| soren | Or just one? | 00:00 |
|---|---|---|
| *** pcrews has quit IRC | 00:01 | |
| vish1_ | soren: no idea | 00:01 |
| vish1_ | wait though | 00:01 |
| vish1_ | another (related?) issue | 00:01 |
| vish1_ | File "/Users/vishvananda/cloudkick/nova/.nova-venv/lib/python2.6/site-packages/boto/utils.py", line 72, in canonical_string | 00:01 |
| vish1_ | provider = boto.provider.get_default() | 00:01 |
| vish1_ | exceptions.AttributeError: 'module' object has no attribute 'provider' | 00:01 |
| vish1_ | that is boto actually failing internally? | 00:02 |
| soren | Semi-related. | 00:02 |
| soren | Yeah. | 00:02 |
| *** miclorb_ has quit IRC | 00:02 | |
| soren | We can fix it from nova, though, but yeah, it's boto's own fault. | 00:02 |
| vish1_ | so 2.0b1 is a pos | 00:02 |
| vish1_ | how? import boto.provider? | 00:03 |
| soren | Should do the trick. | 00:03 |
| *** rmillner has quit IRC | 00:03 | |
| *** lbieber has quit IRC | 00:03 | |
| soren | but it's 2 AM for me. Take whatever I say with a grain of your favourite mineral. | 00:03 |
| vish1_ | trying the hack | 00:04 |
| vish1_ | ok just importing boto | 00:05 |
| vish1_ | instead of boto.utils works | 00:05 |
| soren | Yeah. | 00:05 |
| soren | That's why they didn't catch it. | 00:05 |
| soren | "import boto" gets everything imported, probably. | 00:06 |
| *** miclorb_ has joined #openstack | 00:11 | |
| *** adjohn has joined #openstack | 00:11 | |
| *** iroth has quit IRC | 00:19 | |
| *** gustavomzw has quit IRC | 00:21 | |
| *** gasbakid has quit IRC | 00:23 | |
| *** miclorb_ has quit IRC | 00:29 | |
| vish1_ | wanted feature: | 00:34 |
| vish1_ | flag for hvm vs qemu in libvirt.xml creation | 00:34 |
| vish1_ | --libvirt_type | 00:39 |
| *** _buck has quit IRC | 00:43 | |
| vish1_ | i went ahead and made it :) | 00:56 |
| *** miclorb has joined #openstack | 01:01 | |
| *** maximilien has quit IRC | 01:02 | |
| *** jdmaturen has quit IRC | 01:05 | |
| vish1_ | if there are any people familiar with brew recipes, we could use one for euca2ools | 01:06 |
| vish1_ | here are the manual steps | 01:06 |
| vish1_ | wget http://open.eucalyptus.com/sites/all/modules/pubdlcnt/pubdlcnt.php?file=http://eucalyptussoftware.com/downloads/releases/euca2ools-1.2.tar.gz&nid=1346 | 01:06 |
| vish1_ | tar zxf euca2ools-1.2.tar.gz | 01:06 |
| vish1_ | cd euca2ools-1.2 | 01:06 |
| vish1_ | sed -i="" "s/-g root/-g wheel/g" Makefile | 01:06 |
| vish1_ | sudo make install | 01:06 |
| *** kallistec has quit IRC | 01:07 | |
| *** tobym has quit IRC | 01:08 | |
| *** vish1_ has quit IRC | 01:25 | |
| *** dendrobates is now known as dendro-afk | 01:27 | |
| *** dendro-afk is now known as dendrobates | 01:27 | |
| *** mrayzenoss has joined #openstack | 01:28 | |
| *** mrayzenoss is now known as mray | 01:28 | |
| *** mray has joined #openstack | 01:28 | |
| *** justinsb_ has quit IRC | 01:31 | |
| *** jdmaturen_ has joined #openstack | 01:36 | |
| *** jdmaturen_ has quit IRC | 01:37 | |
| *** jdmaturen has joined #openstack | 01:38 | |
| *** jdmaturen has quit IRC | 01:40 | |
| *** mray has quit IRC | 01:41 | |
| *** Brainspackle has joined #openstack | 01:48 | |
| *** miclorb has quit IRC | 01:48 | |
| *** iroth has joined #openstack | 01:48 | |
| *** Brainspackle is now known as viewlogic | 01:49 | |
| *** kallistec has joined #openstack | 01:51 | |
| *** lbieber has joined #openstack | 01:51 | |
| *** miclorb_ has joined #openstack | 01:56 | |
| *** silassewell has quit IRC | 01:59 | |
| *** miclorb_ has quit IRC | 02:01 | |
| *** miclorb has joined #openstack | 02:07 | |
| *** littleidea has quit IRC | 02:09 | |
| *** lbieber has quit IRC | 02:10 | |
| *** blpiatt has joined #openstack | 02:12 | |
| *** littleidea has joined #openstack | 02:13 | |
| *** silassewell has joined #openstack | 02:13 | |
| *** littleidea has quit IRC | 02:16 | |
| *** littleidea_ has joined #openstack | 02:16 | |
| *** littleidea_ has quit IRC | 02:19 | |
| *** miclorb has quit IRC | 02:22 | |
| *** miclorb has joined #openstack | 02:23 | |
| *** miclorb has quit IRC | 02:23 | |
| *** mrayzenoss has joined #openstack | 02:28 | |
| *** mrayzenoss is now known as mray | 02:28 | |
| *** lbieber has joined #openstack | 02:35 | |
| *** sophiap has joined #openstack | 02:39 | |
| *** littleidea has joined #openstack | 02:48 | |
| *** littleidea has quit IRC | 02:54 | |
| *** _jcsmith has joined #openstack | 02:54 | |
| _jcsmith | anybody know a graphical mac/windows cloud files file transfer/browser app that works with a non-rackspace auth server? the command line stuff works well, just want to do a demo to non techies | 02:57 |
| _jcsmith | for swift | 02:57 |
| *** JimCurry has joined #openstack | 02:59 | |
| *** wattersjames_ has joined #openstack | 03:01 | |
| *** littleidea has joined #openstack | 03:02 | |
| *** wattersjames has quit IRC | 03:03 | |
| *** miclorb_ has joined #openstack | 03:04 | |
| creiht | _jcsmith: unfortunately none that I am aware of :/ | 03:10 |
| _jcsmith | ok :) | 03:11 |
| creiht | I'm looking around, but can't seem to find anything | 03:11 |
| _jcsmith | I looked too, if people felt there was a need for a windows one I could write something in C#, but that means I'd have to admit to being a former windows developer ;) | 03:14 |
| creiht | There is supposedly an iphone app that does, but not sure if that will work for you | 03:14 |
| creiht | haha | 03:14 |
| creiht | It wouldn't be hard to whip something up with pyQT | 03:14 |
| creiht | or something like that | 03:14 |
| _jcsmith | yeah, I'd prefer the python wrote | 03:15 |
| _jcsmith | I went to the nova meetup in SF today, there was a good showing of people, I look forward to the swift one | 03:16 |
| creiht | swift one? | 03:16 |
| creiht | Hadn't heard about it yet | 03:16 |
| _jcsmith | meetup.com | 03:16 |
| _jcsmith | openstack group | 03:16 |
| _jcsmith | at nasa ames | 03:16 |
| creiht | ahh cool | 03:17 |
| _jcsmith | its at the end of august | 03:17 |
| creiht | well hopefully they will send out some swift devs :) | 03:17 |
| _jcsmith | that would be awesome, you deserve it, you've answered a lot of my newbie questions | 03:18 |
| creiht | hah | 03:18 |
| *** tobym has joined #openstack | 03:19 | |
| *** StylusEater has left #openstack | 03:22 | |
| _jcsmith | creiht: are you familiar with python-cloudfiles? | 03:23 |
| creiht | _jcsmith: a little | 03:23 |
| _jcsmith | off the top of your head do you know how to specify a custom auth url? | 03:24 |
| cw | _jcsmith: fake out DNS ? or are certs involved? | 03:24 |
| blpiatt | _jcsmith have you checked out the openstack web control panel for interfacing with object storage? you can ping @greenisus about it on twitter or here, or email mike.mayo@rackspace.com | 03:24 |
| cw | _jcsmith: please tell me you're not still working in the city? :-) | 03:25 |
| _jcsmith | cw: no, back in the east bay. I fixed the auth issue | 03:25 |
| cw | oh sweet, let me fire up IM | 03:25 |
| _jcsmith | want to use something other than 'st' to talk to the cluster | 03:25 |
| _jcsmith | blpiatt: no, but I'll check it out! | 03:26 |
| creiht | _jcsmith: I think you send authurl when creating the connection | 03:26 |
| cw | _jcsmith: -> | 03:26 |
| creiht | _jcsmith: you might also checkout swift/common/client.py for a simpler bear metal api | 03:27 |
| blpiatt | _jcsmith he lives in the bay area too, he was at the meetup tonight so you could have met him today and didn't know it ;) | 03:27 |
| _jcsmith | is client.py the same as 'st'? I got that working | 03:27 |
| creiht | st uses client.py | 03:27 |
| cw | blpiatt: we were in the back left corner | 03:27 |
| *** jsgotangco has joined #openstack | 03:28 | |
| *** littleidea has quit IRC | 03:28 | |
| blpiatt | cw, I'm in san antonio so I didn't make it :( I just talked to him and said "make sure everyone has pizza tonight!" | 03:28 |
| _jcsmith | oh yeah, thanks for the pizza Rackspace :) | 03:29 |
| jsgotangco | mm pizza | 03:30 |
| creiht | blpiatt: Do you know if they are planning on sending out any swift devs to the swift meetup? | 03:31 |
| blpiatt | I'm not sure how we're handling getting proper coverage to all of the stuff yet, we need to figure out how to ramp up people all over as meetups are going to be everywhere soon and we won't be able to cover them all | 03:32 |
| creiht | ahh | 03:32 |
| blpiatt | Awesome that the first was in the bay so the team from Ames could make it up as well as Mike heading over | 03:32 |
| cw | lee was there too | 03:32 |
| _jcsmith | any excuse to go to texas works for me :) | 03:33 |
| cw | get in line | 03:33 |
| blpiatt | Yeah, Lee lives in Palo Alto as well, we have a few folks in that area too :) | 03:33 |
| blpiatt | Xen.org is going to have their next summit in TX so come on out for that | 03:33 |
| *** mray has quit IRC | 03:33 | |
| *** pvo has joined #openstack | 03:38 | |
| *** ChanServ sets mode: +v pvo | 03:38 | |
| blpiatt | pvo, good evening | 03:39 |
| pvo | why hello there. | 03:39 |
| _jcsmith | pvo: thanks for hosting us at your digs tonight | 03:40 |
| pvo | _jcsmith: I have digs? I'm in San Antonio... | 03:40 |
| _jcsmith | oops | 03:40 |
| pvo | you guys were at cloudkick? | 03:41 |
| _jcsmith | a case of mistaken identity | 03:41 |
| _jcsmith | yeah | 03:41 |
