| nelson__ | Hrm. I have setuptools installed. When I look at trunk/README, it says that I can build the documentation using "python setup.py build_sphinx" | 00:01 |
|---|---|---|
| nelson__ | does that work for anybody else? | 00:01 |
| *** MikeZ1 has quit IRC | 00:04 | |
| creiht | nelson__: install python sphinx | 00:07 |
| nelson__ | I'll bet that will fix it. Sounds like something to add to my documentation fixes. :) | 00:08 |
| nelson__ | File "/home/nelson/openstack/trunk/doc/source/./_theme/layout.html", line 2, in top-level template code | 00:09 |
| nelson__ | {% set css_files = css_files + ['_static/tweaks.css'] %} | 00:09 |
| nelson__ | jinja2.exceptions.UndefinedError: 'css_files' is undefined | 00:09 |
| nelson__ | which sounds like it expects that css_files was defined by sphinx_doc/layout.html | 00:11 |
| creiht | nelson__: is this for swift, or nova? | 00:12 |
| nelson__ | swift | 00:12 |
| creiht | hrm | 00:13 |
| creiht | nelson__: which version of sphinx? | 00:14 |
| creiht | nelson__: current trunk builds for me | 00:15 |
| *** jimbaker has quit IRC | 00:17 | |
| *** MikeZ has joined #openstack | 00:18 | |
| *** Cybo has quit IRC | 00:30 | |
| *** ccustine has quit IRC | 00:32 | |
| nelson__ | python-sphinx (0.6.4-1) | 00:32 |
| *** reldan has quit IRC | 00:33 | |
| nelson__ | On Ubuntu 10.04.1 | 00:33 |
| *** reldan has joined #openstack | 00:35 | |
| nelson__ | argh. And auth-server is silently crashing. | 00:40 |
| Ryan_Lane | nelson__: you running trunk, or a stable release? | 00:41 |
| nelson__ | trunk | 00:41 |
| * Ryan_Lane nods | 00:41 | |
| *** dendrobates is now known as dendro-afk | 00:41 | |
| nelson__ | Doh! The config file is b0rked. swift-auth-server is printing an appropriate error to fd1, but swift-init isn't printing that to stderr | 00:43 |
| *** winston-d has joined #openstack | 00:44 | |
| *** Cybo has joined #openstack | 00:45 | |
| nelson__ | fixing the config file solved the problem and swift-auth-server is now running, but ... but ... swift-init ought to print an error message, | 00:45 |
| nelson__ | particularly if the user install instructions are going to tell people to run "swift-init auth start" | 00:45 |
| nelson__ | and there's nothing in any /var/log/ file mentioning swift. | 00:47 |
| *** dirakx has joined #openstack | 00:48 | |
| nelson__ | and without meaning to whine *too* loudly, but there's no man page for /usr/bin/swift* in the 'swift' package. | 00:49 |
| *** Cybo has quit IRC | 00:49 | |
| *** gaveen has quit IRC | 00:51 | |
| nelson__ | haha, oops, the first thing swift-init does is close fd1. :) | 00:51 |
| * nelson__ rewrites swift-init so that it relies on Dan Bernstein's daemontools. | 00:52 | |
| *** jdurgin has quit IRC | 00:52 | |
| nelson__ | sudo apt-get install daemontools daemontools-run | 00:56 |
| *** Jordandev has joined #openstack | 00:56 | |
| *** reldan has quit IRC | 00:56 | |
| alekibango | why not runit? | 00:56 |
| termie | upstart | 00:57 |
| nelson__ | because daemontools handles logging of program output. | 00:58 |
| nelson__ | it also lets you send signals to running programs, | 00:58 |
| *** dirakx has quit IRC | 00:59 | |
| nelson__ | without disgusting hackery like /var/run/*.pid | 00:59 |
| termie | nelson__: making things rely on a specific implementation is generally a bad thing | 00:59 |
| termie | there are quite a lot of people who vehemently hate daemontools | 00:59 |
| nelson__ | daemontools runs everywhere | 00:59 |
| termie | that isn't much of an argument | 01:00 |
| nelson__ | I'm sure they have bad reasons for hating daemontools ... starting with the personality of the author. | 01:00 |
| termie | there are plenty of things to dislike about it | 01:00 |
| nelson__ | daemontools running everywhere is a GREAT argument against "making things rely on a specific implementation is generally a bad thing" | 01:00 |
| nelson__ | termie: like what? | 01:01 |
| nelson__ | daemontools solves a lot of stupid little problems all at once. | 01:01 |
| termie | nelson__: it has a very separate environment from the rest of the system and does not integrate with standard setups nor is supported by the os writers | 01:01 |
| alekibango | nelson__: and introduces some of its own :) | 01:02 |
| nelson__ | so put it in /etc/service like the debian/ubuntu packages does. Next? | 01:02 |
| alekibango | nelson__: did you read license? | 01:02 |
| nelson__ | public domain. next? | 01:02 |
| *** anotherjesse has joined #openstack | 01:02 | |
| alekibango | really?, i need to check that :) | 01:02 |
| termie | nelson__: well first off, talk about the decision on the list | 01:03 |
| termie | alekibango: yeah it is public domain | 01:03 |
| nelson__ | termie: that's a reason not to change anything, not a reason not to use daemontools. Next? | 01:03 |
| nelson__ | So far, #define plenty 2 | 01:03 |
| *** jc_smith has quit IRC | 01:03 | |
| Xenith | I think the question is if nelson__ is making daemontools required or not. | 01:04 |
| Xenith | Nothing wrong with making it supported but optional | 01:04 |
| termie | nelson__: daemontools is a perfectly reasonable way to run the code, it is not a reasonable dependency | 01:04 |
| *** Cybo has joined #openstack | 01:04 | |
| nelson__ | and neither of those reasons were true so far. | 01:04 |
