joined #openstack | 04:05 | |
*** mattf has joined #openstack | 04:05 | |
*** jonesy_ has joined #openstack | 04:05 | |
*** matiu has joined #openstack | 04:05 | |
*** miclorb_ has joined #openstack | 04:05 | |
*** viewlogic has joined #openstack | 04:05 | |
*** jaypipes has joined #openstack | 04:05 | |
*** armmer has joined #openstack | 04:05 | |
*** vishvananda has joined #openstack | 04:05 | |
*** BartVB has joined #openstack | 04:05 | |
*** jeevan_ullas has joined #openstack | 04:05 | |
*** croteb has joined #openstack | 04:05 | |
*** almaisan-away has joined #openstack | 04:05 | |
*** Phil_FL has joined #openstack | 04:05 | |
*** rackerhacker has joined #openstack | 04:05 | |
*** jonesy has joined #openstack | 04:05 | |
*** poli has joined #openstack | 04:05 | |
*** bytee has joined #openstack | 04:05 | |
*** ttx has joined #openstack | 04:05 | |
*** zul has joined #openstack | 04:05 | |
*** clayg has joined #openstack | 04:05 | |
*** mcgurrin has joined #openstack | 04:05 | |
*** snowboarder04 has joined #openstack | 04:05 | |
*** sander__ has joined #openstack | 04:05 | |
*** stefano_s has joined #openstack | 04:05 | |
*** cparedes has joined #openstack | 04:05 | |
*** jsmith-away has joined #openstack | 04:05 | |
*** wreese has joined #openstack | 04:05 | |
*** ikk has joined #openstack | 04:05 | |
*** justinsb has joined #openstack | 04:05 | |
*** _cerberus_ has joined #openstack | 04:05 | |
*** samppah_ has joined #openstack | 04:05 | |
*** ralph has joined #openstack | 04:05 | |
*** ghchinoy has joined #openstack | 04:05 | |
*** termie has joined #openstack | 04:05 | |
*** brown.freenode.net sets mode: +v mtaylor | 04:05 | |
*** londo has joined #openstack | 04:05 | |
*** anticw has joined #openstack | 04:05 | |
*** benoitc has joined #openstack | 04:05 | |
*** anm_ has joined #openstack | 04:05 | |
*** btorch has joined #openstack | 04:05 | |
*** omar has joined #openstack | 04:05 | |
*** cory_ has joined #openstack | 04:05 | |
*** _0x44 has joined #openstack | 04:05 | |
*** larsemil has joined #openstack | 04:05 | |
*** gein has joined #openstack | 04:05 | |
*** filler has joined #openstack | 04:05 | |
*** cloud0 has joined #openstack | 04:05 | |
*** e-vad has joined #openstack | 04:05 | |
*** CnC has joined #openstack | 04:05 | |
*** uvirtbot has joined #openstack | 04:05 | |
*** thom has joined #openstack | 04:05 | |
*** jeremydei has joined #openstack | 04:05 | |
*** scotticus has joined #openstack | 04:05 | |
*** chmouel has joined #openstack | 04:05 | |
*** g0rdy has joined #openstack | 04:05 | |
*** admboom has joined #openstack | 04:05 | |
*** dubs has joined #openstack | 04:05 | |
*** chilts has joined #openstack | 04:05 | |
*** Garo_ has joined #openstack | 04:05 | |
*** brainproxy has joined #openstack | 04:05 | |
*** gholt has joined #openstack | 04:05 | |
*** eday has joined #openstack | 04:05 | |
*** brown.freenode.net sets mode: +v _0x44 | 04:05 | |
*** ChanServ sets mode: +v _cerberus_ | 04:05 | |
*** cw_ has joined #openstack | 04:05 | |
*** larissa` has joined #openstack | 04:06 | |
*** veszig` has quit IRC | 04:06 | |
*** veszig` has joined #openstack | 04:06 | |
*** _cerberus_ has quit IRC | 04:06 | |
*** termie has quit IRC | 04:06 | |
*** ralph has quit IRC | 04:06 | |
*** jtimberman has quit IRC | 04:06 | |
*** lbieber has quit IRC | 04:06 | |
*** PiotrSikora has quit IRC | 04:06 | |
*** miclorb_ has quit IRC | 04:06 | |
*** zul has quit IRC | 04:06 | |
*** londo has quit IRC | 04:06 | |
*** cory_ has quit IRC | 04:06 | |
*** vishvananda has quit IRC | 04:06 | |
*** rackerhacker has quit IRC | 04:06 | |
*** gholt has quit IRC | 04:06 | |
*** thom has quit IRC | 04:06 | |
*** samppah_ has quit IRC | 04:06 | |
*** mjmac has quit IRC | 04:06 | |
*** _anm has quit IRC | 04:06 | |
*** justinsheehy has quit IRC | 04:06 | |
*** mtaylor has quit IRC | 04:06 | |
*** jonesy_ has quit IRC | 04:06 | |
*** viewlogic has quit IRC | 04:06 | |
*** matiu has quit IRC | 04:06 | |
*** croteb has quit IRC | 04:06 | |
*** almaisan-away has quit IRC | 04:06 | |
*** Phil_FL has quit IRC | 04:06 | |
*** jonesy has quit IRC | 04:06 | |
*** mcgurrin has quit IRC | 04:06 | |
*** anticw has quit IRC | 04:06 | |
*** g0rdy has quit IRC | 04:06 | |
*** gein has quit IRC | 04:06 | |
*** e-vad has quit IRC | 04:06 | |
*** tjyang has quit IRC | 04:06 | |
*** rajijoom has quit IRC | 04:06 | |
*** _0x44 has quit IRC | 04:06 | |
*** filler has quit IRC | 04:06 | |
*** soren has quit IRC | 04:06 | |
*** polvi has quit IRC | 04:06 | |
*** poli has quit IRC | 04:06 | |
*** CnC has quit IRC | 04:06 | |
*** eday has quit IRC | 04:06 | |
*** scotticus has quit IRC | 04:06 | |
*** uvirtbot has quit IRC | 04:06 | |
*** jeremydei has quit IRC | 04:06 | |
*** bytee has quit IRC | 04:06 | |
*** snowboarder04 has quit IRC | 04:06 | |
*** sander__ has quit IRC | 04:06 | |
*** stefano_s has quit IRC | 04:06 | |
*** cparedes has quit IRC | 04:06 | |
*** btorch has quit IRC | 04:06 | |
*** jaypipes has quit IRC | 04:06 | |
*** ttx has quit IRC | 04:06 | |
*** cloud0 has quit IRC | 04:06 | |
*** jsmith-away has quit IRC | 04:06 | |
*** ghchinoy has quit IRC | 04:06 | |
*** littleidea has quit IRC | 04:06 | |
*** armmer has quit IRC | 04:06 | |
*** jeevan_ullas has quit IRC | 04:06 | |
*** clayg has quit IRC | 04:06 | |
*** wreese has quit IRC | 04:06 | |
*** justinsb has quit IRC | 04:06 | |
*** benoitc has quit IRC | 04:06 | |
*** anm_ has quit IRC | 04:06 | |
*** larsemil has quit IRC | 04:06 | |
*** ikk has quit IRC | 04:06 | |
*** brainproxy has quit IRC | 04:06 | |
*** admboom has quit IRC | 04:06 | |
*** chilts has quit IRC | 04:06 | |
*** chmouel has quit IRC | 04:06 | |
*** mattf has quit IRC | 04:06 | |
*** BartVB has quit IRC | 04:07 | |
*** omar has quit IRC | 04:07 | |
*** dubs has quit IRC | 04:07 | |
*** Garo_ has quit IRC | 04:07 | |
*** larissa has quit IRC | 04:07 | |
*** larissa` is now known as larissa | 04:07 | |
*** redbo_ has joined #openstack | 04:07 | |
*** soren has joined #openstack | 04:07 | |
*** jtimberman has joined #openstack | 04:07 | |
*** mjmac has joined #openstack | 04:07 | |
*** tjyang has joined #openstack | 04:07 | |
*** polvi has joined #openstack | 04:07 | |
*** rajijoom has joined #openstack | 04:07 | |
*** lbieber has joined #openstack | 04:07 | |
*** littleidea has joined #openstack | 04:07 | |
*** PiotrSikora has joined #openstack | 04:07 | |
*** _anm has joined #openstack | 04:07 | |
*** justinsheehy has joined #openstack | 04:07 | |
*** mtaylor has joined #openstack | 04:07 | |
*** mattf has joined #openstack | 04:07 | |
*** jonesy_ has joined #openstack | 04:07 | |
*** matiu has joined #openstack | 04:07 | |
*** miclorb_ has joined #openstack | 04:07 | |
*** viewlogic has joined #openstack | 04:07 | |
*** jaypipes has joined #openstack | 04:07 | |
*** armmer has joined #openstack | 04:07 | |
*** vishvananda has joined #openstack | 04:07 | |
*** BartVB has joined #openstack | 04:07 | |
*** jeevan_ullas has joined #openstack | 04:07 | |
*** croteb has joined #openstack | 04:07 | |
*** almaisan-away has joined #openstack | 04:07 | |
*** Phil_FL has joined #openstack | 04:07 | |
*** rackerhacker has joined #openstack | 04:07 | |
*** jonesy has joined #openstack | 04:07 | |
*** poli has joined #openstack | 04:07 | |
*** bytee has joined #openstack | 04:07 | |
*** ttx has joined #openstack | 04:07 | |
*** zul has joined #openstack | 04:07 | |
*** clayg has joined #openstack | 04:07 | |
*** mcgurrin has joined #openstack | 04:07 | |
*** snowboarder04 has joined #openstack | 04:07 | |
*** sander__ has joined #openstack | 04:07 | |
*** stefano_s has joined #openstack | 04:07 | |
*** cparedes has joined #openstack | 04:07 | |
*** jsmith-away has joined #openstack | 04:07 | |
*** wreese has joined #openstack | 04:07 | |
*** ikk has joined #openstack | 04:07 | |
*** justinsb has joined #openstack | 04:07 | |
*** _cerberus_ has joined #openstack | 04:07 | |
*** samppah_ has joined #openstack | 04:07 | |
*** ralph has joined #openstack | 04:07 | |
*** ghchinoy has joined #openstack | 04:07 | |
*** verne.freenode.net sets mode: +v mtaylor | 04:07 | |
*** termie has joined #openstack | 04:07 | |
*** londo has joined #openstack | 04:07 | |
*** anticw has joined #openstack | 04:07 | |
*** benoitc has joined #openstack | 04:07 | |
*** anm_ has joined #openstack | 04:07 | |
*** btorch has joined #openstack | 04:07 | |
*** omar has joined #openstack | 04:07 | |
*** cory_ has joined #openstack | 04:07 | |
*** _0x44 has joined #openstack | 04:07 | |
*** larsemil has joined #openstack | 04:07 | |
*** gein has joined #openstack | 04:07 | |
*** filler has joined #openstack | 04:07 | |
*** cloud0 has joined #openstack | 04:07 | |
*** e-vad has joined #openstack | 04:07 | |
*** CnC has joined #openstack | 04:07 | |
*** uvirtbot has joined #openstack | 04:07 | |
*** thom has joined #openstack | 04:07 | |
*** jeremydei has joined #openstack | 04:07 | |
*** scotticus has joined #openstack | 04:07 | |
*** chmouel has joined #openstack | 04:07 | |
*** g0rdy has joined #openstack | 04:07 | |
*** admboom has joined #openstack | 04:07 | |
*** dubs has joined #openstack | 04:07 | |
*** chilts has joined #openstack | 04:07 | |
*** Garo_ has joined #openstack | 04:07 | |
*** brainproxy has joined #openstack | 04:07 | |
*** gholt has joined #openstack | 04:07 | |
*** eday has joined #openstack | 04:07 | |
*** verne.freenode.net sets mode: +v _0x44 | 04:07 | |
*** ChanServ sets mode: +v _cerberus_ | 04:07 | |
*** _cerberus_ has quit IRC | 04:09 | |
*** termie has quit IRC | 04:09 | |
*** ralph has quit IRC | 04:09 | |
*** jtimberman has quit IRC | 04:09 | |
*** lbieber has quit IRC | 04:09 | |
*** PiotrSikora has quit IRC | 04:09 | |
*** miclorb_ has quit IRC | 04:09 | |
*** zul has quit IRC | 04:09 | |
*** londo has quit IRC | 04:09 | |
*** cory_ has quit IRC | 04:09 | |
*** vishvananda has quit IRC | 04:09 | |
*** rackerhacker has quit IRC | 04:09 | |
*** gholt has quit IRC | 04:09 | |
*** thom has quit IRC | 04:09 | |
*** samppah_ has quit IRC | 04:09 | |
*** redbo_ has quit IRC | 04:09 | |
*** mjmac has quit IRC | 04:09 | |
*** _anm has quit IRC | 04:09 | |
*** justinsheehy has quit IRC | 04:09 | |
*** mtaylor has quit IRC | 04:09 | |
*** jonesy_ has quit IRC | 04:09 | |
*** viewlogic has quit IRC | 04:09 | |
*** matiu has quit IRC | 04:09 | |
*** croteb has quit IRC | 04:09 | |
*** almaisan-away has quit IRC | 04:09 | |
*** Phil_FL has quit IRC | 04:09 | |
*** jonesy has quit IRC | 04:09 | |
*** mcgurrin has quit IRC | 04:09 | |
*** anticw has quit IRC | 04:09 | |
*** g0rdy has quit IRC | 04:09 | |
*** gein has quit IRC | 04:09 | |
*** e-vad has quit IRC | 04:09 | |
*** tjyang has quit IRC | 04:09 | |
*** rajijoom has quit IRC | 04:09 | |
*** _0x44 has quit IRC | 04:09 | |
*** filler has quit IRC | 04:09 | |
*** soren has quit IRC | 04:09 | |
*** polvi has quit IRC | 04:09 | |
*** poli has quit IRC | 04:09 | |
*** CnC has quit IRC | 04:09 | |
*** eday has quit IRC | 04:09 | |
*** scotticus has quit IRC | 04:09 | |
*** uvirtbot has quit IRC | 04:09 | |
*** jeremydei has quit IRC | 04:09 | |
*** bytee has quit IRC | 04:09 | |
*** snowboarder04 has quit IRC | 04:09 | |
*** sander__ has quit IRC | 04:09 | |
*** stefano_s has quit IRC | 04:09 | |
*** cparedes has quit IRC | 04:09 | |
*** btorch has quit IRC | 04:09 | |
*** jaypipes has quit IRC | 04:09 | |
*** ttx has quit IRC | 04:09 | |
*** cloud0 has quit IRC | 04:09 | |
*** jsmith-away has quit IRC | 04:09 | |
*** ghchinoy has quit IRC | 04:09 | |
*** littleidea has quit IRC | 04:10 | |
*** armmer has quit IRC | 04:10 | |
*** jeevan_ullas has quit IRC | 04:10 | |
*** clayg has quit IRC | 04:10 | |
*** wreese has quit IRC | 04:10 | |
*** justinsb has quit IRC | 04:10 | |
*** benoitc has quit IRC | 04:10 | |
*** anm_ has quit IRC | 04:10 | |
*** larsemil has quit IRC | 04:10 | |
*** ikk has quit IRC | 04:10 | |
*** brainproxy has quit IRC | 04:10 | |
*** admboom has quit IRC | 04:10 | |
*** chilts has quit IRC | 04:10 | |
*** chmouel has quit IRC | 04:10 | |
*** mattf has quit IRC | 04:10 | |
*** BartVB has quit IRC | 04:10 | |
*** omar has quit IRC | 04:10 | |
*** dubs has quit IRC | 04:10 | |
*** Garo_ has quit IRC | 04:10 | |
*** veszig has quit IRC | 04:10 | |
*** redbo has quit IRC | 04:10 | |
*** drewlande has quit IRC | 04:10 | |
*** cw has quit IRC | 04:10 | |
*** pandemicsyn has quit IRC | 04:11 | |
*** sophiap has quit IRC | 04:11 | |
*** notmyname has quit IRC | 04:11 | |
*** dendrobates has quit IRC | 04:11 | |
*** LoRez has quit IRC | 04:11 | |
*** loglaunch has quit IRC | 04:11 | |
*** rektide has quit IRC | 04:11 | |
*** chromakode has quit IRC | 04:11 | |
*** pandemicsyn has joined #openstack | 04:12 | |
*** chromakode has joined #openstack | 04:12 | |
*** sophiap has joined #openstack | 04:12 | |
*** notmyname has joined #openstack | 04:12 | |
*** LoRez has joined #openstack | 04:12 | |
*** rektide has joined #openstack | 04:12 | |
*** dendrobates has joined #openstack | 04:12 | |
*** loglaunch has joined #openstack | 04:12 | |
*** verne.freenode.net sets mode: +vv pandemicsyn notmyname | 04:12 | |
*** ChanServ sets mode: +v dendrobates | 04:12 | |
*** glange is now known as 84XAANVW0 | 04:13 | |
*** jdmaturen has joined #openstack | 04:14 | |
*** 36DAAB5CX has joined #openstack | 04:14 | |
*** glange has joined #openstack | 04:14 | |
*** drewlande has joined #openstack | 04:14 | |
*** redbo_ has joined #openstack | 04:14 | |
*** soren has joined #openstack | 04:14 | |
*** jtimberman has joined #openstack | 04:14 | |
*** mjmac has joined #openstack | 04:14 | |
*** tjyang has joined #openstack | 04:14 | |
*** polvi has joined #openstack | 04:14 | |
*** rajijoom has joined #openstack | 04:14 | |
*** lbieber has joined #openstack | 04:14 | |
*** littleidea has joined #openstack | 04:14 | |
*** PiotrSikora has joined #openstack | 04:14 | |
*** _anm has joined #openstack | 04:14 | |
*** justinsheehy has joined #openstack | 04:14 | |
*** mtaylor has joined #openstack | 04:14 | |
*** mattf has joined #openstack | 04:14 | |
*** jonesy_ has joined #openstack | 04:14 | |
*** matiu has joined #openstack | 04:14 | |
*** miclorb_ has joined #openstack | 04:14 | |
*** viewlogic has joined #openstack | 04:14 | |
*** jaypipes has joined #openstack | 04:14 | |
*** armmer has joined #openstack | 04:14 | |
*** vishvananda has joined #openstack | 04:14 | |
*** BartVB has joined #openstack | 04:14 | |
*** jeevan_ullas has joined #openstack | 04:14 | |
*** croteb has joined #openstack | 04:14 | |
*** almaisan-away has joined #openstack | 04:14 | |
*** Phil_FL has joined #openstack | 04:14 | |
*** rackerhacker has joined #openstack | 04:14 | |
*** jonesy has joined #openstack | 04:14 | |
*** poli has joined #openstack | 04:14 | |
*** bytee has joined #openstack | 04:14 | |
*** ttx has joined #openstack | 04:14 | |
*** zul has joined #openstack | 04:14 | |
*** clayg has joined #openstack | 04:14 | |
*** mcgurrin has joined #openstack | 04:14 | |
*** snowboarder04 has joined #openstack | 04:14 | |
*** sander__ has joined #openstack | 04:14 | |
*** stefano_s has joined #openstack | 04:14 | |
*** cparedes has joined #openstack | 04:14 | |
*** jsmith-away has joined #openstack | 04:14 | |
*** wreese has joined #openstack | 04:14 | |
*** ikk has joined #openstack | 04:14 | |
*** justinsb has joined #openstack | 04:14 | |
*** verne.freenode.net sets mode: +v mtaylor | 04:14 | |
*** _cerberus_ has joined #openstack | 04:14 | |
*** samppah_ has joined #openstack | 04:14 | |
*** ralph has joined #openstack | 04:14 | |
*** ghchinoy has joined #openstack | 04:14 | |
*** termie has joined #openstack | 04:14 | |
*** londo has joined #openstack | 04:14 | |
*** anticw has joined #openstack | 04:14 | |
*** benoitc has joined #openstack | 04:14 | |
*** anm_ has joined #openstack | 04:14 | |
*** btorch has joined #openstack | 04:14 | |
*** omar has joined #openstack | 04:14 | |
*** cory_ has joined #openstack | 04:14 | |
*** _0x44 has joined #openstack | 04:14 | |
*** larsemil has joined #openstack | 04:14 | |
*** gein has joined #openstack | 04:14 | |
*** filler has joined #openstack | 04:14 | |
*** cloud0 has joined #openstack | 04:14 | |
*** e-vad has joined #openstack | 04:14 | |
*** CnC has joined #openstack | 04:14 | |
*** uvirtbot has joined #openstack | 04:14 | |
*** thom has joined #openstack | 04:14 | |
*** jeremydei has joined #openstack | 04:14 | |
*** scotticus has joined #openstack | 04:14 | |
*** chmouel has joined #openstack | 04:14 | |
*** g0rdy has joined #openstack | 04:14 | |
*** admboom has joined #openstack | 04:14 | |
*** dubs has joined #openstack | 04:14 | |
*** chilts has joined #openstack | 04:14 | |
*** Garo_ has joined #openstack | 04:14 | |
*** brainproxy has joined #openstack | 04:14 | |
*** eday has joined #openstack | 04:14 | |
*** gholt has joined #openstack | 04:14 | |
*** verne.freenode.net sets mode: +v _0x44 | 04:14 | |
*** ChanServ sets mode: +v _cerberus_ | 04:14 | |
*** sureshgv has quit IRC | 04:14 | |
*** jtimberman has quit IRC | 04:15 | |
*** jtimberman has joined #openstack | 04:15 | |
*** glange has quit IRC | 04:15 | |
*** _cerberus_ has quit IRC | 04:15 | |
*** termie has quit IRC | 04:15 | |
*** ralph has quit IRC | 04:15 | |
*** jdmaturen has quit IRC | 04:15 | |
*** redbo_ has quit IRC | 04:15 | |
*** mjmac has quit IRC | 04:15 | |
*** _anm has quit IRC | 04:15 | |
*** justinsheehy has quit IRC | 04:15 | |
*** mtaylor has quit IRC | 04:15 | |
*** jonesy_ has quit IRC | 04:15 | |
*** viewlogic has quit IRC | 04:15 | |
*** matiu has quit IRC | 04:15 | |
*** croteb has quit IRC | 04:15 | |
*** almaisan-away has quit IRC | 04:15 | |
*** Phil_FL has quit IRC | 04:15 | |
*** jonesy has quit IRC | 04:15 | |
*** mcgurrin has quit IRC | 04:15 | |
*** anticw has quit IRC | 04:15 | |
*** g0rdy has quit IRC | 04:15 | |
*** gein has quit IRC | 04:15 | |
*** e-vad has quit IRC | 04:15 | |
*** 36DAAB5CX has quit IRC | 04:15 | |
*** tjyang has quit IRC | 04:15 | |
*** rajijoom has quit IRC | 04:15 | |
*** _0x44 has quit IRC | 04:15 | |
*** filler has quit IRC | 04:15 | |
*** drewlande has quit IRC | 04:15 | |
*** soren has quit IRC | 04:15 | |
*** polvi has quit IRC | 04:15 | |
*** poli has quit IRC | 04:16 | |
*** CnC has quit IRC | 04:16 | |
*** eday has quit IRC | 04:16 | |
*** scotticus has quit IRC | 04:16 | |
*** uvirtbot has quit IRC | 04:16 | |
*** jeremydei has quit IRC | 04:16 | |
*** bytee has quit IRC | 04:16 | |
*** snowboarder04 has quit IRC | 04:16 | |
*** sander__ has quit IRC | 04:16 | |
*** stefano_s has quit IRC | 04:16 | |
*** cparedes has quit IRC | 04:16 | |
*** btorch has quit IRC | 04:16 | |
*** jaypipes has quit IRC | 04:16 | |
*** ttx has quit IRC | 04:16 | |
*** cloud0 has quit IRC | 04:16 | |
*** jsmith-away has quit IRC | 04:16 | |
*** ghchinoy has quit IRC | 04:16 | |
*** littleidea has quit IRC | 04:16 | |
*** armmer has quit IRC | 04:16 | |
*** jeevan_ullas has quit IRC | 04:16 | |
*** clayg has quit IRC | 04:16 | |
*** wreese has quit IRC | 04:16 | |
*** justinsb has quit IRC | 04:16 | |
*** benoitc has quit IRC | 04:16 | |
*** anm_ has quit IRC | 04:16 | |
*** larsemil has quit IRC | 04:16 | |
*** ikk has quit IRC | 04:16 | |
*** brainproxy has quit IRC | 04:16 | |
*** admboom has quit IRC | 04:16 | |
*** chilts has quit IRC | 04:16 | |
*** chmouel has quit IRC | 04:16 | |
*** mattf has quit IRC | 04:16 | |
*** BartVB has quit IRC | 04:16 | |
*** omar has quit IRC | 04:16 | |
*** dubs has quit IRC | 04:16 | |
*** Garo_ has quit IRC | 04:16 | |
*** jaker has quit IRC | 04:17 | |
*** 84XAANVW0 has quit IRC | 04:17 | |
*** seats has quit IRC | 04:17 | |
*** seats has joined #openstack | 04:17 | |
*** jdmaturen has joined #openstack | 04:18 | |
*** 36DAAB5CX has joined #openstack | 04:18 | |
*** glange has joined #openstack | 04:18 | |
*** drewlande has joined #openstack | 04:18 | |
*** redbo_ has joined #openstack | 04:18 | |
*** soren has joined #openstack | 04:18 | |
*** mjmac has joined #openstack | 04:18 | |
*** tjyang has joined #openstack | 04:18 | |
*** polvi has joined #openstack | 04:18 | |
*** rajijoom has joined #openstack | 04:18 | |
*** littleidea has joined #openstack | 04:18 | |
*** _anm has joined #openstack | 04:18 | |
*** justinsheehy has joined #openstack | 04:18 | |
*** mtaylor has joined #openstack | 04:18 | |
*** mattf has joined #openstack | 04:18 | |
*** jonesy_ has joined #openstack | 04:18 | |
*** matiu has joined #openstack | 04:18 | |
*** viewlogic has joined #openstack | 04:18 | |
*** jaypipes has joined #openstack | 04:18 | |
*** armmer has joined #openstack | 04:18 | |
*** BartVB has joined #openstack | 04:18 | |
*** jeevan_ullas has joined #openstack | 04:18 | |
*** croteb has joined #openstack | 04:18 | |
*** almaisan-away has joined #openstack | 04:18 | |
*** Phil_FL has joined #openstack | 04:18 | |
*** jonesy has joined #openstack | 04:18 | |
*** poli has joined #openstack | 04:18 | |
*** bytee has joined #openstack | 04:18 | |
*** ttx has joined #openstack | 04:18 | |
*** clayg has joined #openstack | 04:18 | |
*** mcgurrin has joined #openstack | 04:18 | |
*** snowboarder04 has joined #openstack | 04:18 | |
*** sander__ has joined #openstack | 04:18 | |
*** stefano_s has joined #openstack | 04:18 | |
*** cparedes has joined #openstack | 04:18 | |
*** jsmith-away has joined #openstack | 04:18 | |
*** wreese has joined #openstack | 04:18 | |
*** ikk has joined #openstack | 04:18 | |
*** justinsb has joined #openstack | 04:18 | |
*** _cerberus_ has joined #openstack | 04:18 | |
*** ralph has joined #openstack | 04:18 | |
*** ghchinoy has joined #openstack | 04:18 | |
*** termie has joined #openstack | 04:18 | |
*** anticw has joined #openstack | 04:18 | |
*** benoitc has joined #openstack | 04:18 | |
*** anm_ has joined #openstack | 04:18 | |
*** verne.freenode.net sets mode: +v mtaylor | 04:18 | |
*** btorch has joined #openstack | 04:18 | |
*** omar has joined #openstack | 04:18 | |
*** _0x44 has joined #openstack | 04:18 | |
*** larsemil has joined #openstack | 04:18 | |
*** gein has joined #openstack | 04:18 | |
*** filler has joined #openstack | 04:18 | |
*** cloud0 has joined #openstack | 04:18 | |
*** e-vad has joined #openstack | 04:18 | |
*** CnC has joined #openstack | 04:18 | |
*** uvirtbot has joined #openstack | 04:18 | |
*** jeremydei has joined #openstack | 04:18 | |
*** scotticus has joined #openstack | 04:18 | |
*** chmouel has joined #openstack | 04:18 | |
*** g0rdy has joined #openstack | 04:18 | |
*** admboom has joined #openstack | 04:18 | |
*** dubs has joined #openstack | 04:18 | |
*** chilts has joined #openstack | 04:18 | |
*** Garo_ has joined #openstack | 04:18 | |
*** brainproxy has joined #openstack | 04:18 | |
*** eday has joined #openstack | 04:18 | |
*** verne.freenode.net sets mode: +v _0x44 | 04:18 | |
*** ChanServ sets mode: +v _cerberus_ | 04:18 | |
*** jaker has joined #openstack | 04:19 | |
_anm | in nova/compute/libvirt.xml.template is th element: <type>hvm</type> | 04:19 |
_anm | is that correct? shouldn;t that be <type>kvm</type>? | 04:19 |
_anm | does it matter? | 04:19 |
_anm | could this be why instances will never actually launch? | 04:19 |
_anm | dang, welcome back everybody, i was evidently having a nice conversation with myself | 04:19 |
mtaylor | zomg. gonna kill something | 04:19 |
mtaylor | pvo: hey - if you can merge your branch up with trunk and resolve that conflict we can land that bad boy :) | 04:19 |
pvo | sure... give me just a few. | 04:19 |
_anm | as I was saying: in nova/compute/libvirt.xml.template is the4 element: <type>hvm</type> | 04:19 |
pvo | I think vish and I were racing | 04:19 |
_anm | is that a typo? | 04:20 |
mtaylor | pvo: yup | 04:20 |
_anm | shoudl it be: <type>kvm</type> ? | 04:20 |
_anm | also, I have installed s3cmd on Ubuntu 10.4 and it appears to work fine with nova | 04:25 |
_anm | after running s3cmd --configure, you just have to edit ~/.s3cmd and change the urls from amazon to 127.0.0.1 | 04:26 |
_anm | s3cmd ls accurately lists my buckets | 04:26 |
chilts | so the API is the same as S3? | 04:26 |
* chilts doesn't really like the S3 API (or indeed any of the AWS ones) | 04:26 | |
*** jdmaturen has quit IRC | 04:27 | |
pvo | mtaylor: I deleted that branch and just pushed another and proposed the merge. Probably not the right workflow but it was simple this time | 04:28 |
mtaylor | pvo: :) | 04:28 |
_anm | chilts: it is supposed to be compatible | 04:28 |
chilts | right | 04:28 |
* pvo still a gittard ... bzr is new. | 04:28 | |
_anm | so s3cmd ls returns a list of my buckets | 04:29 |
mtaylor | pvo: that's fine... no need to learn it all at once :) | 04:29 |
chilts | I like the current Rackspace APIs so I guess I just think the AWS ones aren't as good | 04:29 |
mtaylor | pvo: you should see me attempting to use git :) | 04:29 |
pvo | heh. I went from svn -> git... my head was swimming for days | 04:29 |
_anm | bzr actually has a very decent UI called Explorer | 04:30 |
_anm | i tried it today and it just worked | 04:30 |
* mtaylor went from svn -> bitkeeper -> bzr | 04:30 | |
pvo | _anm: cool. I'll check it out. | 04:31 |
pvo | mtaylor: you followed the kernel? | 04:31 |
*** _buck has joined #openstack | 04:31 | |
* _anm went from MS VSS -> cvs -> svn -> git and now bzr | 04:31 | |
pvo | er, minus bzr | 04:31 |
mtaylor | pvo: no - I was on the other project that used bitkeeper - mysql | 04:31 |
_anm | pvo, just run bzr Explorer and if it's installe dit will launch in X | 04:32 |
pvo | _anm: fairly certain it isn't on osx. | 04:32 |
_anm | pvo, thats what I'm on, and it is! | 04:32 |
pvo | it has deps I have to go fix and my eyes are starting to itch. | 04:32 |
_anm | it's actually the best looking one! | 04:32 |
* mtaylor has been using bzr now for about 4 years now though | 04:33 | |
pvo | ok, let me try it. | 04:33 |
mtaylor | so I have long since forgotten any learning curve that might have been there ;) | 04:33 |
_anm | pvo I installed bzr from the bazaar dmg, not from macports, but it is supposed to be in both | 04:34 |
pvo | mtaylor: it isn't as high as leaping to dvcs, but the syntax differences can be frustrating | 04:34 |
mtaylor | pvo: yeah | 04:34 |
pvo | _anm: I tried this http://doc.bazaar.canonical.com/explorer/en/install-osx.html | 04:34 |
pvo | seems to work | 04:34 |
mtaylor | pvo: one of the biggest things that drives me batty when I try to use git are the commands that are named the same but do _vastly_ different things | 04:35 |
mtaylor | pvo: I imagine the same is quite true in reverse | 04:35 |
pvo | mtaylor: heh. exactly | 04:35 |
_anm | pvo thats the one - it just worked for me | 04:35 |
pvo | I was trying to just a 'git show commit' equiv but couldn't figure it out. I gave up after 30 seconds and just rebranched and fixed the 1 one. | 04:35 |
pvo | line | 04:35 |
*** vish1 has joined #openstack | 04:36 | |
vish1 | anyone have any ideas why coverage fails? | 04:37 |
pvo | even though I hate guis, I see how this is useful | 04:37 |
vish1 | i suspect working directory and/or flags issues | 04:37 |
*** miclorb_ has quit IRC | 04:39 | |
mtaylor | vish1: I was thinking it had to do with coverage monkey-patching itself in or whatever it is that it does | 04:43 |
mtaylor | vish1: and perhaps that was overriding some magic that nova was doing? | 04:43 |
mtaylor | pvo: what does "git show commit' do? | 04:43 |
pvo | git show XXXX will show you the diff of a particular commit | 04:44 |
pvo | 'git show' will give you the last commit | 04:44 |
mtaylor | ah. | 04:44 |
mtaylor | just for reference - in bzr is the bzr diff command | 04:45 |
pvo | so it would be bzr diff rev# ? | 04:45 |
mtaylor | bzr diff -r<some revision specifier> | 04:45 |
mtaylor | where <some revision specifier> can be a single rev or a range | 04:45 |
pvo | there we go. | 04:46 |
pvo | thats what I needed | 04:46 |
pvo | I was trying 'bzr diff 170' | 04:46 |
pvo | and it just complained | 04:46 |
pvo | I guess 170 without the r indicated a branch | 04:46 |
mtaylor | also - at some point (and I'm gonna blog on this tomorrow probably) if you get bored read bzr help revisionspec | 04:46 |
* pvo wanders off to install bzr-pager | 04:48 | |
_anm | aha! i think I may have found out why an instance won't actualy launch under nova: the domain element has a type attribute of "hvm" in nova/compute/libvirt.xml.template | 04:49 |
_anm | according to here: http://libvirt.org/formatdomain.html#elements | 04:50 |
_anm | this can only be one of: "xen", "kvm", "qemu", "lxc" and "kqemu" | 04:50 |
_anm | but also states that the value can be driver specific | 04:51 |
_anm | so my question is wheter this shoudl be "hvm" as it is, or "kvm" | 04:51 |
_anm | also: i tipe gud | 04:52 |
vish1 | hvm is right | 04:52 |
vish1 | and it launches | 04:52 |
_anm | ok, nevermind, i just re-read that | 04:52 |
_anm | i suck | 04:52 |
vish1 | but you have to have modeprobe kvm | 04:52 |
vish1 | usr/lib/python*/*-packages/nova/* | 04:53 |
vish1 | anyone know why that is in nova-common.install | 04:53 |
_anm | it launches for me then immediately stops | 04:53 |
vish1 | are you running on hardware? | 04:53 |
vish1 | or in a vm? | 04:53 |
_anm | "hvm" stands for hardware - | 04:53 |
_anm | vish1: on hardware | 04:53 |
_anm | source is from trunk | 04:54 |
vish1 | does /var/log/libvirt/qemu/<id>.log | 04:54 |
vish1 | exist? | 04:54 |
_anm | the directory exists, but no log | 04:55 |
vish1 | in your instances dir | 04:55 |
vish1 | check filesizes on disk kernel and ramdisk | 04:56 |
vish1 | and check to make sure libvirt.xml exists | 04:56 |
vish1 | instances/<instance_id>/ | 04:56 |
*** kashyapc has joined #openstack | 04:56 | |
_anm | hmm, for some reason all my <instance> ids are in ~/openstack/nova, alongside bin/ and | 04:59 |
* _anm scratches head and looks at env variables again | 04:59 | |
*** redbo_ is now known as redbo | 04:59 | |
_anm | and each <instance-id> folder just has a single file, libvirt.xml | 05:00 |
_anm | data/instances in empty, but data/buckets | 05:02 |
_anm | looks fine | 05:02 |
*** ChanServ sets mode: +v redbo | 05:02 | |
_anm | so i need to dig around a bit, clean up, and recheck all my services' startup arguments | 05:03 |
_anm | sigh | 05:03 |
jaypipes | justinsb, vish1: got a little bit more work left, but lots of progress on removing Redis specifics and abstracting the data store and data model: https://code.launchpad.net/~jaypipes/nova/base-datastore-driver | 05:03 |