| blpiatt | I love google translate but it isn't quite perfect yet, checking out Bayoda's website (openstack corporate CLA signed), they have a backup product, translated one portion as "Provides furious fast data loss", somehow that can't really be right :D | 03:41 |
| pvo | good times? lots 'o code? | 03:41 |
| _jcsmith | 'o code, heh | 03:41 |
| blpiatt | cloudkick = polvi, cloud servers dev lead for rackspace = pvo | 03:42 |
| _jcsmith | that explains it :) | 03:42 |
| *** chilts has left #openstack | 03:44 | |
| pvo | sweet. tests passing today. | 03:45 |
| blpiatt | pvo, did you bribe hudson or did the team push some new code? | 03:45 |
| pvo | ~ 30 revs since I last checked out. | 03:47 |
| *** mrayzenoss has joined #openstack | 03:48 | |
| *** mrayzenoss is now known as mray | 03:48 | |
| *** mray has joined #openstack | 03:48 | |
| blpiatt | awesome, love the pace at which stuff is getting revved | 03:48 |
| _jcsmith | btw, to answer my own question about python-cloudfiles, this is how you can specify your the url to your own auth server | 03:49 |
| _jcsmith | conn = cloudfiles.get_connection(username='accountXXX:userXXX', api_key='passwdXXX', authurl='http://api.mytestbox.com:11000/v1.0') | 03:49 |
| creiht | yup | 03:49 |
| *** mray has quit IRC | 03:49 | |
| _jcsmith | thanks creiht | 03:50 |
| creiht | and I just looked at the control panel code, and correct me if I'm wrong, but I don't think it has storage support in it yet | 03:51 |
| blpiatt | creiht, hmm that could be true! sounds like it needs help | 03:52 |
| _jcsmith | is the control panel a django app? | 03:52 |
| creiht | worse, it is ruby :) | 03:53 |
| *** littleidea has joined #openstack | 03:53 | |
| _jcsmith | oh man, no comment :) | 03:53 |
| creiht | and a lot of JS | 03:53 |
| creiht | though to be fair, the ruby is a bit light, | 03:53 |
| blpiatt | it is a ruby backend with a cappuccino front end, written by an iphone dev ;) | 03:53 |
| _jcsmith | neat | 03:53 |
| blpiatt | it is a prototype, we need a functional control panel, not sure what the best thing to write it in is long term, we'll have to talk more about that at the dev summit in november, maybe hack on this until then | 03:54 |
| blpiatt | the ruby backend is a jsonp proxy so you could connect multiple frontends to it, mike was messing with some ideas when he put it together | 03:55 |
| jsgotangco | blpiatt: hey sorry to budge in but you got to talk to yoram earlier yes? | 03:55 |
| _jcsmith | I agree having a control panel is a must to make this as turn key as possible for hosting companies | 03:55 |
| blpiatt | jsgotangco, yes I did, great chat | 03:55 |
| jsgotangco | blpiatt: great thanks looking forward to collaborating | 03:57 |
| blpiatt | we're really excited to help give a big community a place to collaborate on building the best software together | 03:57 |
| *** metoikos has quit IRC | 04:04 | |
| *** jernst has joined #openstack | 04:05 | |
| *** jernst has left #openstack | 04:06 | |
| *** miclorb_ has quit IRC | 04:06 | |
| *** JimCurry has left #openstack | 04:08 | |
| *** miclorb_ has joined #openstack | 04:13 | |
| *** cjwilson has joined #openstack | 04:15 | |
| *** silassewell has quit IRC | 04:16 | |
| *** tobym has quit IRC | 04:19 | |
| *** adjohn has quit IRC | 04:22 | |
| *** tobym has joined #openstack | 04:22 | |
| *** miclorb_ has quit IRC | 04:26 | |
| *** miclorb_ has joined #openstack | 04:26 | |
| *** adjohn has joined #openstack | 04:34 | |
| *** jsmith is now known as jsmith-away | 04:42 | |
| *** blpiatt has quit IRC | 04:45 | |
| *** pvo has quit IRC | 04:55 | |
| *** maximilien has joined #openstack | 05:01 | |
| *** alexsolo has joined #openstack | 05:03 | |
| *** miclorb_ has joined #openstack | 05:09 | |
| *** jbarratt has quit IRC | 05:09 | |
| *** alexsolo has quit IRC | 05:10 | |
| *** abecc has quit IRC | 05:15 | |
| *** lbieber has quit IRC | 05:19 | |
| *** maximilien has quit IRC | 05:21 | |
| *** miclorb_ has quit IRC | 05:22 | |
| *** miclorb has joined #openstack | 05:31 | |
| *** Cloudkylemac has joined #openstack | 05:32 | |
| *** Cloudkylemac has left #openstack | 05:33 | |
| *** kylemac has joined #openstack | 05:34 | |
| *** kylemac is now known as topspinkyle | 05:35 | |
| *** topspinkyle has joined #openstack | 05:35 | |
| *** miclorb has quit IRC | 05:51 | |
| *** miclorb_ has joined #openstack | 05:52 | |
| *** wattersjames has joined #openstack | 05:53 | |
| *** miclorb_ has joined #openstack | 05:53 | |
| *** miclorb_ has quit IRC | 06:25 | |
| *** miclorb has joined #openstack | 06:25 | |
| *** _jcsmith has quit IRC | 06:26 | |
| *** rajijoom has joined #openstack | 06:28 | |
| *** wattersjames has quit IRC | 06:43 | |
| *** fortune has joined #openstack | 06:59 | |
| *** joshuamckenty has joined #openstack | 07:01 | |
| *** _buck has joined #openstack | 07:07 | |
| *** calavera has joined #openstack | 07:19 | |
| *** CnC has quit IRC | 07:48 | |
| *** _buck has quit IRC | 07:59 | |
| *** iroth has quit IRC | 08:18 | |
| *** adjohn has left #openstack | 08:19 | |
| *** littleidea has quit IRC | 08:21 | |
| *** allsystemsarego has joined #openstack | 08:21 | |
| termie | i think mailing lists are broken for me | 08:25 |
| termie | i am not getting anything form the nova mailing list | 08:25 |
| *** gasbakid has joined #openstack | 08:26 | |
| termie | yet my launchpad ui claims i am | 08:26 |
| termie | and my messages haven't gone through or been given a rejection notice | 08:28 |
| termie | and i thought i was just sitting here waiting for replies | 08:28 |
| termie | when nobody has even seen my messages | 08:28 |
| *** joshuamckenty has quit IRC | 08:30 | |
| termie | what is the little diamond shaped symbol that is next to project names on launchpad? | 08:32 |
| termie | it says "bazaar branches" when i hover | 08:33 |
| termie | but that doesn't seem to mean much considering we are on launchpad | 08:33 |
| soren | termie: Where, exactly? | 08:44 |
| termie | if i go to launchpad.net/~termie | 08:45 |
| termie | under Most Active In | 08:45 |
| termie | the three projects i've participated in in launchpad all have this little symbol | 08:45 |
| soren | Right. Lots of project that don't /live/ on Launchpad, are /registered/ on Launchpad, for instance. | 08:45 |
| termie | this is OpenStack Compute(nova), gitbzr-ng and Pantheon | 08:46 |
| soren | Others just host bugs on Launchpad, but vcs elsewhere, or vice versa. | 08:46 |
| termie | so two of those are all in launchpad | 08:46 |
| soren | Those symbols denote which parts of the project live on Launchpad. | 08:46 |
| termie | so everything, effectively, wil have the bazaar symbole then? | 08:47 |
| soren | No. | 08:47 |
| soren | 08:46 < soren> Others just host bugs on Launchpad, but vcs elsewhere, or vice versa. | 08:47 |
| termie | right, i said effectively | 08:47 |
| termie | any project that is being worked on on launchpad | 08:47 |
| soren | Far, far from it. | 08:47 |
| soren | Lots of project that don't /live/ on Launchpad, are /registered/ on Launchpad. | 08:47 |
| termie | okay fine, plenty of other stuff won't have it | 08:47 |
| termie | the point is it is very common | 08:48 |
| termie | and it looks very much like the also difficult to read edit symbol | 08:48 |
| soren | Really. I'm not being pedantic. Thousands of projects are registered on Launchpad without their actually using it. | 08:48 |
| termie | and only provides minimal information once you are already familiar with a project | 08:48 |
| termie | can we ditch the yellow and just make it a gray symbol that doesn't catch the eye? | 08:49 |
| termie | my launchpad overview looks like it is calling out for attention from every link | 08:49 |
| soren | I'm not arguing that the information is superuseful or that it belongs on one's main page. | 08:49 |
| termie | i'm not claiming you are, i think we were off topic | 08:50 |
| soren | Quite so :) | 08:50 |
| termie | the point of all this being that it represents something common and has a visual conflict with another symbol | 08:51 |
| termie | perhaps a chrome extension is hte way to go here | 08:52 |
| termie | .sprite { display: none; } | 08:54 |
| termie | soren: on a different note, any idea why i'm not receiving mail from the mailing list? on my +editemails page it has my email address under 'subscribe with' for both nova and openstack discuss | 08:59 |
| *** jtdowney has joined #openstack | 09:00 | |
| termie | i've tried cycling it | 09:00 |
| termie | so maybe i'll get new messages? | 09:00 |
| *** adjohn has joined #openstack | 09:08 | |
| *** londo has quit IRC | 09:08 | |
| *** londo has joined #openstack | 09:11 | |
| soren | termie: It landed in the moderation queue for some reason. I'm trying to work out why. | 09:11 |
| termie | soren: i'm a member of nova-core rather than "nova" perhaps? | 09:12 |
| termie | although it says inderirect membership and put the list on my mailing lists section | 09:12 |
| soren | termie: Yeah, it should be fne. | 09:12 |
| soren | fine, even. | 09:12 |
| soren | termie: Could you send something else to the list, please? Then I can approve your real e-mail and use the new one for debugging with the launchpad guys. | 09:18 |
| *** adjohn_ has joined #openstack | 09:21 | |
| *** adjohn has quit IRC | 09:21 | |
| *** adjohn_ is now known as adjohn | 09:21 | |
| termie | soren: sent | 09:21 |
| soren | ta | 09:22 |
| soren | termie: Uh... I think maybe someone went ahead and did the moderation thing for me. | 09:22 |