| nelson__ | termie: I'm looking for rationality here, not hatred of djb. | 01:05 |
| nelson__ | (and I will cheerfully acknowledge that he has been an ass in the past) | 01:05 |
| termie | nelson__: i don't know djb, or care about him, i have used daemontools before and I personally don't like it and I am not alone in that | 01:05 |
| termie | nelson__: i offered three reasons that you skipped over | 01:05 |
| nelson__ | why don't you like it? | 01:05 |
| termie | nelson__: scroll back | 01:05 |
| nelson__ | I've answered all three of your reasons. | 01:05 |
| Ryan_Lane | nelson__: we are unlikely to use daemontools | 01:06 |
| termie | nelson__: putting it in /etc/service is not a solution to that problem | 01:06 |
| Ryan_Lane | we are going to use init.d, and start-stop-daemon | 01:06 |
| nelson__ | Ryan_Lane: I'm sorry to hear that. Do you have a rational reason for objecting to a superior solution? | 01:06 |
| nelson__ | Or again, are you a djb-hater? | 01:06 |
| nelson__ | (which is fine ... it's just irrational) | 01:06 |
| Ryan_Lane | because we do things the ubuntu way | 01:07 |
| Ryan_Lane | it's more consistent if we do so | 01:07 |
| alekibango | i cant say i am hater of djb, but his licensing was crazy last time i checked | 01:07 |
| nelson__ | it sucks consistently. | 01:07 |
| Ryan_Lane | alekibango: most of his stuff was released public domain recently | 01:07 |
| Xenith | alekibango: A while back he just threw everything of his into public domain | 01:07 |
| nelson__ | no, seriously, init.d is more than twenty years old, and it started sucking when it was first created. | 01:07 |
| Ryan_Lane | like in the last few years | 01:07 |
| alekibango | thats great news, really | 01:07 |
| Ryan_Lane | nelson__: meh. works for me | 01:08 |
| nelson__ | sucks for me. | 01:08 |
| alekibango | i must say i liked few features of daemontools | 01:08 |
| nelson__ | Look, I'm not suggesting this for no reason. | 01:08 |
| nelson__ | I'm suggesting it because swift-init is broken. | 01:08 |
| Ryan_Lane | I like the init system of solaris and os x personally | 01:08 |
| nelson__ | 1) it has no documentation. | 01:08 |
| nelson__ | 2) it discards error messages. | 01:08 |
| nelson__ | 3) it has no internal help. | 01:08 |
| nelson__ | 4) it doesn't report when a program it started exits. | 01:09 |
| nelson__ | there's a GAZILLION reasons to prefer daemontools to swift-init in particular, and init.d in general. | 01:09 |
| Ryan_Lane | all of those should simply be fixed. | 01:09 |
| Ryan_Lane | no distro uses daemontools by default | 01:09 |
| nelson__ | init.d can't be fixed. It's broken by design. | 01:09 |
| nelson__ | well that's FAIL, then, isn't it? | 01:10 |
| Ryan_Lane | it would be silly to add that requirement by default | 01:10 |
| nelson__ | it's silly to FAIL when success is within your grasp. | 01:10 |
| Ryan_Lane | nelson__: that's something to take up with fedora, ubuntu, debian, rhel, suse, and all other major distros :) | 01:10 |
| nelson__ | "hey, let's do this good thing." "No, we prefer to fail." | 01:10 |
| nelson__ | Ryan_Lane: the reason they don't use it by default is because of conservative sysadmins who want everything done the same way it was done twenty years ago. | 01:11 |
| Ryan_Lane | I'd love a system like launchd in linux | 01:11 |
| nelson__ | yeah, I hate it when ifconfig eth0:0 goes away, but y'know "ip address" makes a lot more sense. | 01:12 |
| Ryan_Lane | either way, making things work properly in init.d is likely best | 01:12 |
| nelson__ | except that init.d doesn't actually solve the problem of starting long-running problems. | 01:12 |
| nelson__ | Why do you insist on using something that doesn't actually work? | 01:13 |
| nelson__ | ... just because everybody else is using it? | 01:13 |
| Ryan_Lane | nelson__: they do work for me though | 01:13 |
| termie | nelson__: we don't insist on using anythign at this point, and i'm perfectly willing to accept that swift-init is doing somethign wrong | 01:13 |
| termie | nelson__: but the solution is to make our code compatible with whichever methods people choose to run it with | 01:13 |
| Ryan_Lane | termie: +1 | 01:14 |
| nelson__ | Ryan_Lane: okay .... where does the stdout of, say, rsyncd go? | 01:14 |
| termie | nelson__: within the realm of reason (i'd say supporting daemontools, init.d and upstart is a good start) | 01:14 |
| Ryan_Lane | nelson__: if the daemon is written properly, whereever that daemon says it goes. hopefully to the system logger | 01:14 |
| nelson__ | or, more to the point, swift-auth-server ? | 01:14 |
| termie | nelson__: but none of those are exclusive or really very complex to support | 01:14 |
| termie | i'd love to stop the argument over which system is better because it is largely a waste of our time | 01:15 |
| nelson__ | Ryan_Lane: when you use daemontools, the output ALWAYS goes into the .../log/main directory. | 01:15 |
| *** MikeZ1 has joined #openstack | 01:15 | |
| Ryan_Lane | nelson__: I don't believe that is correct though | 01:15 |
| nelson__ | termie: because ... you refuse to accept that something else is better. | 01:15 |
| Ryan_Lane | termie: agreed | 01:15 |
| termie | as simply making our services act normally will automatically support them all | 01:15 |