vish1 | _anm: sounds like path issues | 05:03 |
_anm | yeah | 05:04 |
_anm | i'll reset everythimg and see how it goes | 05:04 |
vish1 | jaypipes: are you familiar with debian packaging? | 05:04 |
jaypipes | vish1: heh, no :( mtaylor is that man. | 05:04 |
vish1 | jaypipes: I think using flags for min and max ports should ultimately go away | 05:10 |
jaypipes | vish1: don't disagree. :) | 05:10 |
jaypipes | vish1: that wasn't the point of my patch, though ;) | 05:10 |
vish1 | jaypipes: if we have a general idea of a pool of objects, we can create admin-type commands for adding objects to the pool | 05:10 |
jaypipes | yep | 05:10 |
*** pvo has quit IRC | 05:10 | |
vish1 | jaypipes: understood, i just want keep resource pools in mind | 05:11 |
jaypipes | vish1: indeed. | 05:11 |
vish1 | jaypipes: also the min max method is really unfeasible for things that appear and disappear often | 05:13 |
jaypipes | vish1: not sure I follow you... | 05:13 |
vish1 | so say i allocate the first 200 ports | 05:14 |
vish1 | then port 17, 25, and 143 get freed | 05:14 |
vish1 | there is no way to reuse them | 05:14 |
vish1 | if you are incrementing min available | 05:14 |
*** jhalogen has quit IRC | 05:15 | |
*** lbieber has quit IRC | 05:16 | |
vish1 | that is the conceptual power of a pool and popping off the pool | 05:16 |
vish1 | which we lose | 05:16 |
vish1 | in general I like where you're going with PersistentObject | 05:17 |
vish1 | though | 05:17 |
vish1 | :) | 05:17 |
mtaylor | jaypipes, vish1 aroo? | 05:19 |
jaypipes | vish1: ah, gotcha. I'll refactor that. Just the input I needed :) | 05:20 |
jaypipes | vish1: thx | 05:20 |
*** lbieber has joined #openstack | 05:21 | |
vish1 | jaypipes: cool. I think a base class that handles pooled objects and has atomic pops would be great 'ResourcePool?' because we need it in a few places relating to networking | 05:21 |
jaypipes | vish1: yep | 05:22 |
vish1 | mtaylor: there is a weird line in nova-common.install which is breaking my deb package building | 05:22 |
*** miclorb_ has joined #openstack | 05:22 | |
vish1 | do you have any idea what this is supposed to do? -usr/lib/python*/*-packages/nova/* | 05:22 |
mtaylor | vish1: looking | 05:23 |
*** lbieber has quit IRC | 05:23 | |
vish1 | i get this error: dh_install: nova-common missing files (usr/lib/python*/*-packages/nova/*), aborting | 05:23 |
vish1 | i'm building from within a venv | 05:24 |
mtaylor | that shouldn't affect the debuild | 05:24 |
mtaylor | but I don't know why someone would attempt to ... | 05:24 |
mtaylor | just remove that | 05:24 |
mtaylor | python file installation should be happening with python support | 05:25 |
vish1 | ok it is probably legacy from some old installscript | 05:25 |
vish1 | i'll make a patch that removes it | 05:25 |
vish1 | mtaylor: do you generally use bzr-builddeb? | 05:25 |
mtaylor | vish1: always. | 05:25 |
mtaylor | vish1: it's on my todo list to do the same stuff to the nova tree for bzr bd that I did for the swift tree | 05:26 |
vish1 | can you give me a two minute overview of how to do it? | 05:26 |
vish1 | we've been using the builddeb script that is in nova | 05:26 |
_anm | so, bzr merge trunk is the same as svn update? I noticed a bug I was waiting on was fixed in the nova trunk, and want to fetch it | 05:26 |
mtaylor | vish1: sure. | 05:26 |
vish1 | which uses dpkg-buildpackage -b -rfakeroot -tc -uc -D | 05:26 |
mtaylor | _anm: you may just be able to do bzr pull | 05:26 |
mtaylor | vish1: actually just try running bzr bd | 05:27 |
_anm | bzr pull, ok | 05:27 |
vish1 | bzr bd | 05:27 |
vish1 | Building using working tree | 05:27 |
vish1 | Looking for a way to retrieve the upstream tarball | 05:27 |
vish1 | Using apt to look for the upstream tarball. | 05:27 |
vish1 | apt could not find the needed tarball. | 05:27 |
vish1 | Trying to use get-orig-source to retrieve needed tarball. | 05:27 |
vish1 | dh get-orig-source | 05:27 |
vish1 | dh: Unknown sequence get-orig-source (choose from: binary binary-arch binary-indep build clean install) | 05:27 |
vish1 | make: *** [get-orig-source] Error 1 | 05:27 |
vish1 | Trying to run get-orig-source rule failed | 05:27 |
vish1 | No watch file to use to retrieve upstream tarball. | 05:27 |
vish1 | bzr: ERROR: Unable to find the needed upstream tarball: nova_0.2.3.orig.tar.gz. | 05:27 |
PiotrSikora | someone should really add git interface to launchpad ;) | 05:27 |
mtaylor | PiotrSikora: that's probably not going to happen | 05:28 |
*** littleidea has quit IRC | 05:28 | |
vish1 | github.com/termie/bzr-git-ng | 05:28 |
PiotrSikora | mtaylor: yeah, i figured this out ;) | 05:28 |
mtaylor | PiotrSikora: well, sorry, I suppose the other thing could - like bzr-git-ng | 05:28 |
PiotrSikora | vish1: i was just about to post this :) | 05:28 |
redbo | mtaylor: how come only the main swift package is in our ppa, not any of the others | 05:28 |
_anm | wow, the entire src tree has changed drstically since friday - things are moving fast! | 05:29 |
mtaylor | redbo: oh - it didn't build properly | 05:29 |
mtaylor | redbo: there are actually no swift packages | 05:29 |
mtaylor | vish1: so, eventually there are a few things we need to do to get the bzr bd workflow up and going and happy... | 05:30 |
vish1 | cool | 05:31 |
vish1 | i'm trying to convert our buildprocess over to use the new repo | 05:31 |
mtaylor | vish1: for now I'd keep using your builddeb.sh script, and I'll see if I can make a tree for you to look at with the other changes | 05:31 |
mtaylor | ah | 05:31 |
vish1 | so we can deploy all of the awesome changes | 05:31 |
vish1 | :) | 05:31 |
vish1 | that sounds good | 05:31 |
vish1 | i think this will work for the moment | 05:32 |
mtaylor | cool | 05:32 |
mtaylor | vish1: once we get the bzr bd thing sorted, I can have hudson publish debs on every trunk push ... | 05:32 |
mtaylor | redbo: looking in to the swift ppa packages right now... | 05:33 |
mtaylor | redbo: ok... the ppa with the nightly builds in it is ok | 05:34 |
mtaylor | redbo: although it hasn't built in two days - so I guess they're not nightlies | 05:34 |
mtaylor | https://edge.launchpad.net/~swift-core/+archive/trunk/+packages | 05:34 |
* mtaylor will fix | 05:34 | |
*** veszig` is now known as veszig | 05:36 | |
_anm | is fake_users=1 noe deprecated? | 05:38 |
_anm | after refreshing my souce tree nova-api balks | 05:38 |
_anm | s/noe/now/ | 05:38 |
mtaylor | redbo: ok. I hadn't uploaded new packages to the ppa since I re-worked the branches :) sorry bout that | 05:50 |
vish1 | yes fake_users=1 is gone | 05:55 |
_anm | now some nova-apps accept --nodaemonize, but others require --nodaemon | 05:55 |
* jaypipes calling it a night...gotta add get_keys() call to datastore drivers tomorrow... | 05:56 | |
_anm | should I report that as a bug, or let it slide for a while? | 05:56 |
vish1 | the only one that still uses --nodaemonize is nova-api | 05:56 |
vish1 | and hopefully that will die soon | 05:56 |
_anm | ok | 05:56 |
vish1 | when we get rid of tornado | 05:56 |
_anm | just trying to keep http://wiki.openstack.org/InstallFromSource up to date with trunk - a pretty hard task | 05:57 |
vish1 | aye | 05:57 |
vish1 | although the virtualenv stuff makes installing pretty easy | 05:57 |
vish1 | i did it with about 4 commands | 05:57 |
*** adjohn has joined #openstack | 05:57 | |
vish1 | at least to get to tests passing | 05:58 |
vish1 | actually running the system needs a little more coaxing | 05:58 |
vish1 | :)_ | 05:58 |
vish1 | s/:)_/:)/ | 05:58 |
_anm | virtualenv? whats this? | 06:01 |
vish1 | it is magic | 06:02 |
vish1 | :) | 06:02 |
* _anm is curious how vish1 installs nova with 4 commands | 06:03 | |
vish1 | let me check | 06:03 |
vish1 | so in the current trunk | 06:03 |
vish1 | i think i had to do apt-get install virtualenv | 06:04 |
vish1 | apt-get install pip | 06:04 |
vish1 | ah no | 06:04 |
vish1 | it is easy_install virtualenv | 06:04 |
vish1 | easy_install pip | 06:04 |
vish1 | apt-get build-dep python-m2crypto | 06:05 |
vish1 | python tools/install_venv.py | 06:06 |
vish1 | i think were the four | 06:06 |
vish1 | make test works | 06:06 |
_anm | hmmm, nice | 06:08 |
_anm | ok, i redid everything and have this: | 06:08 |
_anm | vish1: http://pastebin.org/421498 | 06:09 |
_anm | the instance shows pending for a while, then shutdown | 06:09 |
vish1 | and you get no logs in libvirt/qemu right? | 06:10 |
_anm | /var/log/libvirt/qemu is still empty | 06:10 |
vish1 | what is the output of -- dpkg -l | grep libvirt | 06:10 |
vish1 | are you running all of the daemons as root? | 06:11 |
*** greenisus has joined #openstack | 06:11 | |
_anm | root@openstack:~/openstack/data/networks# dpkg -l | grep libvirt | 06:11 |
_anm | ii libvirt-bin 0.7.5-5ubuntu27 the programs for the libvirt library | 06:11 |
_anm | ii libvirt0 0.7.5-5ubuntu27 library for interfacing with different virtu | 06:11 |
_anm | ii python-libvirt 0.7.5-5ubuntu27 libvirt Python bindings | 06:11 |
_anm | all daemons are running as root | 06:12 |
vish1 | ok you have old version of libvirt | 06:12 |
vish1 | but that may or may not be the issue | 06:12 |
vish1 | i would definitely install the 8.1 version from soren's ppa | 06:12 |
_anm | ok | 06:12 |
vish1 | but first | 06:12 |
vish1 | try | 06:12 |
vish1 | from the instance dir | 06:12 |
vish1 | virsh create libvirt.xml | 06:12 |
vish1 | and see if you get any errors | 06:13 |
_anm | error: Failed to create domain from libvirt.xml | 06:15 |
_anm | error: internal error no supported architecture for os type 'hvm' | 06:15 |
_anm | update to 8.2? | 06:15 |
*** matiu has quit IRC | 06:16 | |
vish1 | nope | 06:16 |
vish1 | you just need to | 06:16 |
vish1 | modprobe kvm | 06:16 |
vish1 | and you should be good | 06:16 |
vish1 | assuming your hadware supports virtualization extensions | 06:17 |
vish1 | and it is turned on in the bios | 06:17 |
_anm | it does, and should be | 06:17 |
_anm | vish1: I'll be a you-know-what | 06:19 |
vish1 | works? | 06:19 |
_anm | the tech I asked to build this server built it on an old dell 1750 | 06:19 |
_anm | 1750 = no HW Virtualization | 06:20 |
vish1 | haha | 06:20 |
_anm | support | 06:20 |
vish1 | use qemu | 06:20 |
* _anm bashes head on table | 06:20 | |
_anm | it'll still work? | 06:20 |
vish1 | just requires a change to libvirt.xml.template | 06:20 |
vish1 | yeah | 06:20 |
vish1 | it actually runs inside a vm | 06:20 |
vish1 | with qemu | 06:20 |
_anm | just change type to qemu? | 06:20 |
_anm | from hvm? | 06:21 |
vish1 | i think the install guide has a sed for it | 06:21 |
_anm | yea it does | 06:21 |
vish1 | don't remember exactly | 06:21 |
vish1 | i think it is qemu | 06:21 |
vish1 | you can also change the libvirt.xml in the dir | 06:21 |
vish1 | and try a virsh create to see if it works | 06:21 |
vish1 | ok not quite 4 commands | 06:27 |
vish1 | still have to install redis separately from the ppa or source | 06:27 |
_anm | ok, i now have a log in /var/log/libvirt/qemo | 06:28 |
_anm | qemu | 06:28 |
vish1 | yay | 06:29 |
vish1 | but still not running? | 06:29 |
_anm | error: Failed to add tap interface to bridge 'br0': No such device | 06:30 |
_anm | using virsh create | 06:30 |
_anm | for some reason i have a br100, but not br0 | 06:31 |
_anm | br100 wasnt there 20 mins ago | 06:31 |
_anm | not sure what created br100 | 06:32 |
*** littleidea has joined #openstack | 06:37 | |
_anm | this is how the api was started: | 06:38 |
vish1 | check the libvirt.xml | 06:38 |
_anm | python bin/nova-api --ca_path=/home/nikmartin/openstack/nova/CA --keys_path=/home/nikmartin/openstack/nova/../data/keys --simple_network=true --simple_network_gateway=192.168.1.1 --simple_network_netmask=255.255.255.0 --simple_network_network=192.168.1.0 --simple_network_ips=192.168.1.220,192.168.1.221,192.168.1.222 --simple_network_bridge=br0 --simple_network_broadcast=192.168.1.255 --nodaemonize --verbose start | 06:38 |
_anm | ok | 06:38 |
vish1 | does it specify br0? | 06:38 |
vish1 | you are using simple_network? | 06:39 |
vish1 | with simple_network you have to create the bridge yourself | 06:39 |
_anm | yes | 06:39 |
_anm | ah | 06:39 |
*** littleidea has quit IRC | 06:39 | |
_anm | what happens if i omit simple_network=true, it does a bunch of mgic? | 06:40 |
vish1 | yeah assuming bridge_dev is set properly | 06:40 |
vish1 | it does some sexy magic | 06:40 |
vish1 | creates a vlan | 06:40 |
vish1 | creates a bridge | 06:40 |
vish1 | starts a dhcp server | 06:40 |
_anm | i'm trying to document the easiest way for someone to get up and running using source from trunk | 06:40 |
_anm | to save them the pain I've gone through | 06:41 |
_anm | :) | 06:41 |
_anm | i'll try that, but first, sleep. Im on CST time, and have been up since 0600 Monday. It's now 01:42 Tuesday | 06:42 |
vish1 | oh my | 06:43 |
vish1 | night! | 06:43 |
*** miclorb_ has quit IRC | 06:43 | |
_anm | vish1: thanks for your help! I'm glad to be part of this effort! | 06:45 |
vish1 | _anm: np, we need to decrease the barrier to entry for sure | 06:46 |
vish1 | hopefully we have a run script that starts up all workers locally for testing | 06:47 |
vish1 | we're getting there | 06:47 |
*** rajijoom has quit IRC | 06:49 | |
*** miclorb has joined #openstack | 06:52 | |
*** rajijoom has joined #openstack | 06:53 | |
*** BartVB has quit IRC | 06:54 | |
*** nettok has quit IRC | 06:58 | |
*** vish1 has quit IRC | 07:09 | |
*** jsgotangco has quit IRC | 07:12 | |
*** z0 has joined #openstack | 07:18 | |
*** mray has quit IRC | 07:26 | |
*** mrayzenoss has joined #openstack | 07:26 | |
*** mrayzenoss is now known as mray | 07:26 | |
*** mray has joined #openstack | 07:26 | |
*** viewlogic has quit IRC | 07:39 | |
*** almaisan-away is now known as al-maisan | 07:44 | |
*** metoikos has quit IRC | 07:51 | |
*** calavera has joined #openstack | 07:54 | |
*** allsystemsarego has joined #openstack | 07:55 | |
*** jtdowney has joined #openstack | 08:01 | |
*** BartVB has joined #openstack | 08:05 | |
*** cw_ is now known as cw | 08:09 | |
*** greenisus has quit IRC | 08:12 | |
*** miclorb has quit IRC | 08:15 | |
*** _anm has quit IRC | 08:26 | |
*** miclorb_ has joined #openstack | 08:43 | |
*** adjohn_ has joined #openstack | 09:03 | |
*** adjohn has quit IRC | 09:06 | |
*** adjohn_ is now known as adjohn | 09:06 | |
*** avsm has joined #openstack | 09:16 | |
*** miclorb_ has quit IRC | 09:17 | |
*** miclorb has joined #openstack | 09:17 | |
*** BartVB has quit IRC | 09:19 | |
*** miclorb has quit IRC | 09:22 | |
soren | termie: Did you say you were in the same $TZ as me? | 09:25 |
*** metoikos has joined #openstack | 09:38 | |
*** rnewson has joined #openstack | 09:41 | |
*** jtdowney has quit IRC | 09:42 | |
*** terje has quit IRC | 09:42 | |
*** metoikos has quit IRC | 09:49 | |
*** BartVB has joined #openstack | 09:56 | |
*** BartVB1 has joined #openstack | 10:05 | |
*** BartVB has quit IRC | 10:06 | |
*** jtdowney has joined #openstack | 10:11 | |
*** jtdowney has joined #openstack | 10:11 | |
*** jtdowney has quit IRC | 10:13 | |
*** jtdowney has joined #openstack | 10:14 | |
*** metoikos has joined #openstack | 10:19 | |
uvirtbot | New bug: #610402 in nova ""make test" requires virtualenv" [Undecided,New] https://launchpad.net/bugs/610402 | 10:20 |