| soren | termie: And your second one shot straight through. Did you also receive them back? | 09:24 |
| termie | i hven't received anything at all, no :/ | 09:25 |
| termie | although somebody claims to have gotten my last email | 09:25 |
| termie | or rather, my email about gflags | 09:25 |
| soren | Yup, I got both that one and your test e-mail. | 09:26 |
| termie | want to reply to the test email so i can see if i get anything? | 09:26 |
| soren | Han gon. | 09:27 |
| soren | Hang on, even. | 09:27 |
| termie | ah man, i was really hoping i was just learning a new cool phrase | 09:27 |
| *** metoikos has joined #openstack | 09:28 | |
| soren | termie: Is it flowing both ways now? | 09:36 |
| soren | mtaylor: https://edge.launchpad.net/~soren/+archive/ppa/+build/1893305 | 09:37 |
| soren | mtaylor: Apparantly our unit tests require a running redis-server. | 09:38 |
| *** jsgotangco has quit IRC | 09:44 | |
| termie | aye, something i'd love to change but -shrug- | 09:45 |
| termie | soren: nope, still haven't received any mail | 09:49 |
| termie | nor is it in spam or anything like that | 09:50 |
| termie | i do get bug emails | 09:50 |
| termie | and merge emails | 09:50 |
| termie | ... also my "Most active in" seems to have disappeared in the last 5 minutes | 10:00 |
| termie | which i only mention as possibly useful debug information | 10:00 |
| *** gasbakid has quit IRC | 10:05 | |
| *** jtdowney has quit IRC | 10:21 | |
| *** gasbakid has joined #openstack | 10:24 | |
| termie | soren: you expect i'll have to wait until u.s. morning to see mailing list email? | 10:25 |
| *** cjwilson has left #openstack | 10:31 | |
| *** stewart has quit IRC | 10:33 | |
| termie | i'll just reply to messages without threading for now | 10:34 |
| *** stewart has joined #openstack | 10:34 | |
| *** stewart has quit IRC | 10:41 | |
| *** stewart has joined #openstack | 10:44 | |
| *** avsm has joined #openstack | 10:48 | |
| soren | termie: The one guy I know does mailing list stuff on LP (Barry Warzaw, of mailman fame) is on US time, at least. | 10:53 |
| soren | termie: There might be others, I'm not sure. | 10:53 |
| soren | termie: IIRC, he's an early riser, though, so it may not be too long. | 10:54 |
| termie | oh man, i used to work on mailman | 10:55 |
| termie | here's a pretty fancy piece of it that is still there http://bazaar.launchpad.net/~mailman-coders/mailman/2.1/annotate/head:/Mailman/MailList.py#L98 | 10:56 |
| termie | at least when i last was using it (which was something like 6 years ago) trying to subclass that class would cause a coredump | 10:57 |
| termie | nearly the only time i've ever managed to coredump python | 10:58 |
| termie | i think i managed another time when writing an extension | 10:58 |
| *** zul has quit IRC | 10:59 | |
| *** jtdowney has joined #openstack | 11:00 | |
| *** Slidey has left #openstack | 11:02 | |
| *** jsmith-away is now known as jsmith | 11:03 | |
| *** miclorb has joined #openstack | 11:04 | |
| termie | http://term.ie/data/cleanpad_before.png http://term.ie/data/cleanpad_after.png | 11:07 |
| *** jtdowney has quit IRC | 11:11 | |
| *** miclorb has quit IRC | 11:15 | |
| *** jonesy_ has quit IRC | 11:16 | |
| *** miclorb has joined #openstack | 11:19 | |
| *** miclorb has quit IRC | 11:24 | |
| *** tobym has quit IRC | 11:26 | |
| *** miclorb has joined #openstack | 11:32 | |
| *** miclorb has quit IRC | 11:43 | |
| *** justinsheehy has quit IRC | 11:45 | |
| *** carolinad28 has joined #openstack | 11:48 | |
| *** miclorb has joined #openstack | 11:50 | |
| *** jsmith is now known as jsmith-away | 12:20 | |
| *** miclorb has quit IRC | 12:32 | |
| *** calavera has quit IRC | 12:32 | |
| *** justinsheehy has joined #openstack | 12:44 | |
| *** fortune is now known as communitystring | 12:47 | |
| *** communitystring is now known as CommunityString | 12:47 | |
| *** maximilien has joined #openstack | 12:51 | |
| *** CommunityString has left #openstack | 12:52 | |
| *** joshuamckenty has joined #openstack | 13:23 | |
| *** jonesy_ has joined #openstack | 13:26 | |
| *** jsmith-away is now known as jsmith | 13:26 | |
| *** gundlach has quit IRC | 13:34 | |
| *** abecc has joined #openstack | 13:37 | |
| *** gasbakid has quit IRC | 13:42 | |
| *** pvo has joined #openstack | 13:43 | |
| *** pvo has joined #openstack | 13:43 | |
| *** ChanServ sets mode: +v pvo | 13:43 | |
| *** gasbakid has joined #openstack | 13:45 | |
| *** joshuamckenty has quit IRC | 13:45 | |
| soren | Every example of eventlet code I've come across either deals with raw sockets or speaks to a server through eventlet.urllib2 or some variation thereof. What if I want to speak another protocol to something? Does it just happen to work because the socket module has been replaced? | 13:51 |
| creiht | soren: Yes | 13:51 |
| creiht | In most cases, you can use a python library that implements that protocol | 13:52 |
| creiht | unless it does something out of the ordinary (like uses a c extension to handle socket communication) | 13:52 |
| soren | ..or uses some amount of global state or whatnot, I suppose. | 13:53 |
| creiht | I wouldn't think global state would matter | 13:53 |
| *** Podilarius has joined #openstack | 13:53 | |
| soren | creiht: No? | 13:53 |
| creiht | Unless that global state is used for that one connection I guess | 13:54 |
| soren | creiht: If its interactions with socket suddenly arent' blocking.. | 13:54 |
| soren | right, exactly. | 13:54 |
| soren | Ok. | 13:54 |
| creiht | soren: do you have a certain protocol in mind? | 13:54 |
| creiht | Or just wanted to see examples? | 13:55 |
| soren | creiht: Not really. | 13:55 |
| soren | (To your first question) | 13:56 |
| soren | What about file operations or db access? | 13:57 |
| creiht | for file operations we do reads/writes in chunks and make sure to let eventlet get control between operations with eventlet.sleep() | 13:57 |
| creiht | that seems to work pretty well | 13:58 |
| creiht | db access can be a little more problematic, as db libs usually do weird things | 13:58 |
| creiht | The easiest thing is to just access them from a thread pool (like you would in twisted with deferr to thread) | 13:59 |
| creiht | There has been work on both postgres and mysql async support, but I've never messed with either | 13:59 |
| *** adjohn has quit IRC | 13:59 | |
| soren | creiht: In Twisted, I use adbapi, but that may just be deferToThread behind the scenes. I'm not sure. | 14:00 |
| creiht | soren: yeah looks like it uses thread | 14:02 |
| *** jbryce has quit IRC | 14:03 | |
| creiht | soren: eventlet has something similar with db_pool | 14:03 |
| *** lbieber has joined #openstack | 14:03 | |
| creiht | I think that is where the MySQL and pyscopg support comes in, but have never used it | 14:03 |
| *** carolinad28 has quit IRC | 14:03 | |
| *** jbryce has joined #openstack | 14:03 | |
| *** jsgotangco has joined #openstack | 14:08 | |
| soren | I'll probably miss twisted.web most of all. WSGI feels really amputated next to it. | 14:08 |
| *** jsmith is now known as jsmith-away | 14:08 | |
| creiht | wsgi is quite freeing once you learn it | 14:09 |
| *** _buck has joined #openstack | 14:09 | |
| creiht | By using wsgi, it opens up tons of other libraries available for use | 14:09 |
| creiht | soren: which parts of twisted.web do you think you will miss? | 14:10 |
| *** lbieber has quit IRC | 14:10 | |
| *** binaryWarrior has joined #openstack | 14:11 | |
| soren | Resource and the render_* methods. | 14:11 |
| soren | It just saves a lot of copy/paste stuff every time I need to do something with http. | 14:12 |
| soren | Sure, it's reasonable easy to write a dispatcher that does mostly the same thing with WSGI, but it's easier to use one that someone else wrote for me :) | 14:13 |
| creiht | soren: but that is exactly, you can use any of the wsgi dispatchers that are already written | 14:13 |
| *** rajijoom has quit IRC | 14:13 | |
| creiht | You don't have to write that | 14:13 |
| creiht | and you aren't tied to a specific one | 14:13 |
| creiht | We didn't use one in swift because we a.) were trying to minimize dependencies and b.) had a very simple request structure anyways | 14:15 |
| PiotrSikora | hmm... is there any way to use swift on FS without xattr right now? | 14:17 |
| creiht | PiotrSikora: no | 14:17 |
| creiht | which FS? | 14:17 |
| PiotrSikora | UFS ;) | 14:17 |
| PiotrSikora | make it FFS | 14:18 |
| *** binaryWarrior has quit IRC | 14:18 | |
| creiht | soren: You can even use eventlet with most of the wsgi frameworks available (like pylons, restish, werkzeug, etc.) | 14:18 |
| creiht | heh | 14:18 |
| *** binaryWarrior has joined #openstack | 14:18 | |
| PiotrSikora | creiht: are there any plans to? like back-falling to storing meta-data in sqlite? | 14:19 |
| creiht | PiotrSikora: which os? | 14:19 |
| PiotrSikora | OpenBSD ;) | 14:19 |
| *** binaryWarrior has quit IRC | 14:19 | |
| *** binaryWarrior has joined #openstack | 14:20 | |
| creiht | PiotrSikora: http://en.wikipedia.org/wiki/Extended_file_attributes | 14:20 |
| creiht | says that ffs supports xattrs | 14:20 |
| creiht | though it calls them flags | 14:20 |
| creiht | One of the main reasons that we use xattrs is it makes it fairly easy to make sure object operations are atomic | 14:22 |
| PiotrSikora | creiht: wikipedia lies | 14:22 |