| Ryan_Lane | exactly | 01:15 |
| termie | nelson__: no, because i am not going to force somebody else to agree with me | 01:15 |
| alekibango | support all methods -- thats good for me | 01:15 |
| alekibango | provide patches | 01:15 |
| nelson__ | termie: and since you're agreed that you want to use FAIL, then we'll continue to have to deal with FAIL. | 01:15 |
| termie | nelson__: at this point you are just trolling | 01:16 |
| nelson__ | sigh. | 01:16 |
| nelson__ | nahhhh, I'm frustrated that you're not being rational. | 01:16 |
| nelson__ | "We're using init.d because everybody else is using it." | 01:16 |
| Ryan_Lane | nelson__: meh. if they write logs to the system logger, or add an option for where logs go, daemontools is unneccessary for that | 01:16 |
| nelson__ | That's monkey-talk. | 01:16 |
| nelson__ | more monkey-talk. | 01:16 |
| termie | Ryan_Lane: if you could stop talking to him it would make my /ignore feel a lot more successful ;) | 01:16 |
| nelson__ | I have work to do. | 01:16 |
| alekibango | easy ppl... | 01:16 |
| Ryan_Lane | termie: :) | 01:17 |
| *** MikeZ has quit IRC | 01:17 | |
| Jordandev | fast question: any plan for Bexar to have a Dashboard? (like nebula dashboard?) or another GUI? | 01:17 |
| Ryan_Lane | Jordandev: there are a few being worked on right now | 01:18 |
| anotherjesse | Jordandev: we're just finishing open sourcing the nebula django dashboard as django apps | 01:18 |
| Ryan_Lane | Jordandev: there's one that was supposed to be released by the end of last year (did that happen?) written in django | 01:18 |
| alekibango | there is no need to be harsh on others for having relations with some technology or the other... those are only tools -- some work for some people, some work for another... | 01:18 |
| Jordandev | coool | 01:18 |
| alekibango | Ryan_Lane: it didnt happen yet | 01:18 |
| Ryan_Lane | alekibango: k | 01:19 |
| Jordandev | and when it will? | 01:19 |
| alekibango | really soon now (tm) | 01:19 |
| Ryan_Lane | Jordandev: I'm also writing one, but it isn't likely to be as useful as the django for non wikimedia foundation folks at first | 01:19 |
| anotherjesse | Jordandev / alekibango the guy leading it is back from vacation | 01:19 |
| *** rlucio has quit IRC | 01:19 | |
| alekibango | anotherjesse: ping me if you will want some testing of it please | 01:19 |
| alekibango | i am back from vacation too, pretty ruined my body there, need some rest... | 01:20 |
| alekibango | heh | 01:20 |
| Jordandev | still in vacation but ;) | 01:20 |
| anotherjesse | alekibango: we still need to re-work the auth for it | 01:20 |
| nelson__ | alekibango: I'm just frustrated that I'm being asked to provide a quality solution, and my technical choices are vetoed left and right. | 01:20 |
| anotherjesse | but then again we need to move it to the openstack api :) | 01:20 |
| alekibango | darn... some version to play with would be sweet... iam really curious | 01:21 |
| alekibango | nelson__: i know what you mean | 01:21 |
| Jordandev | Ryan_Lane: we can test it and contribute | 01:21 |
| Ryan_Lane | I'd love that | 01:22 |
| Jordandev | any public link? | 01:22 |
| Ryan_Lane | yep | 01:22 |
| Ryan_Lane | It's for a specific project right now | 01:22 |
| Ryan_Lane | http://ryandlane.com/blog/2011/01/02/building-a-test-and-development-infrastructure-using-openstack/ | 01:22 |
| alekibango | nelson__: i often talk to people about things they cant accept... or imagine... and its tough job | 01:22 |
| Ryan_Lane | which is why it won't be as useful for non-wmf people, as it does some pretty specific things | 01:23 |
| Ryan_Lane | like add puppet configuration and dns information when instances are created | 01:23 |
| alekibango | nelson__: provide patches for daemontools support -- they might get included and i might use them on my servers | 01:23 |
| Ryan_Lane | and creates the users with posix information | 01:23 |
| alekibango | you know, i was thinking about doing this for runit | 01:24 |
| Ryan_Lane | Jordandev: test interface is here: http://nova-controller.tesla.usability.wikimedia.org/sandbox.1/Main_Page | 01:24 |
| alekibango | (which is similar to daemontools in a way) | 01:24 |
| Ryan_Lane | that's where I'm working on features right now anyway | 01:24 |
| Ryan_Lane | I guess I should add information about how to actually use it :D | 01:25 |
| Jordandev | Ryan_Lane: still an empty page! | 01:25 |
| alekibango | Ryan_Lane: i see empty page too | 01:25 |
| nelson__ | alekibango: I'm not generally seeking to improve swift. I'm seeking to improve it for Wikimedia use. But if Ryan_Lane says that he's not going to use it, there's no point in writing it. | 01:26 |
| alekibango | ah, ic | 01:26 |
| Ryan_Lane | try now | 01:27 |
| Ryan_Lane | nelson__: using daemontools for this when we are using init.d for everything else adds complexity, which makes it harder for us to manage things with a small number of admins | 01:28 |
| alekibango | Ryan_Lane: shellaccountnamehelp ? | 01:28 |
| Ryan_Lane | yeah | 01:28 |
| Ryan_Lane | I haven't started doing localization yet | 01:28 |
| alekibango | thats username? | 01:28 |
| nelson__ | except that daemontools works better. | 01:28 |
| Ryan_Lane | yeah. | 01:28 |
| alekibango | nelson__: i loven initng | 01:29 |