*** poli has quit IRC | 10:31 | |
*** gasbakid has joined #openstack | 10:33 | |
*** jonesy_ has quit IRC | 10:33 | |
*** _buck has quit IRC | 10:41 | |
*** poli has joined #openstack | 10:42 | |
*** jtdowney has quit IRC | 10:42 | |
*** adjohn_ has joined #openstack | 10:44 | |
*** adjohn has quit IRC | 10:44 | |
*** adjohn_ is now known as adjohn | 10:44 | |
*** _buck has joined #openstack | 11:03 | |
*** jonesy_ has joined #openstack | 11:10 | |
*** justinsheehy has quit IRC | 11:24 | |
*** bahadir has joined #openstack | 11:24 | |
*** justinsheehy has joined #openstack | 11:29 | |
*** mattf has quit IRC | 11:42 | |
*** al-maisan is now known as almaisan-away | 11:53 | |
*** justinsheehy has quit IRC | 11:53 | |
*** rajijoom has quit IRC | 12:14 | |
*** justinsheehy has joined #openstack | 12:26 | |
*** aliguori has quit IRC | 12:34 | |
termie | soren: yeah, although i've been staying up too late, i am in berlin | 12:37 |
termie | soren: i'l be in cph aug 23 - 30 or thereabouts, too | 12:37 |
*** rajijoom has joined #openstack | 12:47 | |
*** adrian_otto has quit IRC | 12:49 | |
*** bahadir has quit IRC | 12:53 | |
*** adrian_otto has joined #openstack | 12:54 | |
*** jsmith-away is now known as jsmith | 12:55 | |
*** aliguori has joined #openstack | 12:55 | |
*** silassewell has joined #openstack | 13:02 | |
*** poli has quit IRC | 13:05 | |
*** lbieber has joined #openstack | 13:11 | |
uvirtbot | New bug: #610434 in swift "Docs off a little in swift.common.client.get_auth" [Undecided,New] https://launchpad.net/bugs/610434 | 13:11 |
*** croteb has quit IRC | 13:14 | |
kirkland | howdy -- i'm looking for initial ubuntu packages for swift and nova, and I've found https://edge.launchpad.net/~soren/+archive/nova and https://edge.launchpad.net/~swift-core/+archive/ppa | 13:17 |
kirkland | are these the best two to experiment with at the moment? | 13:17 |
termie | kirkland: i remember vishvananda mentioning something about trying to get the nova packaging working right | 13:18 |
termie | kirkland: but soren's may have been a manually created one | 13:19 |
termie | so those options seem reasonable | 13:19 |
*** rajijoom has quit IRC | 13:20 | |
soren | termie: Oh. What's the occasion? | 13:26 |
soren | kirkland: I'm /reasonably/ confident that dpkg-buildpackage from my twisted-web-s3-server branch will do the right thing. | 13:28 |
*** mray has quit IRC | 13:34 | |
*** mattf has joined #openstack | 13:41 | |
creiht | kirkland: https://launchpad.net/~swift-core/+archive/trunk is probably the best place for packages for swift at the moment | 13:45 |
creiht | But I'm not certain if the swift packages are building right yet | 13:45 |
creiht | We are still working through some of the packaging stuff | 13:45 |
*** lbieber has quit IRC | 13:46 | |
*** adjohn has quit IRC | 13:46 | |
*** mcgurrin has quit IRC | 13:47 | |
*** pvo has joined #openstack | 13:48 | |
*** pvo has joined #openstack | 13:48 | |
*** ChanServ sets mode: +v pvo | 13:48 | |
*** adrian_otto has quit IRC | 13:50 | |
*** gundlach has joined #openstack | 13:55 | |
kirkland | creiht: cool, thanks | 14:09 |
*** mrayzenoss has joined #openstack | 14:11 | |
*** mrayzenoss is now known as mray | 14:11 | |
*** mray has joined #openstack | 14:11 | |
*** mray has quit IRC | 14:15 | |
*** mrayzenoss has joined #openstack | 14:16 | |
anm_ | kirkland: This is a fairly accurate doc on getting nova running, but the source is moving VERY fast; I'm doing my best to keep it current: http://wiki.openstack.org/InstallFromSource | 14:17 |
anm_ | but it's all source based, no packages | 14:17 |
anm_ | again, because the source is moving very fast - 30-50 revs a day | 14:17 |
*** yuravk has joined #openstack | 14:18 | |
*** mrayzenoss is now known as mray | 14:19 | |
yuravk | Hi. I'm getting en error while uploading a bundle to non existing bucket | 14:22 |
yuravk | could somebody please assist me on how to create new bucket ? | 14:23 |
notmyname | yuravk: in swift or in nova? | 14:23 |
yuravk | nova | 14:23 |
yuravk | notmyname: nova | 14:25 |
*** tr3buchet has joined #openstack | 14:26 | |
notmyname | yeah. sorry. I'm a swift dev. I don't know about nova details. I was hoping one of those devs would jump in :-) | 14:26 |
pvo | anm_: there is an installer in the tools dir that mtaylor was working on yesterday | 14:26 |
pvo | anm_: it should setup your virtenv and python packages | 14:26 |
anm_ | yuravk: thats a known bug: | 14:27 |
anm_ | look at bug 607527 | 14:27 |
uvirtbot | Launchpad bug 607527 in nova "VPN not created for auto-created project" [Low,Fix released] https://launchpad.net/bugs/607527 | 14:27 |
anm_ | maybe its fixed! pull latest source | 14:28 |
anm_ | sorry, wrong bug | 14:28 |
anm_ | yuravk: hang on a sec | 14:28 |
anm_ | bug 607541 | 14:29 |
uvirtbot | Launchpad bug 607541 in nova "ObjectStore must throw 404 when bucket does not exist (Commit 146)" [High,Fix committed] https://launchpad.net/bugs/607541 | 14:29 |
anm_ | not in trunk yet, but ther's a patch | 14:29 |
yuravk | anm_: I have imported the latest sources today (from Git at http://github.com/openstack/nova.git) | 14:29 |
anm_ | yuravk: yes, this patch is not merged into the trunk yet | 14:29 |
anm_ | https://code.launchpad.net/~justin-fathomdb/nova/bug607541/+merge/30357/+preview-diff/+files/preview.diff | 14:29 |
*** mattf is now known as _mattf | 14:30 | |
anm_ | yuravk: alternatley, if you have an S3 tol that can simply create a bucket, you can use that as a workaround until this patch is committed to trunk | 14:31 |
anm_ | s/tol/tool/ | 14:31 |
anm_ | props to justinsb | 14:31 |
yuravk | anm_: unfortunately i'm not common with S3 tool(s) ... | 14:36 |
*** sirp2 has joined #openstack | 14:37 | |
anm_ | yuravk: then you'll need to patch the source using the patch I listed above, sorry :( | 14:40 |
*** ianw has joined #openstack | 14:40 | |
*** ianw has quit IRC | 14:42 | |
*** ralph has quit IRC | 14:42 | |
yuravk | anm_: ok, thanks. Just one more stupid question: is there any web interface for Nova to manage Clouds, VMs, etc. ? | 14:43 |
*** gustavomzw has joined #openstack | 14:43 | |
*** almaisan-away is now known as al-maisan | 14:44 | |
anm_ | yuravk: in theory, I think any Amazon S3 and EC2 compatible interface will work, but I don't think there's 100% compatibility at the moment - | 14:45 |
anm_ | so far I've used s3cmd to list buckets, create buckets, etc | 14:46 |
anm_ | but have not tried to wire up something like ElasticFox | 14:46 |
anm_ | the tool has to allow changing the URL that it connects to | 14:47 |
anm_ | there was talk yesterday that a tool like cyberduck should work | 14:47 |
yuravk | anm_: oh, thank you. I'm going to get Amazon S3 account ... | 14:49 |
anm_ | yuravk: You shouldn't need an S3 account | 14:49 |
yuravk | anm_: but s3cmd need their keys ? | 14:52 |
*** metoikos has quit IRC | 14:52 | |
anm_ | no, you'll set the keys to the ones in your project | 14:52 |
anm_ | If you follow along this page on Installing: | 14:53 |
anm_ | http://wiki.openstack.org/InstallFromSource | 14:53 |
*** littleidea has joined #openstack | 14:53 | |
anm_ | when you run the command: bin/nova-manage ${NOVA_MANAGE_ARGS} project zip ${OPENSTACK_PROJECTNAME} ${OPENSTACK_USERNAME} me/nova.zip | 14:53 |
anm_ | the resulting file, me/nova.zip will have a file named novarc that contains your keys | 14:54 |
*** gasbakid has quit IRC | 14:54 | |
anm_ | you'll then edit the file ~/.s3cfg and add those keys to it, along with changing the amazon url to your nova server's ip/port | 14:55 |
anm_ | it's a little convoluted | 14:55 |
anm_ | the euca-tools taht's available in ubuntu wiull allow you to do the same stuff, you just need to have those keys setup in your environment variables | 14:56 |
anm_ | if you source novarc (. novarc), your shell will have all the keys and paths set | 14:56 |
*** metoikos has joined #openstack | 15:00 | |
yuravk | anm_: I have sources novarc file. And it contain EC2 keys, EC2 and S3 URLs | 15:02 |
*** gasbakid has joined #openstack | 15:07 | |
yuravk | anm_: but I have no idea how to create new bucket with euca ... | 15:08 |
anm_ | I dont think euca-tools has a "make bucket only" command, which is where this bug came from initally | 15:11 |
anm_ | i dont have any other help in that area, except to say you'll have to apply the patch | 15:12 |
anm_ | someone else may have better advice | 15:12 |
*** jhalogen has joined #openstack | 15:12 | |
anm_ | https://bugs.launchpad.net/nova/+bug/607541 | 15:12 |
uvirtbot | Launchpad bug 607541 in nova "ObjectStore must throw 404 when bucket does not exist (Commit 146)" [High,Fix committed] | 15:12 |
*** pvo_ has joined #openstack | 15:16 | |
*** MaxR_ has joined #openstack | 15:16 | |
yuravk | anm_: ok, thanks | 15:18 |
*** gnrfan has joined #openstack | 15:19 | |
*** pvo_ has quit IRC | 15:20 | |
*** dendrobates is now known as dendro-afk | 15:24 | |
*** calavera has quit IRC | 15:26 | |
*** anm_ has quit IRC | 15:30 | |
*** anm_ has joined #openstack | 15:35 | |
*** greenisus has joined #openstack | 15:36 | |
*** drewlande has quit IRC | 15:40 | |
*** gasbakid has quit IRC | 15:42 | |
*** abecc has joined #openstack | 15:42 | |
*** greenisus has quit IRC | 15:51 | |
MaxR_ | w00t Just got permission from the boss to build out half a rack with Swift. :) | 15:54 |
*** ianw has joined #openstack | 15:54 | |
*** jzb has joined #openstack | 15:54 | |
creiht | MaxR_: Nice! | 15:54 |
*** gasbakid has joined #openstack | 15:55 | |
*** jhalogen has quit IRC | 15:57 | |
*** jdmaturen has joined #openstack | 15:58 | |
anm_ | MaxR_: what type hardware are you going to use? | 15:59 |
creiht | MaxR_: And out of curiosity, what are you guys going to use it for? | 16:00 |
*** dendro-afk is now known as dendrobates | 16:17 | |
*** jzb has quit IRC | 16:19 | |
*** tr3buchet has quit IRC | 16:19 | |
*** jdmaturen has joined #openstack | 16:22 | |
*** lbieber has joined #openstack | 16:22 | |
*** jdmaturen has joined #openstack | 16:22 | |
*** metoikos has quit IRC | 16:28 | |
*** silassewell has quit IRC | 16:30 | |
*** ianw has quit IRC | 16:31 | |
*** jzb has joined #openstack | 16:37 | |
*** rajijoom has joined #openstack | 16:39 | |
*** carolinad28 has joined #openstack | 16:41 | |
*** jbarratt has joined #openstack | 16:41 | |
MaxR_ | anm_: Haven't decided if I want to test it with low-end celerons & small hard drives(hardware we have laying around unused basically) or if I want to build out a PowerEdge rack...leaning more towards the celeron path currently as management is less likely to balk if I start throwing more hardware at it. :P | 16:43 |
MaxR_ | creiht: At the moment, just stress testing and benchmarking.. etc | 16:43 |
anm_ | aren't some of the operations in Swift VERY CPU/IO intensive? | 16:44 |
MaxR_ | I dont know.. yet | 16:44 |
anm_ | lowe end celerons may not give you the benchmarks you are looking for, if you are using 10G in the storage network | 16:44 |
MaxR_ | Thus why I'm building out something larger than the SAIO :) | 16:44 |
anm_ | Gotcha | 16:44 |
MaxR_ | Best I can do readily is 1Gbps :/ | 16:45 |
MaxR_ | This is more of a proof-of-concept to management to show it all works outside of a VM/one-server setup | 16:45 |
MaxR_ | But.. step at a time :) | 16:45 |
anm_ | I see, I'm just using this article as reference: http://swift.openstack.org/deployment_guide.html | 16:46 |
anm_ | I've not done it myself either | 16:46 |
MaxR_ | I've already built out the SAIO | 16:46 |
MaxR_ | It was actually pretty swift(drum crash) to get set up | 16:46 |
MaxR_ | I had it all going in probably about an hour | 16:46 |
*** silassewell has joined #openstack | 16:47 | |
*** MaxR_ is now known as MaxR_Lunch | 16:48 | |
MaxR_Lunch | But yeah.. saying that I'm building it out as celerons, and then expecting decent benchmarks may not be realistic. :) Although I can certainly use it as a proof-of-concept (= | 16:49 |
* MaxR_Lunch is away to lunch and will be back later | 16:50 | |
*** vish1 has joined #openstack | 16:53 | |
*** _jcsmith has joined #openstack | 16:54 | |
*** matclayton has joined #openstack | 16:55 | |
*** greenisus has joined #openstack | 16:56 | |
*** vish1 has quit IRC | 17:04 | |
*** algorist has joined #openstack | 17:08 | |
*** mtaylor has quit IRC | 17:10 | |
*** mtaylor has joined #openstack | 17:12 | |
*** ChanServ sets mode: +v mtaylor | 17:12 | |
*** jdmaturen has quit IRC | 17:14 | |
*** tr3buchet has joined #openstack | 17:15 | |
*** avsm has quit IRC | 17:17 | |
*** al-maisan is now known as almaisan-away | 17:20 | |
*** greenisus has quit IRC | 17:27 | |
*** greenisus has joined #openstack | 17:27 | |
*** stewart has quit IRC | 17:30 | |
*** _buck has quit IRC | 17:31 | |
*** kashyapc has quit IRC | 17:31 | |
*** matclayton has left #openstack | 17:37 | |
*** Kami_ has joined #openstack | 17:39 | |
kirkland | known issue? -> http://pastebin.ubuntu.com/469902/ | 17:40 |
*** greenisus_ has joined #openstack | 17:45 | |
vishvananda | kirkland: i haven't seen that one | 17:47 |
vishvananda | my guess is that you have the wrong version of python-daemon or lockfile | 17:48 |
jaypipes | kirkland: haven't seen that, no. I'd file a bug if it's reproduciable. | 17:48 |
jaypipes | vishvananda: afternoon. | 17:48 |
*** greenisus has quit IRC | 17:48 | |
*** greenisus_ is now known as greenisus | 17:48 | |
vishvananda | jaypipes: hi | 17:48 |
jaypipes | vishvananda: how are things at NASA? | 17:50 |
jaypipes | vishvananda: you're in the Bay Area, right? were you at OSCON? | 17:50 |
*** Podilarius has joined #openstack | 17:51 | |
dendrobates | you do not need a lp login to download the code, only to use the 'lp:' shortcut. Disinformation pisses me off! I am updating the wiki now. | 17:52 |
eday | kirkland: I've seen that, needed a newer python-daemon package | 17:54 |
*** gundlach has quit IRC | 17:55 | |
jaypipes | dendrobates: where did you see that? | 17:55 |
jaypipes | kirkland: pls do put in a bug report, though, since we should put the necessary version of python-daemon in our packaging. | 17:56 |
dendrobates | it has been all over the press and in the github mirror announcement. | 17:57 |
dendrobates | and to be fair our instructions require it. | 17:57 |
dendrobates | 100% BS | 17:58 |
creiht | dendrobates: For a while (not sure it is still true), you had to be logged in to launchpad before you could browse to launchpad.net/swift | 17:58 |
dendrobates | that was when it was closed | 17:58 |
creiht | after it was open | 17:58 |
dendrobates | then it was not open | 17:58 |
creiht | You didn't have to be a member, just logged into launchpad | 17:59 |
creiht | dendrobates: still is | 17:59 |
dendrobates | it was obviously broken, that is our fault for having it closed in the first place and renaming the damn thing 50 times. | 17:59 |
dendrobates | mtaylor: can you look into that? | 18:00 |
creiht | I just opened firefox (which I don't normally use), browsed to launchpad.net/swift, and I get forwarded to login | 18:00 |
dendrobates | I just browsed to https://code.launchpad.net/~hudson-openstack/swift/trunk not logged in | 18:00 |