| creiht | It's on the internet, it must be true! :) | 14:22 |
| PiotrSikora | and OpenBSD flags are predefined and used for something else (file immutability, append-only, etc) | 14:22 |
| creiht | But it becomes more difficult to make things atomic which you are managing a file and some entries in a DB | 14:23 |
| creiht | hrm | 14:23 |
| PiotrSikora | well, swift is "eventaully consistent", so this shouldn't be deal-breaker ;) | 14:24 |
| creiht | It isn't so much about that | 14:24 |
| *** mrayzenoss has joined #openstack | 14:26 | |
| *** mrayzenoss is now known as mray | 14:26 | |
| *** mray has joined #openstack | 14:26 | |
| gholt | Consistent and atomic are definitely different. | 14:26 |
| redbo | I don't know if I'd focus on the atomicity, it's just a good place to put them | 14:26 |
| creiht | hah | 14:26 |
| notmyname | wow, that got a response from the swift devs | 14:27 |
| gholt | Heheh | 14:27 |
| creiht | I guess everyone is awake now :) | 14:27 |
| creiht | PiotrSikora: We tried several approaches, and xattrs were the best solution that we found | 14:28 |
| *** Podilarius has left #openstack | 14:28 | |
| gholt | We did have a sqlite based object server. It was slow, but it had everything in one db, so... | 14:28 |
| gholt | Does BSD have resource forks? :P | 14:28 |
| redbo | we also tried putting xattrs at the end of the file, but it made the hd seek to the end then back to the front | 14:28 |
| creiht | Of course if anyone comes up with a better way, patches are welcome | 14:28 |
| PiotrSikora | gholt: i know it isn't the same | 14:28 |
| PiotrSikora | gholt: but it seems that when something is EV and resources are versioned, then atomicity shouldn't be required | 14:29 |
| PiotrSikora | but i might be wrong ;) | 14:29 |
| redbo | it's not about atomicity | 14:29 |
| gholt | Oh it's not, I was teasing creiht. :) | 14:29 |
| creiht | heh | 14:29 |
| gholt | We can do all sorts of things to the uploaded data before moving it into place, which is atomic. | 14:29 |
| creiht | true | 14:30 |
| redbo | we would have had to move the metadata to the beginning of the file, which requires copying the file over again. If you think not using nginx is inefficient, that's murder. | 14:30 |
| PiotrSikora | redbo: hehe, i agree | 14:30 |
| PiotrSikora | and merging meta-data with actual data seems like a no-no to me | 14:31 |
| PiotrSikora | but i was thinking about making things more portable | 14:31 |
| redbo | the only other option is a second .meta file, which is as bad as having it at the end of the file for performance | 14:31 |
| PiotrSikora | (which means back-falling to less effective way for non-xattr systems) | 14:32 |
| PiotrSikora | redbo: yes, but its better than not being able to use swift at all | 14:32 |
| notmyname | PiotrSikora: I disagree | 14:32 |
| redbo | yeah maybe | 14:32 |
| *** binaryWarrior has quit IRC | 14:33 | |
| gholt | Well, for me at least, I'd prefer not to provide subpar systems. I hope you don't /have to/ use a particular os. | 14:33 |
| *** Podilarius has joined #openstack | 14:33 | |
| creiht | Wow... OpenBSD doesn't support Journaling filesystems period | 14:33 |
| gholt | What would really rock (probably) is if somebody smart figured out how to do all this to a raw partition and threw out all that file system stuff we don't need. :D | 14:34 |
| PiotrSikora | notmyname: do you mind elaborating a bit on that? ;) | 14:34 |
| notmyname | ya, sorry. got asked a question | 14:34 |
| redbo | gholt: quit that | 14:34 |
| notmyname | at some point, portability becomes an excuse to make a system that doesn't work very well. to me, the answer is "don't do that" | 14:34 |
| notmyname | in other words, as gholt said, I think a subpar system is worse than no system at all | 14:35 |
| redbo | I'd be fine with abstrcting the xattr stuff so it could be stored in a different file. It's not *that* much of a performance difference. | 14:35 |
| redbo | well, I wouldn't be fine with doing it myself, but I'd be fine if that existed. | 14:35 |
| PiotrSikora | gholt: actually, i don't... but i prefer using OpenBSD for as my development box :) | 14:35 |
| PiotrSikora | notmyname: but does it really render it subpar? | 14:36 |
| PiotrSikora | we are not talking about disabling main features | 14:36 |
| PiotrSikora | or slashing performance in half | 14:36 |
| gholt | Well, the single db definitely did that. Hehe | 14:36 |
| notmyname | well, for this particular issue, I don't know. I'd want to see comparison numbers. I mean it more as a general guideline. | 14:37 |
| creiht | hah | 14:37 |
| *** topspinkyle has quit IRC | 14:37 | |
| notmyname | good stuff >> no stuff > bad stuff | 14:37 |
| gholt | But yeah, metafiles would probably be okay for a hobby system [teasing] | 14:37 |
| * creiht waits for the person who wants to run swift on Windows | 14:38 | |
| *** sophiap has quit IRC | 14:38 | |
| PiotrSikora | notmyname: good stuff >> so-so stuff > no stuff > bad stuff | 14:38 |
| PiotrSikora | :P | 14:38 |
| notmyname | good stuff >> no stuff > bad stuff >>> swift on windows | 14:38 |
| *** jsgotangco has quit IRC | 14:38 | |
| gholt | Stop it! | 14:38 |
| gholt | Oh wait. | 14:38 |
| redbo | I don't see why it wouldn't run okay | 14:38 |
| gholt | No continue, this is fun. hehe | 14:38 |
| PiotrSikora | :) | 14:39 |
| creiht | PiotrSikora: All of that said, if someone were to produce a patch that added something like a .meta for each uploaded file that contained the metadata that a.) didn't hurt performance too much, and b.) didn't add too much complexity to the code, we would consider it | 14:40 |
| gholt | That's a lot of words for: patches welcome! | 14:41 |
| redbo | I don't care so much about performance. Let people who use the feature worry about that. But yeah, it'd have to be clean :) | 14:41 |
| gholt | [we use that feature!] | 14:41 |
| *** tr3buchet has joined #openstack | 14:42 | |
| * gholt thinks it's strange that we're all so froggy on a Friday morning. | 14:43 | |
| creiht | heh | 14:43 |
| cory_ | lack of sleep | 14:43 |
| PiotrSikora | creiht: yeah, i'm already looking into it ;) | 14:43 |
| creiht | And to be honest, we already have some of that, since we temporarily use .meta files to handle updates to metadata | 14:44 |
| PiotrSikora | :D | 14:44 |
| redbo | permanently temporarily | 14:44 |
| creiht | man... it has been a long time since I have looked at the object server :) | 14:45 |
| gholt | The trick is, since we did it in the past, making sure only one version of the object meta+data is served at any time. Not a difficult trick, but something to keep in mind. | 14:45 |
| redbo | this is going to be a mess, now that I think about it. Good luck. | 14:47 |
| PiotrSikora | well, since files are versioned (and immutable, i assume?) then adding .meta to each version should handle that? | 14:47 |
| notmyname | PiotrSikora: so if I'm hearing you right, you want one big database, served by nginx modules, and doesn't require xattrs? ;-) | 14:49 |
| creiht | oh heh... the .meta file we store for updates is a 0 length file with the metadata in xattrs :) | 14:49 |
| *** gundlach has joined #openstack | 14:49 | |
| PiotrSikora | notmyname: i never said anything about single database | 14:49 |
| notmyname | something about being able to query the system, though, right? ;-) | 14:50 |
| PiotrSikora | notmyname: i was always talking about sharded RDBMS (drizzle/postgresql) | 14:50 |
| creiht | hehe | 14:51 |
| notmyname | ah ok. I'm just giving you a hard time | 14:51 |
| redbo | creiht: I was hacking on the object server this morning! playing with python-sendfile. | 14:51 |
| creiht | oh noes! :) | 14:51 |
| PiotrSikora | notmyname: which IMHO isn't much different from what you guys are doing... just using sqlite instead, its still sharded data in SQL | 14:52 |
| PiotrSikora | erm... data in sharded SQL* | 14:52 |
| notmyname | s/data/metadata | 14:52 |
| PiotrSikora | right :) | 14:52 |
| PiotrSikora | creiht: guess i can't re-use that .meta file then :) | 14:52 |
| creiht | PiotrSikora: not sure | 14:53 |
| redbo | it's not central to our design that our databases aren't shared. It just doesn't add anything, and sqlite is pretty nice that we can just rsync the files around and not worry about configuration or anything. | 14:53 |
| *** avsm has quit IRC | 14:54 | |
| *** antonym is now known as spy | 14:54 | |
| creiht | PiotrSikora: It is more like you would have to work around the .meta file :) | 14:55 |
| *** spy is now known as Guest53283 | 14:55 | |
| redbo | apparently freebsd supports xattrs | 14:57 |
| creiht | PiotrSikora: And if you got the object server updated, you would also have to make replication work with it | 14:57 |
| creiht | redbo: it does :) | 14:57 |
| PiotrSikora | redbo: yeah, i got that last time :) | 14:57 |
| PiotrSikora | creiht: yeah... i know, i know :) | 14:57 |
| creiht | PiotrSikora: The more I look at it, the more messy it seems | 14:57 |
| redbo | "OpenBSD 3.7 supported extended attributes, although support was never built into the default GENERIC kernel." | 14:57 |
| notmyname | what about using ZFS on FreeBSD? | 14:58 |
| termie | soren: i second the "debugging things that make blocking io look like non-blocking io" statement, sends ripples of anticipatory fear through me | 14:58 |
| redbo | oh they completely removed extended attributes | 14:58 |
| PiotrSikora | redbo: yeah, no xattr in there ;) | 14:58 |