| alekibango | loved* | 01:29 |
| Ryan_Lane | wiki user name and shell account name are separate. per request by a ton of people :) | 01:29 |
| alekibango | never liked upstart | 01:29 |
| Ryan_Lane | I haven't started doing localization because we have an army of volunteers, and until I finalize the messages, I'm wasting their time | 01:29 |
| anotherjesse | Ryan_Lane: any other articles about mediawiki + openstack? | 01:30 |
| Ryan_Lane | anotherjesse: just that one right now | 01:30 |
| *** johnpur has quit IRC | 01:30 | |
| Ryan_Lane | I do have a spec sheet here: http://www.mediawiki.org/wiki/Extension:OpenStackManager | 01:30 |
| Ryan_Lane | plus documentation that needs to be updated | 01:31 |
| alekibango | key is what type of key? ssh public key? | 01:32 |
| *** reldan has joined #openstack | 01:32 | |
| Ryan_Lane | yep | 01:32 |
| Ryan_Lane | this is something I need to fix | 01:33 |
| Ryan_Lane | I'm inserting keys directly into ldap right now | 01:33 |
| Ryan_Lane | nova should be doing it instead | 01:33 |
| Ryan_Lane | I need to add that part of the api, and also figure out a way to reliably import the keys | 01:34 |
| Ryan_Lane | and handle key names | 01:34 |
| Ryan_Lane | without adding schema :) | 01:34 |
| Ryan_Lane | alekibango: added you to a project so that you can check out instance creation/modifcation | 01:35 |
| alekibango | thanks :) i added my own project atm | 01:35 |
| Ryan_Lane | ami-gr0xqtsz is a bad image | 01:35 |
| Ryan_Lane | heh | 01:35 |
| Ryan_Lane | I haven't limited access to project creation yet :) | 01:35 |
| Ryan_Lane | and though you can create instances, you can't ssh to them, as they are in a backend-network | 01:36 |
| Ryan_Lane | for security purposes | 01:36 |
| alekibango | you are using nagios? | 01:36 |
| alekibango | try zabbix :) | 01:36 |
| nelson__ | alekibango: swift-auth-server (at least) runs just fine under daemontools. | 01:36 |
| Ryan_Lane | alekibango: I | 01:37 |
| Ryan_Lane | err | 01:37 |
| Ryan_Lane | alekibango: I'll take a look at it | 01:37 |
| alekibango | do, its cool | 01:37 |
| alekibango | zabbix.com | 01:37 |
| Ryan_Lane | we have very few people, and have nagios integrated with puppet, so it's unlikely until someone gets some free time :) | 01:38 |
| alekibango | it takes at least a day to make it right if its complex, but it will be worth it, believe me | 01:38 |
| Ryan_Lane | have I mentioned we like volunteers? :) | 01:38 |
| nelson__ | alekibango: and unlike swift-init, the error message is right there in the daemontool log file. | 01:38 |
| alekibango | nelson__: i just love the simplicity | 01:38 |
| *** anotherjesse has quit IRC | 01:38 | |
| nelson__ | alekibango: I just love that it works. | 01:39 |
| alekibango | yes, and its simple | 01:39 |
| alekibango | simple => no magic (=no problems) | 01:39 |
| nelson__ | alekibango: I know where to find the log output; I know how to start and stop processes; I know how to send signals to the process.; I know how to tell the process to go down and stay down. | 01:39 |
| alekibango | and especially i love that it restarts itself when it fails | 01:40 |
| alekibango | unlike some init scripts | 01:40 |
| alekibango | (or should i say most) | 01:40 |
| nelson__ | alekibango: init scripts aren't running and can't see that a process has exited. | 01:41 |
| alekibango | yes i know | 01:41 |
| alekibango | some of them run some watching daemon which watches the actual service | 01:41 |
| alekibango | which is crazy :) | 01:41 |
| nelson__ | but other people haven't felt that pain, and don't appreciate self-restarting services. | 01:41 |
| alekibango | but the only way to make it work, i hear you nelson | 01:42 |
| alekibango | for openstack cluster, it might make sense to use daemontools - and even depend on them | 01:43 |
| alekibango | as the Nova or SWIFT will be the only service the machine will serve | 01:43 |
| alekibango | (mostly) | 01:43 |
| Ryan_Lane | if services are dying, it for a reason | 01:43 |
| Ryan_Lane | err it's | 01:43 |
| alekibango | Ryan_Lane: yes, but with initscript, you sometimes cant find out why it dies | 01:44 |
| alekibango | or why it is not running | 01:44 |
| Ryan_Lane | how's daemontools help with that? | 01:44 |
| alekibango | a lot :) | 01:44 |
| alekibango | thats where the simplicity comes in | 01:44 |
| Ryan_Lane | don't get me wrong. init.d sucks. | 01:45 |
| alekibango | it captures output of the process into file -> its not lost somewhere in system logs (guess which one) or in buggy init script | 01:45 |
| Ryan_Lane | forcing people to use something non-standard for their distro just isn't a good idea | 01:45 |
| alekibango | (guess where) | 01:45 |
| alekibango | and why | 01:46 |
| Ryan_Lane | heh | 01:46 |
| Ryan_Lane | but... | 01:46 |
| alekibango | am not for forcing | 01:46 |
| Ryan_Lane | if you are writing to the system logger, the admin can choose where to write that file | 01:46 |
| alekibango | but having this possibility might make sense | 01:46 |
| alekibango | Ryan_Lane: yes, but mostly they do not | 01:46 |
| alekibango | and i even lost my nova logs few times | 01:46 |
| alekibango | thanks to some bugs | 01:46 |
| alekibango | they always bite | 01:47 |
| Ryan_Lane | that's a problem fixed by documentation | 01:47 |