dendrobates | no problems | 18:01 |
dendrobates | what are we doing differently? | 18:01 |
mtaylor | dendrobates: it's just the swift project - and there are issues | 18:02 |
mtaylor | dendrobates: spm is looking in to it- there's also weirdness with the swift mailing list - but it's only swift - the other resources are all fine | 18:02 |
dendrobates | you can go straight th the code and specify the URL in bzr to create a branch | 18:02 |
vishvananda | jaypipes: yes and no | 18:03 |
_0x44 | dendrobates: Requiring people know the URL to the code before they can get it is a UI failure. | 18:03 |
vishvananda | jaypipes: i was in dc during oscon | 18:03 |
dendrobates | anyway I'm creating a bzrwithoutlp page on the wiki | 18:03 |
*** _buck has joined #openstack | 18:04 | |
jaypipes | vishvananda: ah | 18:04 |
dendrobates | _0x44: I'm am not talking about a UI. | 18:04 |
_0x44 | dendrobates: I'm not talking about A UI either. I'm talking about user interface/interaction. If someone needs to know the direct URL for a resource before they can do anything with our code, they won't. | 18:05 |
*** kallistec has joined #openstack | 18:05 | |
mtaylor | bzr branch lp:swift should work for anyone | 18:05 |
_0x44 | Saying "It works because I can browse directly to the URL" misses the point. | 18:05 |
vishvananda | dendrobates: i can bzr branch lp:nova with no login | 18:05 |
mtaylor | and lp:swift | 18:05 |
dendrobates | mtaylor: the lp: command wants a login. | 18:05 |
mtaylor | dendrobates: it doesn't for me | 18:06 |
mtaylor | I just did it from a blank os user with no lp-login configured | 18:06 |
dendrobates | _0x44: so you think someone should be able to create a remote branch without specifying a URL, real bright. | 18:06 |
mtaylor | to be fair - going to http://launchpad.net/swift should bloody work | 18:07 |
dendrobates | mtaylor: hm, maybe I didn't look past the warning | 18:07 |
MaxR_Lunch | I don't see what the big deal is. I didn't have LP account.. I registered one to download the code per the instructions. | 18:07 |
MaxR_Lunch | LP login is the price of admission.. | 18:07 |
*** MaxR_Lunch is now known as MaxR | 18:07 | |
*** MaxR is now known as MaxR_ | 18:07 | |
mtaylor | dendrobates: yeah - it was only a warning | 18:07 |
dendrobates | MaxR_: to contribute yes, but not just to look. | 18:07 |
_0x44 | dendrobates: I'm saying that saying people can get to the code directly from the code URL, but not from the project page is a user interface failure that will not encourage people to contribute to the project. Please stop being obtuse. | 18:08 |
MaxR_ | Isnt it being released as a "developers preview" ? | 18:08 |
*** ianw has joined #openstack | 18:08 | |
MaxR_ | And the instructions I followed were for SAIO, which clearly indicates the instructions are for doing swift development | 18:08 |
MaxR_ | I figured the more public instructions would be released when the software is publically released.. | 18:09 |
MaxR_ | I guess it's a matter of interpretation of intention of the current released version (= | 18:09 |
dendrobates | _0x44: no one is saying anything different, we've broken LP. I am talking about command line. | 18:09 |
mtaylor | well, I'm pinging folks at launchpad about the login thing again | 18:10 |
jaypipes | MaxR_: yeah, there are currrently some issues trying to get Swift completely public...issues that originated with the project being marked "Private/Proprietary" before last Monday. Hopefully, mtaylor and others will have the issues resolved shortly. This isn't something that is typical on LP... | 18:10 |
*** imeyer has joined #openstack | 18:10 | |
mtaylor | turns out we're the first people to have created private projects and then want to take them public :) | 18:10 |
MaxR_ | o_O | 18:10 |
jaypipes | not surprised... | 18:10 |
mtaylor | yeah - not a common workflow :) | 18:10 |
jaypipes | :) | 18:11 |
MaxR_ | I guess most people start public and stay public or private and stay private.. | 18:11 |
*** sdelano has joined #openstack | 18:11 | |
MaxR_ | err projects* | 18:11 |
dendrobates | plus we renamed the thing several times | 18:11 |
mtaylor | several. SEVERAL | 18:11 |
mtaylor | oh the pain | 18:11 |
MaxR_ | For curiousity, what were some of the other names you guys came up with? :) | 18:11 |
dendrobates | MaxR_: carlos spicy wiener | 18:12 |
MaxR_ | LOL | 18:12 |
dendrobates | but it was vetoed by the business | 18:12 |
dendrobates | rcloud was an early one. | 18:12 |
dendrobates | than we went from nova to ozone and back to nova | 18:13 |
dendrobates | mtaylor: so I should hang off on the wiki page? | 18:13 |
dendrobates | err hold off | 18:13 |
* ttx still votes for NuAge, but that's my French showing. | 18:14 | |
mtaylor | dendrobates: I dunno. what were you going to put there? | 18:14 |
mtaylor | dendrobates: simple dev instructions should always just be "bzr branch lp:swift" ... and should really imply that doing a bzr lp-login first would be better, as that way you're using bzr+ssh and not http ... but that bzr branch lp:swift should always work to get the latest code | 18:15 |
dendrobates | mtaylor: Just instructions on using the URL to branch and not lp:* to avoid warnings. | 18:15 |
mtaylor | dendrobates: I would hold off on that. the warnings are always going to be present if someone branches from lp and hasn't done lp-login first | 18:15 |
mtaylor | that has nothing to do with the swift privacy thing - it has to do with the protocol being more efficient if you're logged in | 18:16 |
dendrobates | mtaylor: I agree, I wish there was no warning, though | 18:16 |
mtaylor | dendrobates: well... you'll have to take that up with lifeless ... :) | 18:16 |
*** kallistec has quit IRC | 18:17 | |
dendrobates | that could be arranged, but I have a feeling I'd lose the argument. | 18:17 |
mtaylor | I usually lose arguments with lifeless | 18:17 |
*** stewart has joined #openstack | 18:17 | |
dendrobates | as do most people | 18:18 |
*** kallistec has joined #openstack | 18:19 | |
*** selenamarie has joined #openstack | 18:21 | |
eday | selenamarie: hey there! | 18:24 |
mtaylor | hey selenamarie | 18:25 |
*** jdmaturen has joined #openstack | 18:25 | |
selenamarie | eday, mtaylor: hi :) | 18:27 |
*** gundlach has joined #openstack | 18:27 | |
eday | selenamarie: this is where all the cool kids hang out these days :) | 18:29 |
creiht | MaxR_: Back to your setup, sounds great... Let us know if you run into any issues | 18:29 |
creiht | eday: btw... did you need any help with the eventlet stuff? | 18:29 |
eday | creiht: actually.. if you have a moment... leme paste something | 18:29 |
creiht | cool | 18:30 |
selenamarie | eday: i see... :) | 18:31 |
*** jhalogen has joined #openstack | 18:34 | |
eday | creiht: with more than 10 concurrent connections, it dies: http://pastebin.com/ZcJwaVza I assume ignoring sigpipe is not the default, but that should not be happening either since I'm not hanging up on my end :) | 18:36 |
*** jzb has quit IRC | 18:39 | |
creiht | eday: what platform are you running it on? | 18:40 |
redbo | weird. I don't get any errors with it. | 18:42 |
eday | creiht: ubuntu/lucid | 18:42 |
creiht | hrm | 18:43 |
creiht | eday: what are you hitting it with? | 18:43 |
creiht | (for testing) | 18:43 |
eday | creiht: a echo flooding tool I wrote, part of lp:scalestack. establishes multiple connects and pumps data through it as fast as it can on each | 18:44 |
creiht | eday: what's the easiest way to get/run that? | 18:46 |
* jaypipes remarks that LDAP is a ridiculously obtuse interface. | 18:47 | |
creiht | I guess checkout;make;make instal :) | 18:47 |
*** lbieber has quit IRC | 18:47 | |
eday | creiht: yeah ./config/autorun.sh, ./configure, make | 18:49 |
creiht | eday: that's not what the README says ;) | 18:49 |
eday | creiht: then: ./bin/scalestack p=scalestack/.libs echo_flood_tcp.port=12345 echo_flood_tcp.count=500 v to run the echo client | 18:50 |
eday | creiht: yeah it does, minus the install part :) (tarball doesn't require ./config/autorun.sh, but you're not getting it from one :) | 18:51 |
creiht | ahh | 18:51 |
jaypipes | vishvananda: what's the reason behind having role_ldap_subtree == project_ldap_subtree? | 18:54 |
*** jzb has joined #openstack | 18:55 | |
*** jzb has quit IRC | 18:57 | |
creiht | eday: I'm getting an error that it can't find libevent, I have libevent installed though is there anything else that I need? | 18:57 |
creiht | FATAL [event::libevent] Could not find | 18:58 |
eday | creiht: hm, it must not have detected libevent during configure | 18:58 |
creiht | I don't have -dev installed for it | 18:58 |
eday | creiht: that would be it | 18:58 |
*** blpiatt has joined #openstack | 18:59 | |
*** rajijoom has quit IRC | 18:59 | |
eday | creiht: actually... it may simply be a backlog option issue in listen | 18:59 |
* creiht tries again | 18:59 | |
eday | creiht: because the echo client kills all connections on first error, probably causing all those sigpipes | 18:59 |
creiht | eday: now I get a lot of NOTICE messages, but nothing else | 19:01 |
creiht | eday: that would make sense for the sigpipes | 19:01 |
eday | creiht: yup, that was it | 19:02 |
eday | creiht: server = eventlet.listen(('0.0.0.0', 12345), backlog=10000) | 19:02 |
*** mrayzenoss has joined #openstack | 19:02 | |
*** mrayzenoss has joined #openstack | 19:02 | |
creiht | oh cool... so is your test running bette rnow? | 19:03 |
*** mray has quit IRC | 19:04 | |
*** mrayzenoss is now known as mray | 19:04 | |
eday | creiht: yup, no errors now :) | 19:04 |
creiht | cool | 19:05 |
eday | creiht: simple test with about 5k connections shows about 8s for eventlet and twisted, 3s for the C++ server | 19:05 |
creiht | seems reasonable | 19:06 |
creiht | eventlet performance should be similar to twisted | 19:06 |
creiht | eday: how many of those connections are concurrent? or are they serialized? | 19:06 |
*** littleidea has quit IRC | 19:07 | |
eday | creiht: they're all concurrent, although takes a bit for them all to be accepted, so some close before... I'm going to do some other tests so they are all pumping data for a longer amount of time and compare too | 19:13 |
eday | creiht: will probably write up a blog post or something about it | 19:13 |
*** _jcsmith has quit IRC | 19:19 | |
*** binaryWarrior has joined #openstack | 19:24 | |
anm_ | ok, updating the wiki InstallFromSource instructions, when using --simple_network=true, does a bridge simply have to be created using brctl create brn? or does some other stuff have to be done before nova/libvirt will do something magic? | 19:29 |
creiht | eday: by default the greenpool has 1000 green threads, you might try bumping that up (unless your other benchmarks also use a pool of 1000) | 19:30 |
creiht | I would expect raw tcp to be similar to twisted, in our testing a while back, web requests were a bit more performant than twisted | 19:31 |
*** mcgurrin has joined #openstack | 19:32 | |
eday | creiht: the C++ test uses a single system thread with all non-blocking i/o, twisted is pretty much the same, so no thread pools | 19:34 |
vishvananda | just create a bridge | 19:37 |
vishvananda | but you might want to give it an ip | 19:37 |
*** algorist has quit IRC | 19:37 | |
vishvananda | if you want to communicate host to instance | 19:37 |
vishvananda | i usually give it the .1 so it is the gateway for the instance | 19:38 |
jaypipes | vishvananda: what's the reason behind having role_ldap_subtree == project_ldap_subtree? | 19:39 |
creiht | eday: Well with the eventlet example, by using the greenpool, you are basically saying that it will limit to 1000 concurrent connections at a time | 19:42 |
creiht | if you just want raw handling, and I would get rid of the greenpool, and just call eventlet.spawn_n | 19:43 |
eday | creiht: ok, thanks | 19:47 |
eday | creiht: I'll do that instead | 19:48 |
*** stewart has quit IRC | 19:51 | |
*** algorist has joined #openstack | 19:52 | |
*** silassewell has quit IRC | 19:57 | |
*** dendrobates is now known as dendro-afk | 19:59 | |
*** mray has quit IRC | 20:04 | |
*** dendro-afk is now known as dendrobates | 20:13 | |
*** lbieber has joined #openstack | 20:14 | |
*** mrayzenoss has joined #openstack | 20:15 | |
*** mrayzenoss is now known as mray | 20:15 | |
*** mray has joined #openstack | 20:15 | |
uvirtbot | New bug: #610583 in swift "standardize on optparse" [Low,New] https://launchpad.net/bugs/610583 | 20:17 |
*** driftx has left #openstack | 20:17 | |
vishvananda | jaypipes: it is just a default, they are both groups so we stuck them in the same tree | 20:18 |
vishvananda | jaypipes: but they could go anywhere, that is why it is a flag | 20:18 |
vishvananda | jaypipes: now that i think about it, i think the only place we use role_ldap_subtree is on deleting a user | 20:19 |
jaypipes | vishvananda: OK, thx for the explanation...wasn't sure if it was a bug or not...seemed strange. | 20:19 |
jaypipes | ya | 20:20 |