| PiotrSikora | notmyname: ZFS has xattr | 14:58 |
| redbo | zfs won't matter if there's no system call to use them :) | 14:58 |
| *** littleidea has joined #openstack | 14:58 | |
| redbo | oh, does it have a userland library or something? | 14:58 |
| PiotrSikora | erm? | 14:59 |
| termie | soren: but i have to admit after 4 years of twisted i'd be pretty excited to try something else, that's initially one of the reasons we put so much work into tornado before deciding it wasn't worth it | 14:59 |
| PiotrSikora | redbo: FreeBSD has xattr and zfs in base, OpenBSD doesn't have either | 14:59 |
| PiotrSikora | neither* | 14:59 |
| redbo | ah, ok | 14:59 |
| redbo | I didn't notice he said zfs on FreeBSD | 14:59 |
| PiotrSikora | ah, ok :) | 15:00 |
| creiht | termie/soren: It isn't as bad as it sounds | 15:00 |
| jaypipes | PiotrSikora: either was correct :) | 15:00 |
| *** pcrews has joined #openstack | 15:00 | |
| creiht | Much better than trying to debug exceptions in long strings of callbacks :) | 15:01 |
| jaypipes | creiht: ++ | 15:01 |
| creiht | I've done quite a bit of twisted development as well, and much prefer eventlet | 15:02 |
| jaypipes | creiht: or if you've ever had the disctinct pleasure of debugging C++ metaprogramming.... | 15:02 |
| creiht | hah | 15:02 |
| * creiht doesn't miss C++ in general :) | 15:02 | |
| PiotrSikora | jaypipes: thx :P | 15:03 |
| jaypipes | PiotrSikora: :) | 15:04 |
| PiotrSikora | btw, since you guys mentioned ZFS, it would be probably killer FS for swift, since it has built-in versioning (via snapshoting) and replication (via send/recv) | 15:04 |
| *** soren has quit IRC | 15:05 | |
| *** soren has joined #openstack | 15:05 | |
| *** ChanServ sets mode: +v soren | 15:05 | |
| creiht | The problem with ZFS is that it is in a very weird state right now, especially with NetAPP being so sue happy about it right now | 15:05 |
| jaypipes | creiht: Python is indeed a joy to code in. most of the time ;) | 15:05 |
| * creiht may be a little bit tainted from his C++ days of having to use MFC | 15:06 | |
| PiotrSikora | creiht: and who they are going to sue? Oracle? ;) | 15:06 |
| *** sophiap has joined #openstack | 15:06 | |
| creiht | PiotrSikora: They have been suing people distrbuiting NAS systems that use ZFS | 15:06 |
| creiht | And it is still uncertain what Oracle is going to do with ZFS | 15:07 |
| gholt | He really likes ZFS though. | 15:07 |
| creiht | haha | 15:07 |
| PiotrSikora | creiht: they did? bastards! | 15:07 |
| PiotrSikora | creiht: well, they'll probably kill it | 15:07 |
| creiht | yeah | 15:07 |
| PiotrSikora | like they did with rest of their nice stuff | 15:07 |
| creiht | As much as I like ZFS, I'm just not certain how much of a future it has | 15:08 |
| PiotrSikora | that they got with Sun | 15:08 |
| *** mray has quit IRC | 15:08 | |
| notmyname | ya. opensolaris seems dead too | 15:08 |
| creiht | indeed :/ | 15:08 |
| PiotrSikora | yup :( | 15:08 |
| notmyname | maybe we could switch to btrfs | 15:08 |
| PiotrSikora | just when I started using it | 15:08 |
| redbo | btrfs deletes stuff so fast compared to xfs I thought it was broken | 15:09 |
| *** mrayzenoss has joined #openstack | 15:09 | |
| *** mrayzenoss is now known as mray | 15:09 | |
| *** mray has joined #openstack | 15:09 | |
| creiht | hah | 15:09 |
| jaypipes | PiotrSikora: ya, opensolaris seems to be knocking on death's door...shame. | 15:09 |
| notmyname | I've got a home fileserver running opensolaris/zfs for all backups and media and whatnot. I'm hoping i can keep it running for a while | 15:09 |
| jaypipes | redbo: lol | 15:09 |
| cory_ | when does redbofs come out? | 15:09 |
| jaypipes | hehe | 15:09 |
| PiotrSikora | redbo: lol :D | 15:09 |
| gholt | Barfs, actually. | 15:09 |
| jaypipes | :) | 15:09 |
| notmyname | cory_: we've been asking that on the swift team for a while :-) | 15:09 |
| creiht | And then BartOS | 15:09 |
| PiotrSikora | but doesn't btrfs have the same problems as zfs? | 15:10 |
| PiotrSikora | legally speaking | 15:10 |
| PiotrSikora | its also copy-on-write with checksuming on online fs checks | 15:10 |
| * creiht has no idea, except it has the deep pockets of oracle behind it | 15:10 | |
| *** mray has quit IRC | 15:10 | |
| PiotrSikora | those were the problem that NetAPP have with it, AFAIR | 15:11 |
| *** mrayzenoss has joined #openstack | 15:11 | |
| *** mrayzenoss is now known as mray | 15:11 | |
| *** mray has joined #openstack | 15:11 | |
| PiotrSikora | are* | 15:11 |
| *** sophiap has quit IRC | 15:16 | |
| creiht | PiotrSikora: replication is going to be interesting, since you have to make sure both the object file and metadata files are synced over, if one doesn't make it for some reason, it would need to be deleted | 15:18 |
| PiotrSikora | creiht: are you trying to discourage me wrong working on that? :P | 15:19 |
| PiotrSikora | s/wrong/from | 15:19 |
| creiht | PiotrSikora: not at all, just trying to let you know what you are getting into :) | 15:19 |
| PiotrSikora | duh, my english skills sux at the end of the month :P | 15:19 |
| creiht | hah | 15:19 |
| creiht | PiotrSikora: Where are you from, if you don't mind me asking? | 15:19 |
| PiotrSikora | Poland | 15:19 |
| * PiotrSikora waits while creiht opens Google Maps | 15:20 | |
| creiht | hah | 15:20 |
| cory_ | haha | 15:20 |
| * creiht knows where pland is :) | 15:21 | |
| redbo | We never forget Poland. | 15:21 |
| creiht | poland | 15:21 |
| * creiht actually has some Polish ancestry | 15:21 | |
| notmyname | isn't it right between France and Ireland? ;-) | 15:21 |
| PiotrSikora | notmyname :P | 15:21 |
| notmyname | (I actually do know the geography of Europe pretty well) | 15:22 |
| *** mattf has quit IRC | 15:24 | |
| PiotrSikora | oh, rsync has --delay-updates, which should take care of "almost-atomic replication" | 15:31 |
| creiht | hehe | 15:31 |
| PiotrSikora | it renames all files to final filenames once everything was transferred ok | 15:32 |
| redbo | yeah, but we don't want that on all the time | 15:32 |
| creiht | The only problem with that is that we rsync a partition at a time, which could be a lot of files | 15:32 |
| termie | notmyname: ireland is poland-lite | 15:32 |
| PiotrSikora | termie: nope, but london is ;) | 15:33 |
| termie | notmyname: although i guess that it is getting less popular as a destination | 15:33 |
| PiotrSikora | creiht: ugh... partition at a time? | 15:33 |
| creiht | yes | 15:33 |
| PiotrSikora | guess i'll need to understand internals more before i'll try playing with .meta | 15:34 |
| creiht | hehe | 15:34 |
| pvo | is todd from anso around? | 15:34 |
| pvo | don't know his nick | 15:34 |
| creiht | pvo: he is usually xtoddx | 15:34 |
| termie | pvo: xtoddx | 15:34 |
| pvo | ah, right | 15:35 |
| *** mattf has joined #openstack | 15:39 | |
| *** lbieber has joined #openstack | 15:40 | |
| *** littleidea has quit IRC | 15:42 | |
| *** carolinad28 has joined #openstack | 15:43 | |
| *** metoikos has quit IRC | 15:46 | |
| *** sophiap has joined #openstack | 15:46 | |
| jaypipes | PiotrSikora: don't worry, my Polish skills suck at the end of the month, too. | 15:47 |
| PiotrSikora | jaypipes :) | 15:48 |
| jaypipes | termie: so have all your email woes been fixed? | 15:48 |
| *** metoikos has joined #openstack | 15:53 | |
| *** Guest53283 is now known as spy | 15:56 | |
| *** asksol has joined #openstack | 16:02 | |
| *** littleidea has joined #openstack | 16:10 | |
| *** holoway has quit IRC | 16:11 | |
| *** holoway has joined #openstack | 16:13 | |
| *** sophiap has quit IRC | 16:16 | |
| mtaylor | lbieber: morning! | 16:21 |
| lbieber | mtaylor: good morning! | 16:21 |
| *** jdmaturen has joined #openstack | 16:22 | |
| *** jsmith-away is now known as jsmith | 16:38 | |
| *** littleidea has quit IRC | 16:40 | |
| *** blpiatt has joined #openstack | 16:58 | |
| *** jsmith is now known as jsmith-away | 17:09 | |
| *** blpiatt has quit IRC | 17:10 | |
| jaypipes | vish1: around? question for you... | 17:17 |
| *** kallistec has quit IRC | 17:20 | |
| *** Protocol_ has joined #openstack | 17:21 | |
| *** Protocol_ has left #openstack | 17:21 | |
| *** maximilien has quit IRC | 17:23 | |
| *** jc_smith has joined #openstack | 17:25 | |
| *** carolinad28 has quit IRC | 17:33 | |
| *** dendrobates is now known as dendro-afk | 17:43 | |
| *** dendro-afk is now known as dendrobates | 17:43 | |
| *** jdmaturen has quit IRC | 17:47 | |
| *** kallistec has joined #openstack | 17:50 | |
| *** mattf has quit IRC | 17:50 | |
| *** jdmaturen has joined #openstack | 17:50 | |
| *** carolinad28 has joined #openstack | 17:50 | |
| *** doublerr has left #openstack | 17:54 | |
| *** mattf has joined #openstack | 18:02 | |
| eday | mtaylor: you drizzle post ended up on planet openstack :) | 18:02 |
| mtaylor | eday: I broke tag filtering on my blog - fixing is on my todo list | 18:04 |
| mtaylor | eday: I broke _rss feed_ tag filtering on my blog - fixing is on my todo list | 18:04 |
| eday | hehe | 18:04 |
| mtaylor | so sorry for the planet spam everyone :) | 18:04 |
| mtaylor | eday: it's ok - it means my openstack posts will wind up on planetmysql too :) | 18:04 |
| eday | mtaylor: unacceptable, I'm removing you. | 18:05 |
| * mtaylor quickly removes eday's commit access | 18:05 | |
| eday | mtaylor: either that or you get 5 lashes | 18:06 |