| alekibango | Ryan_Lane: not really | 01:47 |
| Ryan_Lane | ah. you mean bugs causing the logs to go to the wrong spot | 01:48 |
| alekibango | right! | 01:48 |
| alekibango | thats the simplicity i am talking about | 01:48 |
| alekibango | complex ==> will be broken | 01:48 |
| alekibango | and init scripts are complex :) | 01:49 |
| Ryan_Lane | inconsistency with a distro is also complexity though ;) | 01:49 |
| alekibango | on my debian, i need to restart nova after first boot | 01:49 |
| alekibango | thanks to not using daemontools :) | 01:49 |
| Ryan_Lane | why? | 01:49 |
| Ryan_Lane | sounds like a bug to fix | 01:49 |
| alekibango | all nova services are trying to create tables at the same moment -- in the empty database | 01:50 |
| alekibango | some of them fail | 01:50 |
| alekibango | and abort | 01:50 |
| alekibango | and whatever | 01:50 |
| Ryan_Lane | that's not a problem in a production system though | 01:51 |
| alekibango | it is | 01:51 |
| alekibango | i need it to be deployed automatically | 01:51 |
| Ryan_Lane | why would the database be empty? | 01:51 |
| alekibango | without glitch | 01:51 |
| alekibango | i install nova in 3 minutes on 3 servers | 01:51 |
| * Ryan_Lane nods | 01:51 | |
| alekibango | including partitioning and raid and stuff | 01:51 |
| alekibango | and i want it to RUN sucessfully immediatelly | 01:52 |
| alekibango | seconds are important for me | 01:52 |
| alekibango | you know, for fast deployment, automated functional testing, etc etc | 01:52 |
| Ryan_Lane | yep | 01:52 |
| alekibango | and after first start, few services just fail somehow :) | 01:53 |
| alekibango | using daemontools with restart might help there | 01:53 |
| Ryan_Lane | fixing the bug would be better :) | 01:53 |
| alekibango | well, might be, but then, another 5 bugs come in | 01:53 |
| alekibango | :) | 01:54 |
| Ryan_Lane | heh | 01:54 |
| alekibango | Ryan_Lane: its not a bug, its feature... maybe | 01:54 |
| alekibango | its feature of SQLAlchemy | 01:54 |
| termie | alekibango: certainly would, as would other approaches, the bug should also be fixed to handle the error case in popular setups as well, as well as any bugs related to it running under less popular but still supported setups | 01:54 |
| Ryan_Lane | lol | 01:54 |
| alekibango | its creating tables and stuff -which is missing in db | 01:54 |
| alekibango | wrong part is, they all are trying to do it at the same moment | 01:55 |
| alekibango | :) | 01:55 |
| alekibango | all those nova services | 01:55 |
| alekibango | maybe we should prepare the db as part of postinstall | 01:55 |
| alekibango | that would make it work better | 01:55 |
| Ryan_Lane | that assumes the database is configured at install time | 01:56 |
| alekibango | ... even installation manual imho mentions restarts | 01:56 |
| alekibango | my database is configured | 01:56 |
| alekibango | but empty | 01:56 |
| termie | alekibango: a tools/prep_db.py would certainly allow such a thing to be written | 01:56 |
| Xenith | I've switched to icinga, wasn't too impressed with zabbix when my old company used it. | 01:56 |
| * Ryan_Lane doesn't like when services auto-start themselves | 01:56 | |
| Ryan_Lane | on install, that is | 01:56 |
| alekibango | Xenith: why? | 01:57 |
| Xenith | alekibango: Personally, I didn't work with it, but my coworkers who were trying to configure it were not happy at all. Always complaining :) | 01:57 |
| alekibango | heh, as i said, its not very easy when you are not doing large deployment | 01:58 |
| *** joearnold has quit IRC | 01:58 | |
| Xenith | Also, it wouldn't shut up. I got tons of SMS messages from it, but that was probably due to the previously mentioned coworkers not being able to configuring it properly :) | 01:58 |
| alekibango | i agree its not easy to configure :) but still -- just look on screenshots :) | 01:58 |
| Xenith | Well, the new icinga-web stuff looks pretty snazzy as well :) | 01:59 |
| Ryan_Lane | yeah, i'm looking at the demo for that now | 01:59 |
| Ryan_Lane | it's quite pretty | 02:00 |
| *** mdomsch has joined #openstack | 02:00 | |
| alekibango | it certainly is... i remember using nagios some 15 years ago and it was crazy | 02:00 |
| Ryan_Lane | nagios has a terrible UI | 02:00 |
| alekibango | thansk, never heard of icinga, will check it later | 02:01 |
| Xenith | That's because the Nagios UI is(was) hard-coded in the C/CGI stuff. | 02:01 |
| alekibango | yes, iremember that | 02:01 |
| *** littleidea_ has quit IRC | 02:01 | |
| Xenith | alekibango: icinga is a fork of nagios, so it can use all the nagios plugins and stuff | 02:01 |
| alekibango | my hairs are standing when i remember looking into source | 02:01 |
| alekibango | something like PHP, but in C (ouch) | 02:02 |
| alekibango | or perl? | 02:02 |
| Ryan_Lane | heh. one of our devs did a security audit of nagios | 02:03 |
| Ryan_Lane | he was less than impressed | 02:03 |
| alekibango | thats the spirit, yes | 02:03 |
| zaitcev | Did he find a useable exploit though? | 02:03 |
| Xenith | That's why you don't run nagios over public internet. | 02:04 |
| Ryan_Lane | he sent a list of bugs to them | 02:04 |
| Xenith | Backplane or VPNs :) | 02:04 |
| Ryan_Lane | Xenith: wuss :) http://nagios.wikimedia.org/ | 02:04 |
| *** reldan has quit IRC | 02:04 | |