*** londo has quit IRC | 20:23 | |
*** glange has quit IRC | 20:23 | |
*** _cerberus_ has quit IRC | 20:23 | |
*** termie has quit IRC | 20:23 | |
*** tr3buchet has quit IRC | 20:23 | |
*** jhalogen has quit IRC | 20:23 | |
*** yuravk has quit IRC | 20:23 | |
*** Podilarius has quit IRC | 20:23 | |
*** justinsheehy has quit IRC | 20:23 | |
*** allsystemsarego has quit IRC | 20:23 | |
*** jaker has quit IRC | 20:23 | |
*** redbo has quit IRC | 20:23 | |
*** mjmac has quit IRC | 20:23 | |
*** almaisan-away has quit IRC | 20:23 | |
*** Phil_FL has quit IRC | 20:23 | |
*** jonesy has quit IRC | 20:23 | |
*** anticw has quit IRC | 20:23 | |
*** g0rdy has quit IRC | 20:23 | |
*** gein has quit IRC | 20:23 | |
*** e-vad has quit IRC | 20:23 | |
*** mcgurrin has quit IRC | 20:23 | |
*** sdelano has quit IRC | 20:23 | |
*** BartVB1 has quit IRC | 20:23 | |
*** 36DAAB5CX has quit IRC | 20:23 | |
*** selenamarie has quit IRC | 20:23 | |
*** tjyang has quit IRC | 20:23 | |
*** _0x44 has quit IRC | 20:23 | |
*** filler has quit IRC | 20:23 | |
*** greenisus has quit IRC | 20:23 | |
*** _mattf has quit IRC | 20:23 | |
*** soren has quit IRC | 20:23 | |
*** polvi has quit IRC | 20:23 | |
*** CnC has quit IRC | 20:23 | |
*** eday has quit IRC | 20:23 | |
*** scotticus has quit IRC | 20:23 | |
*** uvirtbot has quit IRC | 20:23 | |
*** jbarratt has quit IRC | 20:23 | |
*** jeremydei has quit IRC | 20:23 | |
*** bytee has quit IRC | 20:23 | |
*** snowboarder04 has quit IRC | 20:23 | |
*** sander__ has quit IRC | 20:23 | |
*** stefano_s has quit IRC | 20:23 | |
*** cparedes has quit IRC | 20:23 | |
*** btorch has quit IRC | 20:23 | |
*** Kami_ has quit IRC | 20:23 | |
*** MaxR_ has quit IRC | 20:23 | |
*** jonesy_ has quit IRC | 20:23 | |
*** jaypipes has quit IRC | 20:23 | |
*** ttx has quit IRC | 20:23 | |
*** cloud0 has quit IRC | 20:23 | |
*** jsmith has quit IRC | 20:23 | |
*** ghchinoy has quit IRC | 20:23 | |
*** ianw has quit IRC | 20:23 | |
*** rnewson has quit IRC | 20:23 | |
*** armmer has quit IRC | 20:23 | |
*** jeevan_ullas has quit IRC | 20:23 | |
*** clayg has quit IRC | 20:24 | |
*** wreese has quit IRC | 20:24 | |
*** justinsb has quit IRC | 20:24 | |
*** benoitc has quit IRC | 20:24 | |
*** larsemil has quit IRC | 20:24 | |
*** ikk has quit IRC | 20:24 | |
*** brainproxy has quit IRC | 20:24 | |
*** admboom has quit IRC | 20:24 | |
*** chilts has quit IRC | 20:24 | |
*** chmouel has quit IRC | 20:24 | |
*** algorist has quit IRC | 20:24 | |
*** z0 has quit IRC | 20:24 | |
*** omar has quit IRC | 20:24 | |
*** dubs has quit IRC | 20:24 | |
*** Garo_ has quit IRC | 20:24 | |
*** comstud has joined #openstack | 20:24 | |
*** ChanServ sets mode: +v comstud | 20:24 | |
*** londo has joined #openstack | 20:25 | |
*** algorist has joined #openstack | 20:25 | |
*** mcgurrin has joined #openstack | 20:25 | |
*** jhalogen has joined #openstack | 20:25 | |
*** selenamarie has joined #openstack | 20:25 | |
*** sdelano has joined #openstack | 20:25 | |
*** ianw has joined #openstack | 20:25 | |
*** Podilarius has joined #openstack | 20:25 | |
*** greenisus has joined #openstack | 20:25 | |
*** Kami_ has joined #openstack | 20:25 | |
*** tr3buchet has joined #openstack | 20:25 | |
*** jbarratt has joined #openstack | 20:25 | |
*** MaxR_ has joined #openstack | 20:25 | |
*** yuravk has joined #openstack | 20:25 | |
*** _mattf has joined #openstack | 20:25 | |
*** justinsheehy has joined #openstack | 20:25 | |
*** jonesy_ has joined #openstack | 20:25 | |
*** BartVB1 has joined #openstack | 20:25 | |
*** rnewson has joined #openstack | 20:25 | |
*** allsystemsarego has joined #openstack | 20:25 | |
*** z0 has joined #openstack | 20:25 | |
*** jaker has joined #openstack | 20:25 | |
*** 36DAAB5CX has joined #openstack | 20:25 | |
*** glange has joined #openstack | 20:25 | |
*** redbo has joined #openstack | 20:25 | |
*** soren has joined #openstack | 20:25 | |
*** mjmac has joined #openstack | 20:25 | |
*** tjyang has joined #openstack | 20:25 | |
*** polvi has joined #openstack | 20:25 | |
*** jaypipes has joined #openstack | 20:25 | |
*** armmer has joined #openstack | 20:25 | |
*** jeevan_ullas has joined #openstack | 20:25 | |
*** almaisan-away has joined #openstack | 20:25 | |
*** Phil_FL has joined #openstack | 20:25 | |
*** jonesy has joined #openstack | 20:25 | |
*** bytee has joined #openstack | 20:25 | |
*** ttx has joined #openstack | 20:25 | |
*** clayg has joined #openstack | 20:25 | |
*** snowboarder04 has joined #openstack | 20:25 | |
*** sander__ has joined #openstack | 20:25 | |
*** stefano_s has joined #openstack | 20:25 | |
*** cparedes has joined #openstack | 20:25 | |
*** jsmith has joined #openstack | 20:25 | |
*** wreese has joined #openstack | 20:25 | |
*** ikk has joined #openstack | 20:25 | |
*** justinsb has joined #openstack | 20:25 | |
*** verne.freenode.net sets mode: +v redbo | 20:25 | |
*** _cerberus_ has joined #openstack | 20:25 | |
*** ghchinoy has joined #openstack | 20:25 | |
*** termie has joined #openstack | 20:25 | |
*** anticw has joined #openstack | 20:25 | |
*** benoitc has joined #openstack | 20:25 | |
*** btorch has joined #openstack | 20:25 | |
*** omar has joined #openstack | 20:25 | |
*** _0x44 has joined #openstack | 20:25 | |
*** larsemil has joined #openstack | 20:25 | |
*** gein has joined #openstack | 20:25 | |
*** filler has joined #openstack | 20:25 | |
*** cloud0 has joined #openstack | 20:25 | |
*** e-vad has joined #openstack | 20:25 | |
*** CnC has joined #openstack | 20:25 | |
*** uvirtbot has joined #openstack | 20:25 | |
*** jeremydei has joined #openstack | 20:25 | |
*** scotticus has joined #openstack | 20:25 | |
*** chmouel has joined #openstack | 20:25 | |
*** g0rdy has joined #openstack | 20:25 | |
*** admboom has joined #openstack | 20:25 | |
*** dubs has joined #openstack | 20:25 | |
*** chilts has joined #openstack | 20:25 | |
*** Garo_ has joined #openstack | 20:25 | |
*** brainproxy has joined #openstack | 20:25 | |
*** eday has joined #openstack | 20:25 | |
*** verne.freenode.net sets mode: +vv _cerberus_ _0x44 | 20:25 | |
*** gundlach has quit IRC | 20:27 | |
*** glange has quit IRC | 20:27 | |
*** _cerberus_ has quit IRC | 20:27 | |
*** termie has quit IRC | 20:27 | |
*** tr3buchet has quit IRC | 20:27 | |
*** jhalogen has quit IRC | 20:27 | |
*** yuravk has quit IRC | 20:27 | |
*** Podilarius has quit IRC | 20:27 | |
*** justinsheehy has quit IRC | 20:27 | |
*** allsystemsarego has quit IRC | 20:27 | |
*** jaker has quit IRC | 20:27 | |
*** redbo has quit IRC | 20:27 | |
*** mjmac has quit IRC | 20:27 | |
*** almaisan-away has quit IRC | 20:27 | |
*** Phil_FL has quit IRC | 20:27 | |
*** jonesy has quit IRC | 20:27 | |
*** anticw has quit IRC | 20:27 | |
*** g0rdy has quit IRC | 20:27 | |
*** gein has quit IRC | 20:27 | |
*** e-vad has quit IRC | 20:27 | |
*** mcgurrin has quit IRC | 20:27 | |
*** sdelano has quit IRC | 20:27 | |
*** BartVB1 has quit IRC | 20:27 | |
*** 36DAAB5CX has quit IRC | 20:27 | |
*** selenamarie has quit IRC | 20:27 | |
*** tjyang has quit IRC | 20:27 | |
*** _0x44 has quit IRC | 20:27 | |
*** filler has quit IRC | 20:27 | |
*** greenisus has quit IRC | 20:27 | |
*** _mattf has quit IRC | 20:27 | |
*** soren has quit IRC | 20:27 | |
*** polvi has quit IRC | 20:27 | |
*** CnC has quit IRC | 20:27 | |
*** eday has quit IRC | 20:27 | |
*** scotticus has quit IRC | 20:27 | |
*** uvirtbot has quit IRC | 20:27 | |
*** jbarratt has quit IRC | 20:27 | |
*** jeremydei has quit IRC | 20:27 | |
*** bytee has quit IRC | 20:27 | |
*** snowboarder04 has quit IRC | 20:27 | |
*** sander__ has quit IRC | 20:27 | |
*** stefano_s has quit IRC | 20:27 | |
*** cparedes has quit IRC | 20:27 | |
*** btorch has quit IRC | 20:27 | |
*** Kami_ has quit IRC | 20:27 | |
*** MaxR_ has quit IRC | 20:27 | |
*** jonesy_ has quit IRC | 20:27 | |
*** jaypipes has quit IRC | 20:27 | |
*** ttx has quit IRC | 20:27 | |
*** cloud0 has quit IRC | 20:27 | |
*** jsmith has quit IRC | 20:27 | |
*** ghchinoy has quit IRC | 20:27 | |
*** londo has quit IRC | 20:27 | |
*** ianw has quit IRC | 20:27 | |
*** rnewson has quit IRC | 20:27 | |
*** armmer has quit IRC | 20:27 | |
*** jeevan_ullas has quit IRC | 20:27 | |
*** clayg has quit IRC | 20:28 | |
*** wreese has quit IRC | 20:28 | |
*** justinsb has quit IRC | 20:28 | |
*** benoitc has quit IRC | 20:28 | |
*** larsemil has quit IRC | 20:28 | |
*** ikk has quit IRC | 20:28 | |
*** brainproxy has quit IRC | 20:28 | |
*** admboom has quit IRC | 20:28 | |
*** chilts has quit IRC | 20:28 | |
*** chmouel has quit IRC | 20:28 | |
*** algorist has quit IRC | 20:28 | |
*** z0 has quit IRC | 20:28 | |
*** omar has quit IRC | 20:28 | |
*** dubs has quit IRC | 20:28 | |
*** Garo_ has quit IRC | 20:28 | |
*** londo has joined #openstack | 20:28 | |
*** algorist has joined #openstack | 20:28 | |
*** mcgurrin has joined #openstack | 20:28 | |
*** jhalogen has joined #openstack | 20:28 | |
*** selenamarie has joined #openstack | 20:28 | |
*** sdelano has joined #openstack | 20:28 | |
*** ianw has joined #openstack | 20:28 | |
*** Podilarius has joined #openstack | 20:28 | |
*** greenisus has joined #openstack | 20:28 | |
*** Kami_ has joined #openstack | 20:28 | |
*** tr3buchet has joined #openstack | 20:28 | |
*** jbarratt has joined #openstack | 20:28 | |
*** MaxR_ has joined #openstack | 20:28 | |
*** yuravk has joined #openstack | 20:28 | |
*** _mattf has joined #openstack | 20:28 | |
*** justinsheehy has joined #openstack | 20:28 | |
*** jonesy_ has joined #openstack | 20:28 | |
*** BartVB1 has joined #openstack | 20:28 | |
*** rnewson has joined #openstack | 20:28 | |
*** allsystemsarego has joined #openstack | 20:28 | |
*** z0 has joined #openstack | 20:28 | |
*** jaker has joined #openstack | 20:28 | |
*** 36DAAB5CX has joined #openstack | 20:28 | |
*** glange has joined #openstack | 20:28 | |
*** redbo has joined #openstack | 20:28 | |
*** soren has joined #openstack | 20:28 | |
*** mjmac has joined #openstack | 20:28 | |
*** tjyang has joined #openstack | 20:28 | |
*** polvi has joined #openstack | 20:28 | |
*** jaypipes has joined #openstack | 20:28 | |
*** armmer has joined #openstack | 20:28 | |
*** jeevan_ullas has joined #openstack | 20:28 | |
*** almaisan-away has joined #openstack | 20:28 | |
*** Phil_FL has joined #openstack | 20:28 | |
*** jonesy has joined #openstack | 20:28 | |
*** bytee has joined #openstack | 20:28 | |
*** ttx has joined #openstack | 20:28 | |
*** clayg has joined #openstack | 20:28 | |
*** snowboarder04 has joined #openstack | 20:28 | |
*** sander__ has joined #openstack | 20:28 | |
*** stefano_s has joined #openstack | 20:28 | |
*** cparedes has joined #openstack | 20:28 | |
*** jsmith has joined #openstack | 20:28 | |
*** wreese has joined #openstack | 20:28 | |
*** ikk has joined #openstack | 20:28 | |
*** justinsb has joined #openstack | 20:28 | |
*** verne.freenode.net sets mode: +v redbo | 20:28 | |
*** _cerberus_ has joined #openstack | 20:28 | |
*** ghchinoy has joined #openstack | 20:28 | |
*** termie has joined #openstack | 20:28 | |
*** anticw has joined #openstack | 20:28 | |
*** benoitc has joined #openstack | 20:28 | |
*** btorch has joined #openstack | 20:28 | |
*** omar has joined #openstack | 20:28 | |
*** _0x44 has joined #openstack | 20:28 | |
*** larsemil has joined #openstack | 20:28 | |
*** gein has joined #openstack | 20:28 | |
*** filler has joined #openstack | 20:28 | |
*** cloud0 has joined #openstack | 20:28 | |
*** e-vad has joined #openstack | 20:28 | |
*** CnC has joined #openstack | 20:28 | |
*** uvirtbot has joined #openstack | 20:28 | |
*** jeremydei has joined #openstack | 20:28 | |
*** scotticus has joined #openstack | 20:28 | |
*** chmouel has joined #openstack | 20:28 | |
*** g0rdy has joined #openstack | 20:28 | |
*** admboom has joined #openstack | 20:28 | |
*** dubs has joined #openstack | 20:28 | |
*** chilts has joined #openstack | 20:28 | |
*** Garo_ has joined #openstack | 20:28 | |
*** brainproxy has joined #openstack | 20:28 | |
*** eday has joined #openstack | 20:28 | |
*** verne.freenode.net sets mode: +vv _cerberus_ _0x44 | 20:28 | |
soren | I hate the internet. | 20:29 |
*** glange has quit IRC | 20:29 | |
*** _cerberus_ has quit IRC | 20:29 | |
*** termie has quit IRC | 20:29 | |
*** tr3buchet has quit IRC | 20:29 | |
*** jhalogen has quit IRC | 20:29 | |
*** yuravk has quit IRC | 20:29 | |
*** Podilarius has quit IRC | 20:29 | |
*** justinsheehy has quit IRC | 20:29 | |
*** allsystemsarego has quit IRC | 20:29 | |
*** jaker has quit IRC | 20:29 | |
*** redbo has quit IRC | 20:29 | |
*** mjmac has quit IRC | 20:29 | |
*** almaisan-away has quit IRC | 20:29 | |
*** Phil_FL has quit IRC | 20:29 | |
*** jonesy has quit IRC | 20:29 | |
*** anticw has quit IRC | 20:29 | |
*** g0rdy has quit IRC | 20:29 | |
*** gein has quit IRC | 20:29 | |
*** e-vad has quit IRC | 20:29 | |
*** mcgurrin has quit IRC | 20:29 | |
*** sdelano has quit IRC | 20:29 | |
*** BartVB1 has quit IRC | 20:29 | |
*** 36DAAB5CX has quit IRC | 20:29 | |
*** selenamarie has quit IRC | 20:29 | |
*** tjyang has quit IRC | 20:29 | |
*** _0x44 has quit IRC | 20:29 | |
*** filler has quit IRC | 20:29 | |
*** greenisus has quit IRC | 20:29 | |
*** _mattf has quit IRC | 20:29 | |
*** soren has quit IRC | 20:29 | |
*** polvi has quit IRC | 20:29 | |
*** CnC has quit IRC | 20:30 | |
*** eday has quit IRC | 20:30 | |
*** scotticus has quit IRC | 20:30 | |
*** uvirtbot has quit IRC | 20:30 | |
*** jbarratt has quit IRC | 20:30 | |
*** jeremydei has quit IRC | 20:30 | |
*** MaxR_ has quit IRC | 20:30 | |
*** jonesy_ has quit IRC | 20:30 | |
*** jaypipes has quit IRC | 20:30 | |
*** ttx has quit IRC | 20:30 | |
*** cloud0 has quit IRC | 20:30 | |
*** jsmith has quit IRC | 20:30 | |
*** ghchinoy has quit IRC | 20:30 | |
*** londo has quit IRC | 20:30 | |
*** ianw has quit IRC | 20:30 | |
*** rnewson has quit IRC | 20:30 | |
*** armmer has quit IRC | 20:30 | |
*** jeevan_ullas has quit IRC | 20:30 | |
*** clayg has quit IRC | 20:30 | |
*** wreese has quit IRC | 20:30 | |
*** justinsb has quit IRC | 20:30 | |
*** benoitc has quit IRC | 20:30 | |
*** larsemil has quit IRC | 20:30 | |
*** ikk has quit IRC | 20:30 | |
*** brainproxy has quit IRC | 20:30 | |
*** admboom has quit IRC | 20:30 | |
*** chilts has quit IRC | 20:30 | |
*** chmouel has quit IRC | 20:30 | |
*** algorist has quit IRC | 20:30 | |
*** z0 has quit IRC | 20:30 | |
*** omar has quit IRC | 20:30 | |
*** dubs has quit IRC | 20:30 | |
*** Garo_ has quit IRC | 20:30 | |
*** londo has joined #openstack | 20:30 | |
*** algorist has joined #openstack | 20:30 | |
*** jhalogen has joined #openstack | 20:30 | |
*** selenamarie has joined #openstack | 20:30 | |
*** sdelano has joined #openstack | 20:30 | |
*** ianw has joined #openstack | 20:30 | |
*** Podilarius has joined #openstack | 20:30 | |
*** greenisus has joined #openstack | 20:30 | |
*** tr3buchet has joined #openstack | 20:30 | |
*** jbarratt has joined #openstack | 20:30 | |
*** MaxR_ has joined #openstack | 20:30 | |
*** yuravk has joined #openstack | 20:30 | |
*** _mattf has joined #openstack | 20:30 | |
*** justinsheehy has joined #openstack | 20:30 | |
*** jonesy_ has joined #openstack | 20:30 | |
*** BartVB1 has joined #openstack | 20:30 | |
*** rnewson has joined #openstack | 20:30 | |
*** allsystemsarego has joined #openstack | 20:30 | |
*** z0 has joined #openstack | 20:30 | |
*** jaker has joined #openstack | 20:30 | |
*** 36DAAB5CX has joined #openstack | 20:30 | |
*** glange has joined #openstack | 20:30 | |
*** redbo has joined #openstack | 20:30 | |
*** soren has joined #openstack | 20:30 | |
*** mjmac has joined #openstack | 20:30 | |
*** tjyang has joined #openstack | 20:30 | |
*** polvi has joined #openstack | 20:30 | |
*** jaypipes has joined #openstack | 20:30 | |
*** armmer has joined #openstack | 20:30 | |
*** jeevan_ullas has joined #openstack | 20:30 | |
*** almaisan-away has joined #openstack | 20:30 | |
*** Phil_FL has joined #openstack | 20:30 | |
*** jonesy has joined #openstack | 20:30 | |
*** ttx has joined #openstack | 20:30 | |
*** clayg has joined #openstack | 20:30 | |
*** jsmith has joined #openstack | 20:30 | |
*** wreese has joined #openstack | 20:30 | |
*** ikk has joined #openstack | 20:30 | |
*** justinsb has joined #openstack | 20:30 | |
*** _cerberus_ has joined #openstack | 20:30 | |
*** ghchinoy has joined #openstack | 20:30 | |
*** termie has joined #openstack | 20:30 | |
*** anticw has joined #openstack | 20:30 | |
*** benoitc has joined #openstack | 20:30 | |
*** omar has joined #openstack | 20:30 | |
*** _0x44 has joined #openstack | 20:30 | |
*** verne.freenode.net sets mode: +vvv redbo _cerberus_ _0x44 | 20:30 | |
*** larsemil has joined #openstack | 20:30 | |
*** gein has joined #openstack | 20:30 | |
*** filler has joined #openstack | 20:30 | |
*** cloud0 has joined #openstack | 20:30 | |
*** e-vad has joined #openstack | 20:30 | |
*** CnC has joined #openstack | 20:30 | |
*** uvirtbot has joined #openstack | 20:30 | |
*** jeremydei has joined #openstack | 20:30 | |
*** scotticus has joined #openstack | 20:30 | |
*** chmouel has joined #openstack | 20:30 | |
*** g0rdy has joined #openstack | 20:30 | |
*** admboom has joined #openstack | 20:30 | |
*** dubs has joined #openstack | 20:30 | |
*** chilts has joined #openstack | 20:30 | |
*** Garo_ has joined #openstack | 20:30 | |
*** brainproxy has joined #openstack | 20:30 | |
*** eday has joined #openstack | 20:30 | |
*** abecc has quit IRC | 20:31 | |
*** kallistec_rulez has joined #openstack | 20:31 | |
MaxR_ | what the... | 20:32 |