| * mtaylor quickly gives eday his commit access back, realizing he doesn't really want to have to directly fix eday's stuff himself | 18:06 | |
| *** maximilien has joined #openstack | 18:07 | |
| *** greenisus has joined #openstack | 18:09 | |
| *** spy is now known as antonym | 18:09 | |
| greenisus | hey guys, i have someone wanting to contribute a patch on the control panel side, but neither one of us knows how to do that with bzr :( | 18:10 |
| greenisus | anyone know how he would send me the equivalent of a git pull request | 18:10 |
| *** subirjolly has joined #openstack | 18:11 | |
| creiht | greenisus: http://wiki.openstack.org/LifeWithBzrAndLaunchpad | 18:11 |
| greenisus | duh, i should have RTFM | 18:12 |
| creiht | :) | 18:12 |
| *** tc__ has joined #openstack | 18:14 | |
| *** tc__ has quit IRC | 18:18 | |
| *** yyyeahTc has joined #openstack | 18:19 | |
| *** allsystemsarego has quit IRC | 18:19 | |
| uvirtbot | New bug: #611799 in openstack-web-control-panel "Incorrect Background Image URL" [Undecided,New] https://launchpad.net/bugs/611799 | 18:21 |
| mtaylor | eday: oh - I totally forgot I was playing with that vala echo server - are you gonna push your scalestack changes? | 18:21 |
| soren | greenisus: There are the concept of "bundles", but really people should just push a branch to Launchpad. It's way easier. | 18:21 |
| soren | greenisus: Where can I see the control panel stuff? | 18:21 |
| *** tjyang_ has joined #openstack | 18:22 | |
| notmyname | mtaylor: on http://wiki.openstack.org/AddingYourBlog there is reference to a heads dir, but the commands seem to mix it with the images dir. what should I be using? | 18:23 |
| mtaylor | notmyname: images | 18:23 |
| mtaylor | notmyname: fixing page | 18:24 |
| notmyname | hmm...ok. I need to rexeamine it then. I'm not seeing my mug show up | 18:24 |
| greenisus | soren: all the projects are here https://launchpad.net/openstack | 18:24 |
| mtaylor | notmyname: thanks! | 18:24 |
| greenisus | soren: OpenStack Web Control Panel, OpenStack iPhone App, OpenStack Android App, OpenStack iPad App | 18:24 |
| soren | greenisus: Oh, sure, I meant screenshots and stuff. | 18:24 |
| greenisus | oh | 18:24 |
| soren | Sorry, I can see now that I was vague :) | 18:25 |
| *** jsmith-away is now known as jsmith | 18:25 | |
| greenisus | soren: here's one http://c0222312.cdn.cloudfiles.rackspacecloud.com/main-99.png | 18:25 |
| mtaylor | greenisus: damn man. you're _still_ on my automation todo list too. sorry for the long delay. :) | 18:25 |
| greenisus | it's out of date, but that's the general idea. it's what i was showing in austin when i was begging everyone for JSONP support | 18:25 |
| greenisus | it's all good mtaylor. i'm still just catching up with bzr as i go along | 18:25 |
| soren | greenisus: JSONP, as I understand it, is a JSON variant where you basically avoid having to eval stuff, but can just src it? | 18:26 |
| greenisus | just so you know, at this point the control panel and mobile apps are speaking rackspace cloud api only | 18:26 |
| mtaylor | greenisus: best way to do it ... needing to do something, then learning. :) | 18:26 |
| greenisus | mtaylor: that's exactly how i'm learning bzr | 18:26 |
| greenisus | and how i learned Android for that matter | 18:26 |
| mtaylor | that's how I learn _everything_ | 18:26 |
| greenisus | soren: that's right about JSONP. basically you'd make an API call like GET /whatever?callback=myCallback. then, the response would be: "myCallback({ the json for whatever you were asking for };" | 18:27 |
| greenisus | that way, you can make API calls in a web app by inserting <script> tags into the DOM | 18:28 |
| greenisus | so ultimately, if we have JSONP support, we can make a control panel that's entirely html/css/javascript, with no backend at all | 18:28 |
| creiht | soren: the main reason for it is to get around same-origin security in the browser | 18:28 |
| greenisus | we could serve the control panel entirely from CDN | 18:28 |
| greenisus | creiht is right. | 18:28 |
| greenisus | i was really hoping we could get JSONP, so i did this app with a dumb jsonp proxy in ruby that just does the JSONP style and passes it along to the real API | 18:29 |
| greenisus | but if we do JSONP, the proxy can go away | 18:29 |
| greenisus | am i making sense? | 18:29 |
| *** littleidea has joined #openstack | 18:29 | |
| eday | mtaylor: ahh yeah, I never really got it working properly (vala echo). I'll get the scale stack changes up shortly | 18:30 |
| mtaylor | eday: I've got the backlog thing fixed | 18:30 |
| mtaylor | eday: I'm asking the vala people about the other... | 18:30 |
| mtaylor | eday: I _think_ it may be vala utf8 encoding strings by default | 18:30 |
| creiht | The only problem with JSONP (at least with current rackspace style auth) is that you would have to have your auth creds in the javascript | 18:30 |
| mtaylor | eday: looking for a way to skip that | 18:30 |
| justinsb | Anyone seen this while running the Nova unit tests: ImportError: No module named amqplib.client_0_8.connection | 18:30 |
| eday | mtaylor: ahh | 18:31 |
| mtaylor | eday: just a guess though | 18:32 |
| greenisus | yeah that's true. hopefully we can get around that somehow | 18:32 |
| *** rdw has quit IRC | 18:33 | |
| justinsb | (answering my own question) ImportError is because I had Debian's python-amqplib package installed | 18:34 |
| *** lbieber has quit IRC | 18:41 | |
| *** gustavomzw has joined #openstack | 18:43 | |
| soren | greenisus: Yes, making perfect sense. | 19:01 |
| soren | greenisus: I see no reason we couldn't add jsonp support. It sounds like a perfectly reasonable idea. | 19:01 |
| *** jsmith is now known as jsmith-away | 19:02 | |
| soren | creiht: I don't quite see why the auth stuff doesn't use an actual http cookie. | 19:02 |
| greenisus | soren: thank you :) i'm really happy about it | 19:02 |
| creiht | soren: Because that's not how rackspace auth works? :) | 19:03 |
| soren | creiht: Instead of a piece of data that does the exact same thing. You get it from the server, and the server expect you to pass it with each request. | 19:03 |
| soren | creiht: Right, exactly. I just don't get why it doesn't. It seems /really/ weird to me. | 19:03 |
| gholt | Cookies are not restful, right? | 19:03 |
| greenisus | yeah i'm assuming a <script> GET would send a cookie, right? as long as we handled that.... | 19:04 |
| greenisus | gholt: JSONP is not restful. it's a hack to get around browser security policies | 19:04 |
| gustavomzw | irclogs is returning 403 - http://irclogs.openstack.org/logs/2010-07/%23openstack.29.log | 19:05 |
| gholt | Ah, I was more trying to answer of why our auth didn't use cookies. But I did jump in at weird point in the conversation. | 19:05 |
| creiht | Also in the long run, auth is going to be one of those things that is likely to vary a lot from installation to installation | 19:05 |
| creiht | since most places already have some sort of auth that they will want to use | 19:06 |
| greenisus | right. auth is one of the weird places for these projects. i'm assuming we're going to have some sort of reference auth implementation though | 19:08 |
| greenisus | so that's what i'll go with for these apps | 19:08 |
| greenisus | at this point i'm assuming rackspace style, but i'd really like for us to do better | 19:08 |
| greenisus | i'd especially like for us to support some sort of human password auth | 19:08 |
| greenisus | (like for rackspace, letting someone auth to the api with their manage.rackspacecloud.com password) | 19:09 |
| greenisus | the reason for that is that typing a huge API key on a cell phone is a miserable experience | 19:09 |
| soren | OpenID + OAuth are high on my list of wishes for a authentication. | 19:09 |
| creiht | Yeah, not sure what is going to happen on the auth side of things | 19:09 |
| greenisus | yeah i'm down with that | 19:09 |
| creiht | OAuth++ | 19:09 |
| notmyname | greenisus: swift has a reference implementation that is compatible with RAX auth | 19:09 |
| soren | That and HOTP/TOTP | 19:09 |
| greenisus | notmyname: okay good. i need to dig into that | 19:09 |
| greenisus | have to admit i've been very lightweight on looking at swift and nova | 19:10 |
| mtaylor | eday: ooh. I think I may have gotten it working | 19:10 |
| notmyname | greenisus: on a related note, that's one reason I use Cloud Mobile on the iPhone rather than your app | 19:10 |
| greenisus | notmyname: exactly! but i wasn't comfortable screen scraping for an API key | 19:11 |
| notmyname | not sure that would be necessary. typing an API key is painful. i use 1password and copy/paste it | 19:11 |
| notmyname | Cloud Mobile has 3 entry fields: username, api key, and auth server | 19:12 |
| greenisus | notmyname: yeah, i copy/paste it too, but not everyone realizes they can do that. so in the App Store i end up with all these bad reviews because no one wants to type the key | 19:12 |
| greenisus | wait, Cloud Mobile prompts for the API key as well | 19:12 |
| notmyname | ya | 19:12 |
| mtaylor | eday: pull from lp:~mordred/+junk/vala-echo-test and try again | 19:12 |
| notmyname | greenisus: sorry, I was told I wasn't being clear. One reason I like Cloud Mobile better is because I can set the auth server | 19:15 |
| eday | mtaylor: it's saving the buffer, so no | 19:16 |