| Xenith | Well, I have the web front end available on a public IP, of course. :) | 02:05 |
| Ryan_Lane | ah. right. the daemon itself isn't publicly accessible | 02:05 |
| alekibango | tomorrow in news: wikimedia hacked by unanonymouse | 02:05 |
| Ryan_Lane | :D | 02:05 |
| *** littleidea has joined #openstack | 02:06 | |
| alekibango | oh that nagios is really crazy ui :) | 02:10 |
| Ryan_Lane | I pretty much only use Services: unhandled and Hosts: unhandled | 02:12 |
| *** daleolds has quit IRC | 02:23 | |
| *** schisamo has quit IRC | 02:28 | |
| *** jdurgin has joined #openstack | 02:30 | |
| *** jimbaker has joined #openstack | 02:33 | |
| *** anotherjesse has joined #openstack | 02:40 | |
| Ryan_Lane | when upgrading nova, is it automatically supposed to upgrade the database schema too? | 02:51 |
| Ryan_Lane | or is there some other way to go about that? | 02:51 |
| anotherjesse | Ryan_Lane: the idea is that once we migrate the schema we should move from the auto-generate database to a migration based | 02:52 |
| Ryan_Lane | so if all of my nova services are throwing schema errors right now, how do I fix it? :) | 02:53 |
| anotherjesse | yikes - hmm, perhaps then something got merged that changed schema :( | 02:54 |
| anotherjesse | did you just upgrade? | 02:54 |
| Ryan_Lane | "Unknown column 'instances.availability_zone' in 'field list'" | 02:54 |
| Ryan_Lane | yeah. using trunk debs | 02:54 |
| Ryan_Lane | trunk ppa that is | 02:54 |
| Ryan_Lane | if I need to wipe out the database I won't be amazingly upset, but I'd like to know this won't happen when I upgrade in production. heh | 02:55 |
| Ryan_Lane | though I won't be using trunk ppa in production | 02:55 |
| anotherjesse | ya - let me do the same thing to my test cluster - haven't upgraded it in a couple days | 02:56 |
| *** GeorgeSebastian has joined #openstack | 02:56 | |
| Ryan_Lane | I wanted raw image support ^_^ | 02:56 |
| anotherjesse | me to | 02:57 |
| Xenith | I'm looking forward to that. | 02:58 |
| Xenith | Haven't had a chance to play with it yet. | 02:58 |
| *** rchavik has joined #openstack | 02:59 | |
| creiht | nelson__: the error stuff on swift-init is a known bug | 03:06 |
| creiht | nelson__: re: daemontools, we leave those decisions up to the ops guys, and they don't like it. That said, I'm not against having a swift-init, or like script that cooperates better with daemontools. It just isn't what we use right now | 03:07 |
| creiht | nelson__: and for future debugging, you can see the output by running the auth server manually like: | 03:14 |
| creiht | swift-auth-server /etc/swift/auth.conf | 03:14 |
| creiht | will display the console error | 03:14 |
| creiht | I know it is a pain, but it is on our list to fix | 03:14 |
| creiht | swift-init can definately be made better, and of course patches are welcome :) | 03:15 |
| creiht | my personal experience with daemontools in the past is that it was a pain to setup/configure, but once you got it set up it worked reasonably well | 03:16 |
| creiht | I think most people dislike it, just because it is different | 03:16 |
| *** blpiatt has joined #openstack | 03:17 | |
| creiht | All of that said, ops people are all going to want to control their services differently, so I'm much more inclined to keep it flexible, rather than say this is how you have to do it (eg. daemontools) | 03:17 |
| alekibango | +1 | 03:17 |
| creiht | swift-init is just there as a piece of legacy code since it was what grew somewhat organically to suite our needs | 03:18 |
| creiht | and the requests of our ops guys | 03:18 |
| creiht | nelson__, Ryan_Lane: Please let me know if you guys run into any issues with swift | 03:20 |
| anotherjesse | alekibango: my system was up-to-date with that change already :( | 03:20 |
| *** dendro-afk is now known as dendrobates | 03:20 | |
| Ryan_Lane | creiht: I'm not too involved with swift as of right now. nelson__ and a couple other people are more involved | 03:21 |
| Ryan_Lane | I'll let you know though, as I'm sure I'll be knee deep in it at some point in time :D | 03:21 |
| creiht | hehe | 03:22 |
| Ryan_Lane | the plus side of having few people? you get to work with a lot of stuff? the negative side? you have to work with a lot of stuff | 03:22 |
| creiht | We know that there are still probably some rough areas, and areas where the docs need improving | 03:22 |
| Ryan_Lane | minus a question mark somewhere in there. | 03:23 |
| creiht | Ryan_Lane: indeed | 03:23 |
| *** littleidea has quit IRC | 03:27 | |
| *** hadrian has quit IRC | 03:31 | |
| *** hadrian has joined #openstack | 03:32 | |
| *** littleidea has joined #openstack | 03:32 | |
| anotherjesse | creiht: was their any progress on swift-lite for nova? | 03:37 |
| creiht | anotherjesse: I think gholt played with the idea for a bit, but then it got pushed to the side as more pressing stuff came around | 03:38 |
| *** Ryan_Lane is now known as Ryan_Lane|away | 03:38 | |
| *** guynaor has joined #openstack | 03:39 | |
| anotherjesse | creiht: termie spent some time moving all the components of nova to eventlet except the naive nova-objectstore | 03:39 |
| *** guynaor has left #openstack | 03:39 | |
| anotherjesse | creiht: it would be nice to be able to kill nova-objectstore and have a swift-lite | 03:39 |