soren | mtaylor: I realise now that I did a terrible job explaining the options. a) Do you prefer to have a branch which at the root as rules, changelog, copyright, etc. and stick that in trunk/debian so that they're two separate branches, or b) have a branch of nova that in addition to the stuff that's in trunk, also has debian/? | 20:32 |
tr3buchet | it's not me, i swears! | 20:32 |
soren | I'm leaning towards the latter. | 20:32 |
anm_ | vishvananda: thanks | 20:33 |
*** kallistec has quit IRC | 20:33 | |
gasbakid | why | 20:33 |
gasbakid | ??? | 20:33 |
gasbakid | the irc server has problems ??????????? | 20:33 |
anm_ | netsplit | 20:33 |
MaxR_ | Amazing after 20 years of IRC, netsplits still happen.... | 20:34 |
redbo | if only it was p2p | 20:34 |
anm_ | the problem is irc isnt a mesh | 20:34 |
anm_ | redbo: exactly | 20:34 |
*** tobym has joined #openstack | 20:34 | |
MaxR_ | Probably for the same reason why we still all pause keys on our PC keyboards | 20:34 |
MaxR_ | all have* | 20:35 |
anm_ | not me says the MAc user | 20:35 |
*** binaryWarrior has quit IRC | 20:37 | |
*** _mattf has quit IRC | 20:37 | |
*** littleidea has joined #openstack | 20:40 | |
*** wattersjames has joined #openstack | 20:41 | |
Podilarius | i like my pause key. | 20:45 |
antonym | sounds like the us<->eu link has problems | 20:45 |
antonym | had* | 20:46 |
*** jonesy_ has quit IRC | 20:47 | |
MaxR_ | Amazes me when I see problems that are not addressed for so many years.. | 20:49 |
MaxR_ | like.. why is a hard drive LED on the front of the case? Why isn't it on the keyboard or monitor... ya know.. where we look 99.9% of the time | 20:50 |
MaxR_ | My suspicion is that the designer of the HDD LED was a midget.. therefore the light was at eye level :) | 20:50 |
*** justinsheehy has quit IRC | 20:51 | |
*** carolinad28 has quit IRC | 20:52 | |
*** metoikos has joined #openstack | 20:52 | |
*** sophiap has quit IRC | 20:52 | |
mtaylor | soren: b | 20:54 |
mtaylor | soren: check out what I did with swift | 20:54 |
mtaylor | soren: it's fully split with a trunk and then a trunk+debian, and the trunk+debian is managed wtih bzr bd/import_dsc/merge-upstream | 20:55 |
mtaylor | soren: as per instructions from lifeless :) | 20:55 |
anm_ | who is this lifeless character people talk about? | 20:56 |
soren | Robert Collins. | 20:57 |
soren | mtaylor: Oh, neat. Fancy doing the same for Nova? :) | 20:58 |
*** MaxR_ has quit IRC | 21:00 | |
mtaylor | soren: yes. it's very high on my list to do for today | 21:00 |
mtaylor | soren: actually, I was thinking of working with you on getting the 0.9 version thing cut, and then making sure everything is in the tarball ... | 21:00 |
soren | mtaylor: Sounds good. | 21:01 |
soren | mtaylor: I've got a bracn that just bumps the version. | 21:01 |
soren | mtaylor: I've got another where I've split out the packaging. | 21:01 |
soren | mtaylor: by "split out" I mean removed. | 21:01 |
mtaylor | soren: cool. | 21:02 |
soren | mtaylor: ...and then I have another where I've added it back in. | 21:02 |
mtaylor | soren: cause adding it back in is actually something I'd like to do via import_dsc, rather than by hand | 21:02 |
soren | mtaylor: Ah, ok. | 21:02 |
mtaylor | soren: http://www.advogato.org/person/robertc/diary/130.html | 21:02 |
mtaylor | is the general approach I follow these days | 21:03 |
soren | mtaylor: Oh, nice. Let me try that. | 21:04 |
mtaylor | it makes for a really nice workflow as we make new releases ... stores the tarball within the branch, so all you ever need is the packaging branch and you're good to go | 21:05 |
anm_ | i just updated nova trunk, restarted everything, and now: http://pastebin.org/423166 | 21:06 |
anm_ | oh nevermind, i now see the lockfile I shoudl delete | 21:07 |
*** gasbakid has quit IRC | 21:07 | |
anm_ | sorry, my S/N ration is very high | 21:07 |
soren | mtaylor: Yeah, that pristine-tar thing is magic. | 21:09 |
soren | mtaylor: Ok, I just need to get python setup.py sdist to do the right thing now.. | 21:09 |
mtaylor | soren: I poked at this briefly the other day - I believe we're missing | 21:09 |
mtaylor | soren: yes. sdist missing files :) | 21:10 |
soren | A whole bunch, even. | 21:11 |
* soren curses distutils | 21:11 | |
soren | And setuptools while I'm at it. | 21:11 |
*** davidstrauss has joined #openstack | 21:11 | |
Podilarius | mmmm .. the smell of cooking cloudware. love it. | 21:11 |
anm_ | is there "clean up all these images, buckets, and AMIs script anywhere? nova-compute is trying to find AMIs that don exist, and I dont know where it learns this from | 21:12 |
davidstrauss | What OpenStack components are worth exploring if you merely plan to run on OpenStack-powered clouds? | 21:12 |
anm_ | davidstrauss: compute clouds or storage clouds? | 21:13 |
davidstrauss | anm_: compute | 21:13 |
anm_ | nova | 21:13 |
davidstrauss | anm_: I realize that Nova can *run* clouds, but are there parts useful for using, say, the Rackspace Cloud? | 21:14 |
mtaylor | greenisus: I haven't forgotten about autmation for you yet, btw... | 21:14 |
anm_ | davidstrauss: is in management? | 21:14 |
anm_ | as. as in management | 21:15 |
davidstrauss | anm_: Let's say I wanted to build a custom control panel for provisioning machines on OS-compliant clouds. Would, say, the control panel app be useful to me? | 21:15 |
davidstrauss | anm_: Or do all of these parts assume they're running in a mutually trusting environment? | 21:16 |
anm_ | davidstrauss: I don't know that answer to that yet | 21:16 |
anm_ | I use euca-tools from the Eucalyptus project at the moment | 21:17 |
anm_ | to do all management of instances | 21:17 |
anm_ | just like with Amazon EC2 or S3 | 21:18 |
davidstrauss | anm_: ah, ok | 21:18 |
anm_ | I'm pretty sure the goal of OpenStack is for the management API to be compliant - I'm sure some crafty group is locked in a room as we speak writing an RFC that will become the standard cloud api | 21:19 |
uvirtbot | New bug: #610611 in nova "Subnet allocation gives the same subnet for every user" [Medium,New] https://launchpad.net/bugs/610611 | 21:21 |
davidstrauss | What is the process for becoming an organizational member of OpenStack? | 21:22 |
*** justinsheehy has joined #openstack | 21:22 | |
davidstrauss | justinsheehy: hello fellow cassandra user | 21:23 |
*** rnewson has left #openstack | 21:24 | |
anm_ | davidstrauss: I'm not sure it qualifies as any membership in any organization, but you have to sign an agreement tobe able to contribute code and possibly documentation tot he project | 21:25 |
davidstrauss | anm_: http://openstack.org/community/ | 21:26 |
davidstrauss | anm_: So, contribute to get listed? | 21:26 |
vishvananda | it appears that hudson missed a test fail | 21:26 |
*** _jcsmith has joined #openstack | 21:27 | |
zul | openstack is using hudson for daily builds right? | 21:27 |
davidstrauss | zul: CI, not "daily builds" | 21:27 |
zul | davidstrauss: because launchpad will have that functionaility soon | 21:27 |
*** jtdowney has joined #openstack | 21:28 | |
soren | zul: It does already, but it lacks a "build-on-bzr-commit" thing. | 21:28 |
zul | soren: ah ok | 21:29 |
zul | oops screaming 3 year old back later ;) | 21:29 |
mtaylor | right. as we're using launchpad daily builds already - but will get build-on-commit going via hudson | 21:29 |
mtaylor | (only marginally following along) | 21:29 |
mtaylor | vishvananda: it did? | 21:30 |
mtaylor | vishvananda: which thing? | 21:30 |
*** _buck has quit IRC | 21:31 | |
anm_ | im not sure how a company gets a link and logo there, probably related to $$contributions | 21:34 |
anm_ | davidstrauss: I just show up here: http://wiki.openstack.org/Approved%20Contributors | 21:35 |
davidstrauss | anm_: How can we set up a corporate CLA to streamline our contributions and involvement? | 21:36 |
davidstrauss | https://rackspace.echosign.com/public/hostedForm?formid=8JKVVE557D6L | 21:36 |
davidstrauss | nvm | 21:36 |
anm_ | davidstrauss: :) | 21:36 |
vishvananda | mtaylor: i get a couple errors in NetworkTestCase in trunk | 21:37 |
vishvananda | i put in a fix | 21:37 |
vishvananda | mtaylor: so right now it doesn't check for test errors? | 21:38 |
mtaylor | vishvananda: it does... which is why I'm concerned about this | 21:39 |
mtaylor | vishvananda: I wan't to know if there's something we need to set up differently on the test machine or something ... | 21:39 |
* davidstrauss just signed the CLA. :-) | 21:39 | |
*** jsmith is now known as jsmith-away | 21:40 | |
*** allsystemsarego has quit IRC | 21:40 | |
*** johnpur has joined #openstack | 21:41 | |
*** ChanServ sets mode: +v johnpur | 21:41 | |
soren | mtaylor: What if the tarball just included everything that is in bzr? | 21:43 |
mtaylor | soren: well, it probably should anyway, no? | 21:44 |
*** kallistec_rulez is now known as kallistec | 21:44 | |
soren | mtaylor: bzr ls -R | grep -v /$ | while read f; do echo include $f; done > MANIFEST.in | 21:44 |
soren | mtaylor: ftw? | 21:44 |
mtaylor | soren: works for me - although you will be including a bunch of things in that MANIFEST file which you don't need to | 21:45 |
soren | mtaylor: Like what? | 21:45 |
*** viewlogic has joined #openstack | 21:45 | |
mtaylor | soren: the python files in the nova pacakge | 21:45 |
soren | Oh, all the python stuff? :) | 21:45 |
mtaylor | yeah | 21:45 |
soren | Fair point. | 21:45 |
mtaylor | :) | 21:46 |
vishvananda | reviews requested on the last couple bugfixes i sent in | 21:46 |
soren | mtaylor: I can't really decide if there's any point in attempting to weed those out again. I'd like to be able to regenerate that MANIFEST.in as often as I please without having to fiddle with it. | 21:48 |
anm_ | OH SNAP - i'm actually SSH'd in to an image that I launched from nova! WOOT! | 21:49 |
davidstrauss | Someone ought to write a Bazaar view plugin to make that manifest dynamically. | 21:49 |
anm_ | now what? :D | 21:49 |
soren | anm_: You're done. Take the rest of the day off. :) | 21:49 |
anm_ | more docs to write | 21:49 |
vishvananda | anm_: congrats | 21:54 |
*** BartVB has joined #openstack | 21:55 | |
*** BartVB1 has quit IRC | 21:55 | |
soren | mtaylor: lp:~soren/nova/build-tarball | 21:56 |
davidstrauss | soren: I will be on Copenhagen in late August. What part of Denmark are you in? | 21:56 |
anm_ | I did all that in order to build upon the InstallFest doc, to have a absolutely complete HowTo for an end usr trying to get up to speed using trunk source | 21:57 |
anm_ | http://wiki.openstack.org/InstallFromSource | 21:57 |
mtaylor | soren: well... once we're up and going and import-dsc'd ... debuild will show us if there's files missing... | 21:57 |
soren | davidstrauss: The part not connected to Copenhagen :) Other end of the country. | 21:57 |
soren | mtaylor: There isn't. That was my test case. | 21:57 |
termie | soren: drupalcon | 21:58 |
termie | soren: same reason as davidstrauss, actually ;) | 21:58 |
mtaylor | soren: excellent - I really meant per your regenerate MANIFEST.in comment... but I don't feel strongly on this right now | 21:58 |
anm_ | Is tornado required just for serving http requests? is this functionality being added directly to nova? | 21:59 |
davidstrauss | anm_: tornado is for the object store | 21:59 |
anm_ | oh, ok | 21:59 |
davidstrauss | anm_: not directly part of nova | 21:59 |
termie | anm_: tornado in nova is slated for removal | 21:59 |
termie | anm_: the remaining uses of it are in the api code | 22:00 |
anm_ | ok | 22:00 |
davidstrauss | termie: oh, so it is still a dependency? | 22:00 |
termie | anm_: the goal is to replace it with twisted and then get rid of it | 22:00 |
anm_ | good | 22:00 |
soren | anm_: The object store uses twisted web now. Since a few days ago. | 22:00 |
termie | davidstrauss: yeah, unless it has changed in the past 24 hours or so, only the endpoint/api.py code | 22:00 |
anm_ | not that I dont like it, just anotehr dependency | 22:00 |
soren | I don't like it. | 22:01 |
soren | :) | 22:01 |
soren | It's ok not to. :) | 22:01 |
anm_ | :) | 22:01 |
termie | nor i, lots of wasted time writing support code for it | 22:01 |
*** littleidea has quit IRC | 22:01 | |
termie | to replicate things twisted already had (even if twisted is a bit ugly) | 22:02 |
termie | i haven't checked out 1.0 though, did they add tests? | 22:02 |
*** pvo has quit IRC | 22:02 | |
mtaylor | soren: uh, confused. | 22:03 |
mtaylor | soren: that branch _has_ a debian dir and has no MANIFEST.in ... | 22:03 |
soren | mtaylor: I know. | 22:03 |
mtaylor | ok. | 22:03 |
soren | mtaylor: Look at the diff. | 22:03 |
soren | :) | 22:03 |
mtaylor | just so I didn't miss something | 22:03 |
*** johnpur has quit IRC | 22:03 | |
mtaylor | ah | 22:04 |
soren | mtaylor: The diff should make it obvious, I think. | 22:04 |
soren | make sdist will now do the right thing. | 22:04 |
soren | Correction: It will now do what is currently thought to be the right thing. | 22:04 |
soren | I'm all about the disclaimers. | 22:04 |
* mtaylor needs to make a todo list item to get rid of that make file and replace it with python... has a feeling that at some point someone is going to want this code to work on windows | 22:05 | |
*** pvo has joined #openstack | 22:05 | |
*** pvo has joined #openstack | 22:05 | |
*** ChanServ sets mode: +v pvo | 22:05 | |
termie | mtaylor: is there any expectation that this code could possibly work on windows? | 22:06 |
soren | termie: Yes. | 22:07 |
termie | mtaylor: i had none and as such didn't bother writing it in python instead | 22:07 |
mtaylor | termie: hell man, I dunno. I just know there's always _someone_ who wants windows | 22:07 |
mtaylor | I think it's fine for now | 22:07 |
termie | soren: orly? that seems a bit far off | 22:07 |
soren | termie: Quite :) | 22:07 |
mtaylor | but probably both of those functions could, when we're bored, be changed to be setup.py commands | 22:07 |
mtaylor | _SO_ not important right now | 22:08 |
soren | termie: Someone always wants to run stuff on Windows. | 22:08 |
vishvananda | i have an approved patch in that is now dependant on a bug fix. If anyone feels like reviewing the bugfix so merge can proceed, please do so here https://code.launchpad.net/~vishvananda/nova/lp610611/+merge/31092 | 22:08 |
termie | mtaylor: i don't necessarily agree with adding extra commands like that to setup.py so let's defer this conversation ;) | 22:08 |
vishvananda | mtaylor: microsoft wants to support azure with openstack | 22:09 |
mtaylor | termie: totally. I'm not saying I do agree that it's the right thing to do | 22:09 |
termie | vishvananda: well, most windows boxes will run linux just fine | 22:09 |
mtaylor | termie: only that Makefile and Windows == fail | 22:09 |
mtaylor | ++ | 22:09 |
vishvananda | termie: lol | 22:10 |
termie | vishvananda: heya... want ot write a test for that bugfix? | 22:12 |
termie | vishvananda: seems like a data integrity bug | 22:13 |
vishvananda | termie: tests were failing for me | 22:22 |