| eday | mtaylor: nc to it and type a few lines, you'll see what I mean | 19:16 |
| mtaylor | eday: well piss | 19:16 |
| mtaylor | eday: oh - you mean I need to make it flush the buffer before I read more into it don't you? | 19:17 |
| eday | oh, you're peeking, not reading | 19:17 |
| eday | must be a simple way to just read instead of peek | 19:18 |
| mtaylor | eday: potentially - lemme go look at api again | 19:18 |
| soren | mtaylor: Did you have any ideas for the failing unit test on the buildd's? | 19:18 |
| mtaylor | soren: the failedtobuild I saw looked like some weirdness with doc/build not being there ... is there something else you're referencing? | 19:21 |
| soren | mtaylor: Look again | 19:22 |
| soren | mtaylor: redis.exceptions.ConnectionError: Error 111 connecting 127.0.0.1:6379. Connection refused. | 19:23 |
| soren | mtaylor: It wants a running redis-server. | 19:23 |
| soren | Maybe this is less of a problem than I think it is. | 19:23 |
| mtaylor | soren: ah yes! well, we should put in a start-redis command into debian/rules | 19:24 |
| soren | Do buildd's have a policy-rc.d that stops them from running daemons? | 19:24 |
| soren | mtaylor: Yeah, that will fix it on the buildd's, but probably make stuff weird on local dev boxes. | 19:24 |
| mtaylor | I'm not sure- in libmemcached and drizzle I have the test suite start daemons one a random open port and then use that port in the test suite | 19:24 |
| *** stewart_ has joined #openstack | 19:25 | |
| mtaylor | soren: we should probably do something similar, because having make test connect to a system instance of a service is a good way to accidentally hose data when someone is doing a rollout and building from source | 19:25 |
| *** stewart has quit IRC | 19:25 | |
| *** stewart_ is now known as stewart | 19:26 | |
| soren | mtaylor: That's totally what we should do. | 19:26 |
| stewart | mtaylor, i was thinking we should launch a private storage thing for drizzle tests of the cloud engine. | 19:26 |
| soren | mtaylor: Sorry, I'm not thinking entirely clearly. It's been a long week. | 19:26 |
| mtaylor | soren: yes. busy around here and stuff | 19:27 |
| mtaylor | stewart: we have a cloud engine? | 19:27 |
| stewart | mtaylor, a SoC project | 19:27 |
| stewart | mtaylor, also for pbms | 19:27 |
| mtaylor | stewart: but yeah - I'm working on hudson plugins to spin up cloud resources | 19:27 |
| stewart | mtaylor, if we can do it as part of the test run... it means people can actually (easily) run "make test" and test everything too | 19:28 |
| stewart | (assuming that installing a single machine small cloud storage thing isn't batshit insanely hard to do) | 19:28 |
| greenisus | notmyname: oh i see. i'll be adding that to the openstack app. i never did for rackspace because it doesn't change (as far as i know) | 19:28 |
| mtaylor | stewart: well... right- we'll get that sorted - but I also want to test it in hudson against a real cloud thing | 19:28 |
| *** rdw has joined #openstack | 19:29 | |
| greenisus | notmyname: i generally try to have as few settings as possible, because they come at a cost | 19:29 |
| stewart | mtaylor, yeah | 19:29 |
| notmyname | greenisus: the only use case I know of was when were were testing swift internally and I wanted to point it at our staging env :-) | 19:29 |
| notmyname | but that has an audience of about 4 | 19:29 |
| notmyname | it makes a lot of sense on the openstack app, though | 19:29 |
| *** allsystemsarego has joined #openstack | 19:30 | |
| greenisus | notmyname: gotcha. you guys should have reached out to me :) i would have sent you a configurable ad hoc build | 19:30 |
| greenisus | notmyname: you also could have gotten the source at github if you have a mac | 19:30 |
| notmyname | greenisus: while we're talking about features.... the other think I like about cloud mobile is the additional metadata that is shown about containers and objects | 19:31 |
| *** jsmith-away is now known as jsmith | 19:31 | |
| greenisus | notmyname: yeah, i always meant to add that and never got around to it | 19:31 |
| greenisus | notmyname: i want it on servers too, but they're not editable so i didn't do it | 19:32 |
| notmyname | greenisus: will the openstack app allow for multiple accounts? | 19:34 |
| greenisus | notmyname: yes. that's currently in progress, and the ipad app already does | 19:34 |
| greenisus | notmyname: eventually, i'll merge the iphone/ipad apps into a single universal app | 19:34 |
| notmyname | very nice. sounds like I'm going to like it | 19:34 |
| greenisus | yeah i have a lot of big ideas for it. right now i'm working on hooking it up to chef | 19:35 |
| greenisus | provision nodes into a role so it can bootstrap itself | 19:35 |
| greenisus | really hoping others get involved in the mobile apps so we can do more than just the stuff i come up with | 19:38 |
| *** jakedahn has joined #openstack | 19:38 | |
| * notmyname needs to learn ObjC first | 19:38 | |
| greenisus | notmyname: learning objective-c and cocoa is definitely the biggest hurdle :) i find most iphone devs would rather go off on their own trying to strike it rich on the app store anyway | 19:39 |
| *** n0cturne has joined #openstack | 19:43 | |
| *** jsmith is now known as jsmith-away | 19:57 | |
| *** n0cturne has quit IRC | 19:58 | |
| *** n0cturne has joined #openstack | 19:58 | |
| *** Podilarius has quit IRC | 20:01 | |
| *** Podilarius has joined #openstack | 20:07 | |
| *** carolinad28 has quit IRC | 20:08 | |
| creiht | mtaylor: how do I fix the conflic in https://code.launchpad.net/~cthier/swift/admin_guide/+merge/31419 | 20:10 |
| creiht | ? | 20:10 |
| creiht | conflice in .bzrignore | 20:11 |
| uvirtbot | New bug: #611839 in swift "Add stats report documentation to admin guide" [Medium,Triaged] https://launchpad.net/bugs/611839 | 20:11 |
| creiht | man I can't type today... must be friday | 20:12 |
| *** justinsheehy has quit IRC | 20:12 | |
| mtaylor | creiht: you need to re-merge your branch with trunk, fix, and push again | 20:12 |
| mtaylor | creiht: so when you merge with trunk, you should see the conflict in that file - fix it, run "bzr resolve .bzrignore" and then commit | 20:13 |
| creiht | k | 20:13 |
| creiht | so I need to branch from trunk, merge my branch into that new branch and then push that? | 20:13 |
| creiht | mtaylor: ok I think I got it | 20:19 |
| *** antonym has quit IRC | 20:23 | |
| *** antonym has joined #openstack | 20:23 | |
| *** asimov.freenode.net sets mode: +v antonym | 20:23 | |
| *** snowboarder04 has quit IRC | 20:25 | |
| *** thinkscientist has joined #openstack | 20:25 | |
| *** pvo has quit IRC | 20:25 | |
| *** pvo has joined #openstack | 20:25 | |
| *** asimov.freenode.net sets mode: +v pvo | 20:25 | |
| *** lbieber has joined #openstack | 20:25 | |
| *** jonesy_ has quit IRC | 20:45 | |
| *** aliguori has quit IRC | 20:56 | |
| *** justinsheehy has joined #openstack | 20:56 | |
| gustavomzw | following http://wiki.openstack.org/InstallationNova20100729 - install_venv breaks installing python-gflags | 20:59 |
| *** justinsheehy_ has joined #openstack | 21:01 | |
| *** justinsheehy has quit IRC | 21:02 | |
| *** justinsheehy_ is now known as justinsheehy | 21:02 | |
| *** allsystemsarego has quit IRC | 21:02 | |
| *** jakedahn has quit IRC | 21:03 | |
| *** jakedahn has joined #openstack | 21:03 | |
| *** jakedahn has joined #openstack | 21:03 | |
| *** n0cturne has quit IRC | 21:05 | |
| *** 64MAAAAEF has quit IRC | 21:06 | |
| *** aliguori has joined #openstack | 21:11 | |
| gustavomzw | when I run pip install using with_venv.sh it installs | 21:11 |
| jc_smith | I had to do easy_install http://python-gflags.googlecode.com/files/python_gflags-1.3-py2.5.egg | 21:12 |
| *** justinsheehy has quit IRC | 21:15 | |
| vish1 | anyone here? | 21:16 |
| vish1 | I have a dependency question | 21:16 |
| vish1 | should we put python-ldap in the debian dependencies? | 21:16 |
| vish1 | my installs are choking on not having it | 21:18 |
| vish1 | in general the question is, if we have optional plugins, do we install the deps for them? | 21:19 |
| jc_smith | did you run /nova/auth/slap.sh | 21:20 |
| gustavomzw | install_venv using pip -E breaks, running pip install commands using with_venv.sh works | 21:22 |
| *** sirp1 has joined #openstack | 21:23 | |
| *** gnrfan has joined #openstack | 21:23 | |
| creiht | is there a reason to use gflags rather than the built in OptParse? | 21:26 |
| *** justinsheehy has joined #openstack | 21:26 | |
| *** mtaylor has quit IRC | 21:28 | |
| jc_smith | creiht: it was part of the deps for nova, but I found this on the site with regard to OptParse: the ability to define flags in the source file in which they're used | 21:30 |
| *** gnrfan has left #openstack | 21:30 | |
| creiht | k.. just curious | 21:30 |
| jc_smith | creiht: in swift, what is being stored in memcache via the proxy and via the auth server? looks like the proxy is caching response codes? | 21:32 |
| jc_smith | result codes, rather | 21:32 |
| jc_smith | is this for like negative caching for people stating files that don't exist? | 21:33 |
| creiht | In the proxy we store auth tokens, and account/container existance | 21:33 |
| creiht | I think we also store error counts, so that we can error limit nodes if it is erroring out too much | 21:33 |
| * creiht goes to look | 21:34 | |
| *** jk0 has quit IRC | 21:34 | |