| *** gondoi_ has joined #openstack | 03:39 | |
| creiht | anotherjesse: k, I'll bring it up again with gholt, but I know that we are all going to be quite busy for a bit | 03:39 |
| anotherjesse | creiht: nah, it's a new year, there is time ;) | 03:40 |
| creiht | lol | 03:40 |
| anotherjesse | swift-lite would only be for dev mode btw | 03:42 |
| creiht | anotherjesse: of course :) | 03:43 |
| *** gondoi_ has quit IRC | 03:44 | |
| creiht | We've also talked how it would be good for anyone who is developing client side apps for swift to test against | 03:44 |
| anotherjesse | hmm, but won't that cost rax money ... | 03:45 |
| anotherjesse | they could make $0.43 in the testing process | 03:45 |
| creiht | anotherjesse: the approach we talked about was to have a package that could be installed that sets everything up for a 1-replica "cluster" on one machine | 03:46 |
| creiht | replication, and aux services wouldn't have to run | 03:46 |
| creiht | In theory it is mostly packaging work | 03:46 |
| aimon | hi all | 03:50 |
| aimon | creiht: A KickStart iso for that would be pretty great | 03:51 |
| aimon | we are doing similar for our product .. it allows non-HA one resource node configuration | 03:51 |
| aimon | gives people a chance to test it out with out ivolving a bunch of support from us | 03:52 |
| aimon | .. I couldnt help but to jump in.. This thread was exaclty like one we just had internally | 03:52 |
| creiht | aimon: there has actually been talk of such a thing. A live ISO for nova already exists | 03:53 |
| creiht | Not sure if there has been any work on a swift one | 03:53 |
| aimon | would be a great accompnyment.. | 03:55 |
| *** jimbaker has quit IRC | 03:55 | |
| aimon | a few iso installs and you are up and runnng | 03:55 |
| aimon | reduces time to implementation for testing considerably | 03:55 |
| aimon | of courser as you said.. it is mostly a packaging job | 03:55 |
| aimon | we are doing same atm.. TRPMing everything in our case | 03:56 |
| aimon | *RPMing | 03:56 |
| *** trin_cz has quit IRC | 03:56 | |
| aimon | then can easily uise KS file for centos to instal and configure | 03:56 |
| aimon | .. well we have a bit of puppet action also.. but our situation is more complex | 03:57 |
| aimon | we have puplic puppet server for KS boot.. it configures itself | 03:57 |
| aimon | creiht: Who are you btw? Were you at the design summit in Texas? | 03:58 |
| aimon | perhaps we met | 03:58 |
| creiht | <- Chuck from the swift team | 03:58 |
| creiht | and yes I have been to all the design summits | 03:59 |
| anotherjesse | creiht yells at other rackers ... | 03:59 |
| *** hadrian has quit IRC | 03:59 | |
| creiht | lol | 03:59 |
| aimon | hehe | 03:59 |
| aimon | I m with Morph Labs | 03:59 |
| creiht | I wouldn't exactly call it yelling :) | 03:59 |
| anotherjesse | you yell because you care | 04:00 |
| aimon | I am still amazed by that summit :) | 04:00 |
| gholt | Badgering? | 04:00 |
| aimon | strongly engouraging <-- ftw | 04:00 |
| aimon | *encouraging | 04:00 |
| notmyname | "creiht is a very aggressive person. He/She attacked others 9 times." from http://12.am/openstack/ | 04:00 |
| gholt | http://www.badgerbadgerbadger.com/ | 04:00 |
| aimon | omg lol | 04:01 |
| creiht | hahah | 04:01 |
| aimon | btw does B. Piatt stalk this forum? | 04:01 |
| aimon | *channel | 04:01 |
| gholt | Probably.... | 04:02 |
| creiht | aimon: periodically | 04:02 |
| gholt | blpiatt: ^^ | 04:02 |
| aimon | cool | 04:02 |
| aimon | LOL love the Amanita Muscaria (Fly Agaric) reference in the bdger dance | 04:02 |
| *** jdurgin has quit IRC | 04:09 | |
| *** dendrobates is now known as dendro-afk | 04:16 | |
| *** MikeZ1 has quit IRC | 04:23 | |
| blpiatt | aimon, I'm around sometimes :) | 04:30 |
| blpiatt | creiht, I'm working on a testbed for openstack so we can have an environment to test API clients, UIs, and run full end to end system tests with performance metrics | 04:33 |
| blpiatt | we'll get that setup rather than a Rackspace Cloud testbed as the product offerings won't be against today's trunk and we'll want to be testing against those when doing new development | 04:34 |
| creiht | blpiatt: cool | 04:35 |
| creiht | I have hardware in our lab now to do a reference swift setup | 04:35 |
| creiht | just don't have time now to mess with it | 04:35 |
| alekibango | blpiatt, creiht: i have fai-project.org installing openstack system automatically if you are interested... i would love to get automated functional testing of openstack... | 04:35 |
| creiht | alekibango: I think jay pipes is taking the lead on getting the automated functional testing going | 04:36 |
| alekibango | i will talk to him then :) | 04:36 |
| blpiatt | alekibango, looks like good stuff, FAI is in the same bare metal provisioning space as cobbler and barracus? | 04:36 |
| alekibango | similar, but for debian, ubuntu | 04:36 |
| alekibango | (and also can work wth other distros if you provide some images) | 04:37 |
| alekibango | debian and ubuntu works very well, using packages | 04:37 |
| blpiatt | creiht, alekibango: yeah, jay is going to lead the functional testing, emellor wants to help, so does jordanrinke | 04:37 |
| alekibango | so i do | 04:37 |
| blpiatt | ok, so more like barracus than cobbler | 04:38 |