vishvananda | termie: not sure why they weren't failing on hudson | 22:22 |
*** littleidea has joined #openstack | 22:23 | |
mtaylor | we need a more comprehensive and disgusting test suite for hudson to run... something with multiple boxes and stuff | 22:23 |
mtaylor | to really beat the hell out of it | 22:23 |
vishvananda | soren: i have no idea about the tarball patch... looks good but outside what i have done before so I'm going to let others review it | 22:27 |
*** littleidea has quit IRC | 22:28 | |
*** mtaylor has quit IRC | 22:31 | |
*** metoikos has quit IRC | 22:32 | |
soren | vishvananda: No problem. | 22:38 |
*** pvo has quit IRC | 22:38 | |
*** sophiap has joined #openstack | 22:38 | |
soren | vishvananda: Did I explicitly ask you to review it? | 22:38 |
*** jonesy_ has joined #openstack | 22:38 | |
soren | I didn't mean to. | 22:38 |
blpiatt | mtaylor, do you need hardware for that test suite to run on? | 22:39 |
*** tr3buchet has quit IRC | 22:40 | |
*** stewart has joined #openstack | 22:47 | |
*** dendrobates is now known as dendro-afk | 22:49 | |
*** mtaylor has joined #openstack | 22:49 | |
*** ChanServ sets mode: +v mtaylor | 22:49 | |
*** pvo has joined #openstack | 22:49 | |
*** pvo has joined #openstack | 22:49 | |
*** ChanServ sets mode: +v pvo | 22:49 | |
*** mray has quit IRC | 22:50 | |
*** miclorb has joined #openstack | 22:50 | |
*** sophiap has quit IRC | 22:51 | |
*** silassewell has joined #openstack | 22:52 | |
soren | mtaylor: Almost got the debian-split thing sorted out. | 22:52 |
*** BartVB has quit IRC | 22:52 | |
mtaylor | soren: excellent | 22:55 |
pvo | mtaylor: do you need boxes? | 22:55 |
pvo | where is hudson running now? | 22:56 |
mtaylor | pvo: probably | 22:56 |
mtaylor | it's running on a cloud server right now | 22:56 |
pvo | ok, so it is local...ish. | 22:56 |
mtaylor | but we'll want some boxes to set up integration testing at some point | 22:56 |
pvo | you're probably in ORD if you just built. | 22:56 |
pvo | yea, we'll need to find some dedicated gear for the different hypervisor integration tests. | 22:56 |
pvo | unless we're yo dawg-ing it | 22:57 |
mtaylor | pvo: yup. and, once we have boxes to run the integration testing on, we'll just fire up hudson slaves there, so it wouldn't be important for the master to be local per-se | 22:58 |
pvo | do the hudson slaves need to be public or just accessible from hudson-master? | 22:59 |
soren | mtaylor: lp:~soren/nova/remove-debian-dir and lp:~soren/nova/ubuntu-packaging | 22:59 |
pvo | I haven't played with that configuration yet | 22:59 |
soren | mtaylor: Oh, let me move that packaging one to nova-core. | 22:59 |
mtaylor | pvo: just accessible from hudson-master | 23:00 |
pvo | mtaylor: ok, that might be easier. | 23:00 |
mtaylor | soren: perfect | 23:00 |
mtaylor | eday: read slashdot today? | 23:00 |
*** dendro-afk is now known as dendrobates | 23:03 | |
pvo | mtaylor: the What if oracle bought all Opensource article? | 23:05 |
mtaylor | pvo: heh | 23:07 |
mtaylor | actually, the Java NIO article : http://developers.slashdot.org/story/10/07/27/1925209/Java-IO-Faster-Than-NIO ... it's java, but also sort of, well... | 23:07 |
mtaylor | I dunno | 23:07 |
*** blpiatt is now known as blp-away | 23:09 | |
soren | pvo: In case you didn't already, can you do a "bzr whoami blahblah" on your dev box? | 23:13 |
pvo | $ bzr whoami | 23:14 |
pvo | Paul Voccio <pvoccio@castor.rackspace.corp> | 23:14 |
pvo | bzr whoami blahblah | 23:14 |
pvo | "blahblah" does not seem to contain an email address. This is allowed, but not recommended. | 23:14 |
pvo | you mean, actually configure it. : ) | 23:14 |
pvo | thats better. | 23:15 |
soren | Smart arse :) | 23:15 |
pvo | hey hey, I wasn't sure at first. I thought you were asking me to validate something, then I realized you meant my config wasn't right | 23:15 |
pvo | we're good now. | 23:16 |
soren | mtaylor: Ok, I've split out debian/, done the import-dsc thing now (with a bit of a twist: there are no releases yet, but the upstream tarball has been imported). Is anything missing? | 23:16 |
mtaylor | soren: nope. does bzr bd make happiness? | 23:17 |
mtaylor | on the packaging branch? | 23:17 |
soren | mtaylor: Lemme check. | 23:17 |
soren | I don't expect to have that branch guarded by Tarmac, by the way. | 23:18 |
mtaylor | soren: well, if you imported a tarball you made from a rev, you should tag that rev as 0.9.0 and then bump the rev - so that when we release again we can merge the new tarball in to the merge | 23:18 |
mtaylor | soren: agree | 23:19 |
*** lbieber has quit IRC | 23:19 | |
soren | mtaylor: Hmm.. | 23:19 |
soren | mtaylor: It failed miserably. | 23:19 |
mtaylor | heh | 23:20 |
soren | *sigh* | 23:20 |
*** _jcsmith has quit IRC | 23:21 | |
soren | mtaylor: How do I extract the tarball? | 23:21 |
mtaylor | soren: as a step by itself? not fully sure I've ever tried | 23:21 |
mtaylor | soren: it exists as a rev in the branch ... with tag upstream-0.9.0 | 23:22 |
*** littleidea has joined #openstack | 23:23 | |
soren | mtaylor: Y'know... Let's just forget that ubuntu-packaging branch ever existed.. | 23:23 |
mtaylor | soren: mmm. yes | 23:24 |
soren | mtaylor: I think I may have used a weird tarball (from my MANIFEST.in experiments) when doing the import. | 23:24 |
mtaylor | soren: hehe | 23:24 |
soren | mtaylor: There. Gone. I'll fix it tomorrow. | 23:25 |
soren | mtaylor: It's 1:30 here, so I think I'm calling it a day. | 23:25 |
soren | g'night, guys. | 23:25 |
mtaylor | soren: night! | 23:26 |
mtaylor | soren: if you want, I can make the deb branch from your remove deb branch | 23:27 |
*** jtdowney has quit IRC | 23:28 | |
*** jsmith-away is now known as jsmith | 23:28 | |
vishvananda | mtaylor: deb branch would be awesome | 23:29 |
mtaylor | vishvananda: k. I'll see if I can get that done | 23:29 |
*** pvo has quit IRC | 23:30 | |
vishvananda | all: it would be great if we got one of the lp609749 patches approved | 23:32 |
vishvananda | right now it dies if you try to start two instances at once | 23:32 |
*** greenisus has quit IRC | 23:33 | |
*** greenisus has joined #openstack | 23:33 | |
*** devcamcar has joined #openstack | 23:36 | |
*** blp-away has quit IRC | 23:39 | |
*** jdmaturen has quit IRC | 23:42 | |
*** jdmaturen has joined #openstack | 23:42 | |
termie | the debate is warming up over in the move_tests branch merge arena | 23:42 |
termie | devcamcar: oh hai | 23:43 |
anm_ | justinsb: fyi, using your installfest instructions and some help from others, it actually IS possible to get nova into a state where it will launch and an instance | 23:43 |
devcamcar | termie: howdy! <3 | 23:43 |
redbo | mtaylor: the swift deb doesn't include the swift python package | 23:44 |
*** devcamcar has quit IRC | 23:44 | |
mtaylor | redbo: hrm. ok. I will fix | 23:45 |
mtaylor | redbo: must for a moment though - so it'll be just a bit | 23:47 |
*** mtaylor has quit IRC | 23:51 | |
*** pvo has joined #openstack | 23:51 | |
*** pvo has joined #openstack | 23:51 | |
*** ChanServ sets mode: +v pvo | 23:51 | |
*** devcamcar has joined #openstack | 23:51 | |
*** devcamcar has left #openstack | 23:52 | |
*** devcamcar has joined #openstack | 23:52 | |
*** greenisus has quit IRC | 23:53 | |
*** devcamcar has quit IRC | 23:53 | |
*** greenisus has joined #openstack | 23:56 | |
*** stewart has quit IRC | 23:57 | |
*** stewart has joined #openstack | 23:58 | |
*** greenisus_ has joined #openstack | 00:00 | |
*** z0 has quit IRC | 00:01 | |
*** zebziggle has joined #openstack | 00:02 | |
*** vishvananda has quit IRC | 00:04 | |
*** greenisus has quit IRC | 00:04 | |
*** greenisus_ is now known as greenisus | 00:04 | |
*** vishvananda has joined #openstack | 00:07 | |
*** vish1 has joined #openstack | 00:08 | |
*** z0 has joined #openstack | 00:10 | |
*** stewart has quit IRC | 00:11 | |
*** sdelano has quit IRC | 00:13 | |
*** imeyer has quit IRC | 00:14 | |
*** imeyer has joined #openstack | 00:15 | |
*** greenisus has quit IRC | 00:18 | |
*** zebziggle has left #openstack | 00:28 | |
*** zebziggle has joined #openstack | 00:28 | |
*** zebziggle has quit IRC | 00:31 | |
*** Guest29615 has joined #openstack | 00:32 | |
*** Guest29615 is now known as zebziggle | 00:33 | |
*** zebziggle has left #openstack | 00:33 | |
*** zebziggle has joined #openstack | 00:33 | |
*** lbieber has joined #openstack | 00:38 | |
*** gnrfan has quit IRC | 00:40 | |
*** zebziggle has left #openstack | 00:42 | |
*** lbieber has quit IRC | 00:42 | |
*** pvo has quit IRC | 00:44 | |
*** pvo has joined #openstack | 00:45 | |
*** ChanServ sets mode: +v pvo | 00:45 | |
*** binaryWarrior has joined #openstack | 00:49 | |
*** z0 has quit IRC | 00:51 | |
*** binaryWarrior has quit IRC | 00:53 | |
*** kallistec has quit IRC | 00:53 | |
*** pvo has quit IRC | 00:56 | |
silassewell | Wondering if anyone could tell me the difference between a graceful shutdown and non-graceful shutdown. Does it just try to let current clients finish their requests or is it something more than that? | 01:02 |
*** jdmaturen has quit IRC | 01:03 | |
_0x44 | silassewell in the RS/OpenStack API a graceful shutdown will attempt to gracefully shutdown the slice and if it fails, fails. | 01:03 |
_0x44 | Errr... sends a shutdown -h now to the slice | 01:04 |
_0x44 | non-graceful attempts a graceful shutdown, but if that takes too long just terminates the VM. | 01:04 |
*** algorist has quit IRC | 01:10 | |
silassewell | _0x44: thanks | 01:10 |
_0x44 | silassewell: No problem | 01:11 |
*** mrayzenoss has joined #openstack | 01:17 | |
*** mrayzenoss is now known as mray | 01:17 | |
*** mray has joined #openstack | 01:17 | |
*** justinsheehy has quit IRC | 01:21 | |
*** adjohn has joined #openstack | 01:25 | |
*** justinsheehy has joined #openstack | 01:26 | |
*** wattersjames has quit IRC | 01:28 | |
*** kallistec has joined #openstack | 01:31 | |
*** pvo has joined #openstack | 01:34 | |
*** ChanServ sets mode: +v pvo | 01:34 | |
vish1 | so i attempted to build using the script, but unfortunately the packages no longer install files into the path | 01:35 |
vish1 | i guess that one line was important | 01:35 |
vish1 | or maybe i just forgot to install nova-common | 01:37 |
vish1 | :) | 01:37 |
vish1 | although you'd think the package would require it | 01:39 |
*** viewlogic has quit IRC | 01:43 | |
*** jhalogen has quit IRC | 01:44 | |
*** mtaylor has joined #openstack | 01:47 | |
*** ChanServ sets mode: +v mtaylor | 01:47 | |
*** miclorb has quit IRC | 01:58 | |
*** miclorb_ has joined #openstack | 01:58 | |
*** davidstrauss has quit IRC | 01:59 | |
*** gustavomzw has left #openstack | 02:01 | |
*** miclorb_ has quit IRC | 02:03 | |
*** wattersjames has joined #openstack | 02:04 | |
vish1 | i need a deb packaging genius if anyone is around | 02:06 |
*** wilsocj1 has joined #openstack | 02:06 | |
*** cjwilson has joined #openstack | 02:11 | |
mtaylor | vish1: whazzup? | 02:12 |
mtaylor | vish1: (sorry, I haven't gotten to the deb branch yet) | 02:12 |
*** stewart has joined #openstack | 02:14 | |
vish1 | mtaylor: so i removed this thing from nova-common.install | 02:21 |
vish1 | which was apparently what was installing the .py files | 02:21 |
vish1 | mtaylor: this line --> usr/lib/python*/*-packages/nova/* | 02:22 |
vish1 | it used to work | 02:22 |
vish1 | now (perhaps because of makefile) that line gives a file not found error | 02:23 |
mtaylor | vish1: so... we really shouldn't be needing to explicitly specifying the location of the .py files ... the rules files should be doing it all through python-support now | 02:24 |
mtaylor | vish1: but I haven't gone through and done a walkthrough of the whole packaging ... | 02:24 |
mtaylor | vish1: lemme take a look right now | 02:25 |
vish1 | ok | 02:26 |
vish1 | i'm using a crummy workaround | 02:26 |
vish1 | nova/ usr/lib/python2.6/dist-packages/ | 02:26 |
*** pvo has quit IRC | 02:27 | |
*** steph021 has joined #openstack | 02:27 | |
mtaylor | heh. yeah - we'll get this all going properly ... | 02:29 |
_0x44 | Going through all the libvirt ./configure dependencies piece-meal is a bit of a pain. :\ | 02:33 |
*** tobym has quit IRC | 02:38 | |
mtaylor | vish1: ok. I see the problem... should have a solution for you in about an hour or so | 02:50 |
mtaylor | biab | 02:50 |
*** mtaylor has quit IRC | 02:56 | |
vish1 | mtaylor: thanks | 02:56 |
*** jhalogen has joined #openstack | 03:00 | |
*** algorist has joined #openstack | 03:02 | |
*** stewart has quit IRC | 03:04 | |
*** tobym has joined #openstack | 03:05 | |
*** mray has quit IRC | 03:06 | |
*** jsmith is now known as jsmith-away | 03:08 | |
chilts | heh, just saw my nick at 7:40 on this video -> http://www.rackspacecloud.com/blog/2010/07/26/oscon-2010-keynote-toward-an-open-cloud/ | 03:08 |
chilts | that's pretty funny :) | 03:08 |
*** jhalogen has quit IRC | 03:09 | |
*** davidstrauss has joined #openstack | 03:11 | |
*** lbieber has joined #openstack | 03:11 | |
*** stewart has joined #openstack | 03:17 | |
*** pvo has joined #openstack | 03:22 | |
*** ChanServ sets mode: +v pvo | 03:22 | |
*** algorist has quit IRC | 03:23 | |
*** algorist has joined #openstack | 03:25 | |
*** miclorb_ has joined #openstack | 03:27 | |
*** wattersjames has quit IRC | 03:27 | |
*** ianw has quit IRC | 03:33 | |
_0x44 | I've updated http://wiki.openstack.org/InstallFromSource with instructions to allow people to build nova on slices. | 03:37 |
*** mtaylor has joined #openstack | 03:54 | |
*** ChanServ sets mode: +v mtaylor | 03:54 | |
*** miclorb_ has quit IRC | 04:13 | |
*** miclorb has joined #openstack | 04:14 | |
eday | mtaylor: what in particular was on /. today? the java io/nio thing? | 04:16 |
mtaylor | eday: yeah | 04:16 |
eday | mtaylor: yup, there were some things about that and SEDA yesterday on twitter | 04:19 |
eday | mtaylor: i tweeted the guy saying it's fine until you hit the 32k thread limit in java... sometimes you need 50k connections :) | 04:20 |
eday | or more, if your app manages lots of idle cons | 04:20 |
mtaylor | heh | 04:20 |
jbarratt | huh, trying to catch up on the chat backlog, and the openstack logs are 403'ing | 04:20 |
jbarratt | http://irclogs.openstack.org/logs/2010-07/%23openstack.27.log | 04:21 |
*** algorist has quit IRC | 04:21 | |
*** adjohn has quit IRC | 04:21 | |
mtaylor | oh for crying out loud | 04:25 |
mtaylor | chromakode: ^^^^^ | 04:25 |
mtaylor | chromakode: suggestions for how I handle that moving forward? | 04:26 |
*** adjohn has joined #openstack | 04:28 | |
*** pvo has quit IRC | 04:30 | |
mtaylor | vish1: ok. I found several of the problems... | 04:42 |
mtaylor | vish1: a few of them required fixes to the based branch - I'm proposing that for merge right now | 04:43 |
mtaylor | vish1: then I've got a debian branch which builds all of the packages again | 04:43 |
mtaylor | vish1: but I need the fixes in before I can push the debian branch - as I need to rebuild the debian branch on top of it | 04:44 |
*** lbieber has quit IRC | 04:54 | |
*** pvo has joined #openstack | 05:01 | |
*** pvo has joined #openstack | 05:01 | |
*** ChanServ sets mode: +v pvo | 05:01 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!