| *** jk0 has joined #openstack | 21:34 | |
| *** asimov.freenode.net sets mode: +v jk0 | 21:34 | |
| jc_smith | oh, I can go dig, no worries | 21:34 |
| creiht | And we also use it for rate limiting | 21:34 |
| jc_smith | you just seem to know everything about swift ;) | 21:34 |
| *** justinsb_ has joined #openstack | 21:34 | |
| creiht | hah | 21:34 |
| creiht | I know little bits, and think I remember other bits | 21:35 |
| *** jc_smith has quit IRC | 21:35 | |
| creiht | :) | 21:35 |
| creiht | for account/container existence we cache the result code | 21:35 |
| *** zenmatt has joined #openstack | 21:35 | |
| creiht | 200, it is there, 404 it isn't :) | 21:36 |
| *** jc_smith has joined #openstack | 21:37 | |
| creiht | jc_smith: for account/container existence we cache the result code | 21:37 |
| creiht | that way the check for existence is the same weather we got the result from a webservice call or from memcache | 21:38 |
| jc_smith | ah, ok thats what I was looking at | 21:38 |
| *** mattf has quit IRC | 21:38 | |
| creiht | and I guess the error_limiting is in process | 21:38 |
| creiht | not in memcache | 21:38 |
| notmyname | if my error limiting stuff is still in it, rate limiting uses memcache to store the current rate counts for accounts, etc | 21:39 |
| creiht | notmyname: correct | 21:39 |
| creiht | jc_smith: and I like refreshing my memory, because you aren't the only one that comes in with questions :) | 21:39 |
| jc_smith | if I had two proxies, would there be two separate memcaches independant of one another? | 21:40 |
| creiht | No the memcaches are shared accross the proxies | 21:40 |
| creiht | I actually have docs awaiting merge that explain some of this :) | 21:41 |
| creiht | The memcached servers are in the proxy-server.conf | 21:41 |
| notmyname | you should get someone to review it :-) | 21:41 |
| * creiht smacks notmyname | 21:41 | |
| creiht | :) | 21:41 |
| creiht | jc_smith: https://code.launchpad.net/~cthier/swift/admin_guide/+merge/31419 | 21:41 |
| creiht | if you want a preview :) | 21:42 |
| jc_smith | hmm, my proxy-server.conf doesn't have a memcache specified | 21:42 |
| *** _buck has quit IRC | 21:42 | |
| jc_smith | I assume that it uses the local one then | 21:42 |
| creiht | It defaults to localhost if not specified | 21:42 |
| creiht | otherwise it is a comma separated list of ip:port | 21:43 |
| notmyname | I see a typo in your docs! I'll have to deny your merge request :-) | 21:43 |
| creiht | hah | 21:44 |
| notmyname | line 279 of the diff | 21:44 |
| creiht | hehe | 21:44 |
| jk0 | :D | 21:45 |
| gustavomzw | InstallationNova20100729 is the cleanest guide | 21:45 |
| *** tjyang has quit IRC | 21:45 | |
| jc_smith | creiht: great doc, cw an I were just talking this morning about what to do about disk failures and stuff, this will help a lot | 21:45 |
| creiht | cool | 21:46 |
| creiht | jc_smith: let me know if there are other questions | 21:46 |
| jc_smith | thanks | 21:46 |
| creiht | so we can get them in | 21:46 |
| creiht | Though, I may take a break from docs next week :) | 21:46 |
| redbo | we were gonna automate it, but then we were like pandemicsyn is on call all weekend anyway | 21:46 |
| creiht | haha | 21:47 |
| creiht | notmyname: fix pushed... should be updated shortly | 21:47 |
| * creiht waits for lp's eventual consistency to kick in | 21:48 | |
| *** justinsb_ has quit IRC | 21:48 | |
| notmyname | yay | 21:48 |
| *** jhalogen has quit IRC | 21:49 | |
| creiht | and there it is | 21:53 |
| redbo | how do I merge trunk into my branch? | 21:56 |
| creiht | hrm... mtaylor is gone already | 21:57 |
| _0x44 | redbo: bzr merge lp:foo/trunk? | 21:57 |
| notmyname | creiht: reviewed and approved | 21:57 |
| creiht | With the way we do merge proposals, I wonder if it is alright to merge directly from trunk to your branch | 21:58 |
| creiht | notmyname: awesome thanks | 21:58 |
| redbo | should be fine | 21:58 |
| creiht | redbo: if you have trunk locally you can try: bzr merge ../trunk | 21:59 |
| creiht | from your branch | 21:59 |
| redbo | _0x44: thanks, bzr merge lp:swift seemed to work | 21:59 |
| *** Phil_FL has quit IRC | 21:59 | |
| _0x44 | redbo: No problem, please don't tell the ozone guys I helped ;) | 21:59 |
| creiht | hah | 21:59 |
| * jk0 saw it | 21:59 | |
| * jk0 is going to tell | 21:59 | |
| *** selenamarie has quit IRC | 22:00 | |
| *** selenamarie has joined #openstack | 22:00 | |
| * _0x44 is never going to hear the end of that. "You helped with bzr? Who are you?" | 22:01 | |
| creiht | haha | 22:01 |
| pvo | _0x44: impostor! | 22:01 |
| jc_smith | creiht: for memcache on the proxies, would it be best to put the ip of both proxies? so they are pointing to each other? | 22:02 |
| creiht | correct | 22:02 |
| creiht | So in our case, we run memcache on all of our proxies, and each proxy-server.conf has the same line, listing the ip:port of all the memcache servers | 22:03 |
| redbo | you know, we probably don't cache anything in swift long enough for that memcached consistent hashing to have any point. | 22:03 |
| _0x44 | pvo: I would have gotten away with it too, if it hadn't been for you meddling kids and your talking Wisconsiner too. | 22:03 |
| creiht | As a side note, if this is on a public network, you might want to run memcache on a private network, if available | 22:04 |
| jc_smith | creight: thanks | 22:04 |
| creiht | redbo: I think it helps for highly concurrent users, but in general you are probably right | 22:04 |
| redbo | well, modular hashing would distribute fine and take less work | 22:05 |
| notmyname | redbo is normally probably right | 22:05 |
| creiht | redbo: would it still work to handle failure cases sanely? | 22:05 |
| redbo | you think it handles failure cases sanely? | 22:06 |
| creiht | well somewhat sanely | 22:06 |
| *** littleidea has quit IRC | 22:06 | |
| creiht | Does it still handoff to another memcache server, if one of them goes down? | 22:06 |
| redbo | the only thing consistent hashing gives you is if you add or remove a server, less stuff moves around | 22:06 |
| *** antonym is now known as spy | 22:06 | |
| creiht | true | 22:06 |
| * creiht wonders why antonym likes to spy so much | 22:07 | |
| _0x44 | creiht: It's because he's required to maintain a 100' distance from most of us. | 22:07 |
| creiht | heh | 22:07 |
| redbo | We never add or remove memcached servers anyway. | 22:08 |
| creiht | true | 22:08 |
| spy | creiht: finally reclaimed my nick | 22:08 |
| creiht | spy: ahh cool | 22:08 |
| spy | one of the freenode staff hadn't used it in 5 years | 22:08 |
| spy | gave it to me when i asked for it heh | 22:09 |
| redbo | and the once every bazillion years we do add or remove memcached nodes, we'd lose like... our auth token caches. | 22:10 |
| creiht | hehe | 22:10 |
| creiht | Maybe at the time we were thinking that we would use memcache more? | 22:11 |
| redbo | yeah | 22:11 |
| redbo | eh, no compelling reason to remove it now though. | 22:13 |
| creiht | There are some new docs up: http://swift.openstack.org/ | 22:16 |
| creiht | Notably a start to an Administrator's Guide, and added some to the Deployment guide | 22:17 |
| *** Podilarius has left #openstack | 22:19 | |
| *** gundlach has quit IRC | 22:22 | |
| *** joearnold has joined #openstack | 22:25 | |
| *** pvo has quit IRC | 22:26 | |
| creiht | ohhh... how do I get openstack in my nick? | 22:26 |
| *** mray has quit IRC | 22:29 | |
| *** gustavomzw has quit IRC | 22:29 | |
| *** jdmaturen has quit IRC | 22:33 | |
| *** strtok has joined #openstack | 22:35 | |
| *** subirjolly has quit IRC | 22:36 | |
| rbergeron | is the list of planet pages in the sidebar on planet.openstack.org pulled from planetplanet.org? or manually selected? | 22:38 |
| jk0 | rbergeron: under the Subscriptions? | 22:48 |
| rbergeron | jk0: no, planetarium. | 22:48 |
| rbergeron | sorry, should have scrolled up to see the distinction. ;) | 22:49 |
| rbergeron | but yes, the planetarium - the fedora link (sorry to toot the horn here) is like... woefully such a 2007 link, but I didn't want to bother anyone if it's a feed from elsewhere. | 22:49 |
| jk0 | rbergeron: I think those are just sitting in the template, but I could be wrong | 22:56 |
| *** lbieber has quit IRC | 22:58 | |
| *** msparks has joined #openstack | 23:01 | |
| msparks | contributing. | 23:01 |
| *** anm_ has left #openstack | 23:04 | |
| *** thinkscientist has quit IRC | 23:17 | |
| *** pvo has joined #openstack | 23:18 | |
| *** pvo has joined #openstack | 23:18 | |
| *** ChanServ sets mode: +v pvo | 23:18 | |
| *** wattersjames has joined #openstack | 23:19 | |
| *** thinkscientist has joined #openstack | 23:22 | |
| *** littleidea has joined #openstack | 23:22 | |
| *** wattersjames_ has joined #openstack | 23:26 | |
| *** jakedahn has quit IRC | 23:27 | |
| *** wattersjames_ has quit IRC | 23:28 | |
| *** jonesy_ has joined #openstack | 23:28 | |
| *** wattersjames_ has joined #openstack | 23:29 | |
| *** jakedahn has joined #openstack | 23:29 | |
| *** wattersjames has quit IRC | 23:30 | |
| *** wattersjames_ is now known as wattersjames | 23:30 | |
| *** kevnfx has joined #openstack | 23:32 | |
| *** aliguori has quit IRC | 23:32 | |
| *** wattersjames has joined #openstack | 23:38 | |
| *** ghchinoy has left #openstack | 23:49 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