| alekibango | i dont know barracus :) | 04:38 |
| blpiatt | http://baracus-project.org/Site/Baracus.html | 04:38 |
| blpiatt | Novell just recently open sourced it | 04:38 |
| alekibango | interesting.. | 04:40 |
| alekibango | blpiatt: fai is not using images with debian/ubuntu | 04:41 |
| alekibango | it installs packages | 04:41 |
| alekibango | its 'the debian way' | 04:41 |
| alekibango | (ubuntu also works well now) | 04:41 |
| alekibango | images = pain | 04:43 |
| anotherjesse | baracus = perl? | 04:43 |
| anotherjesse | http://gitorious.org/baracus/baracus/blobs/master/root/usr/sbin/bapower | 04:43 |
| aimon | hey blpiatt! | 04:43 |
| alekibango | fai = shell (or whatever) | 04:44 |
| blpiatt | anotherjesse, yeah barracus = perl (mostly) | 04:44 |
| blpiatt | alekibango, with barracus you'd use a base image then preseed to install the packages | 04:45 |
| alekibango | similar for fai + nondeb distro... | 04:46 |
| blpiatt | yeah, sounds like it, I haven't looked at FAI before, I'll do so tomorrow | 04:47 |
| alekibango | its not very easy, but after some fight, it works well for you | 04:47 |
| alekibango | and its highly configurable | 04:47 |
| alekibango | i think we should build and TEST automatically functionality of all versions of nova trunk and branches with merge requests... | 04:50 |
| blpiatt | agreed, I want to come up with a NovaMIPS and SwiftMIPS so we can run a set of performance tests as well to see if our changes are improving speed or not on each merge to trunk | 04:51 |
| alekibango | good idea | 04:51 |
| blpiatt | that way if we have performance tuning to do in the future we can go see the impact of each change made in the past to know where to start looking | 04:51 |
| * alekibango needs to sleep now, see you tomorow... | 04:52 | |
| winston-d | blpiatt: what will NovaMIPS & SwiftMIPS be like? | 04:53 |
| creiht | MIPS? | 04:53 |
| xtoddx | hardware architecture? | 04:54 |
| blpiatt | winston-d, creiht: I use MIPS in reference to BogoMIPS, it'll be some metric we define, it isn't "millions of instructions per second" | 04:54 |
| creiht | ahh | 04:54 |
| winston-d | ok | 04:55 |
| creiht | hehe | 04:55 |
| creiht | well for swift, we already have swift-bench as a start | 04:55 |
| blpiatt | creiht, nice, I think we have a lot of the pieces we need, it'll be good to pull it all together and automate it | 04:56 |
| winston-d | creiht: is swift-bench in swift branch? can I use it? | 04:56 |
| creiht | winston-d: yup | 04:56 |
| winston-d | what does swift-bench actually bench? | 04:56 |
| creiht | swift | 04:56 |
| creiht | :) | 04:56 |
| winston-d | I mean what metrics does it report | 04:57 |
| creiht | reminds me that I need to write some docs for it | 04:57 |
| winston-d | creiht: Write Docs for Swift-Bench, Please | 04:57 |
| creiht | It will do a number of PUTS GETS and DELETES of objects to a swift cluster | 04:57 |
| creiht | doh | 04:58 |
| *** kashyapc has joined #openstack | 04:58 | |
| creiht | looks like some import changed and swift-bench is broken :/ | 04:58 |
| winston-d | I'm looking at swift revno 148 source. where is swift-bench? | 04:59 |
| creiht | winston-d: in bin | 05:00 |
| winston-d | OK. find it. thanks | 05:01 |
| creiht | it is configurable for number of objects and concurrency | 05:01 |
| winston-d | Hmm, nice. I'd like to have a simple test in my swift setup | 05:02 |
| *** MikeZ has joined #openstack | 05:02 | |
| creiht | winston-d: at your rev, it is possible it may still be working | 05:03 |
| winston-d | I've a swift testbed with one proxy, 6 account nodes. any test configuration recommendation for this setup? | 05:05 |
| creiht | start with running swift-bench from one machine that has good connectivity to the proxy | 05:06 |
| winston-d | OK | 05:06 |
| *** jimbaker has joined #openstack | 05:08 | |
| *** Cybo has quit IRC | 05:14 | |
| winston-d | creiht: to use swift-bench, do I need to create user account? | 05:15 |
| creiht | yeah it is probably a good idea to create a user account | 05:17 |
| creiht | for benchmarking | 05:17 |
| winston-d | 2011-01-04 13:46:45,695 INFO swift-bench 5 PUTS [0 failures], 2.3/s | 05:28 |
| winston-d | 2011-01-04 13:47:00,988 INFO swift-bench 48 PUTS [0 failures], 2.7/s | 05:28 |
| winston-d | 2011-01-04 13:47:16,282 INFO swift-bench 97 PUTS [0 failures], 3.0/s | 05:28 |
| winston-d | 2011-01-04 13:47:17,389 INFO swift-bench 100 PUTS **FINAL** [0 failures], 2.9/s | 05:28 |
| winston-d | 2011-01-04 13:47:19,831 INFO swift-bench 30 GETS [0 failures], 12.3/s | 05:28 |
| winston-d | 2011-01-04 13:47:35,318 INFO swift-bench 319 GETS [0 failures], 17.8/s | 05:28 |
| winston-d | 2011-01-04 13:47:50,346 INFO swift-bench 652 GETS [0 failures], 19.8/s | 05:28 |
| winston-d | 2011-01-04 13:48:05,835 INFO swift-bench 937 GETS [0 failures], 19.3/s | 05:28 |
| winston-d | 2011-01-04 13:48:08,591 INFO swift-bench 1000 GETS **FINAL** [0 failures], 19.5/s | 05:28 |
| winston-d | 2011-01-04 13:48:11,011 INFO swift-bench 10 DEL [0 failures], 4.1/s | 05:28 |
| winston-d | 2011-01-04 13:48:26,242 INFO swift-bench 59 DEL [0 failures], 3.3/s | 05:28 |
| winston-d | 2011-01-04 13:48:41,182 INFO swift-bench 100 DEL **FINAL** [0 failures], 3.1/s | 05:28 |
| winston-d | creiht: how does this result look like? | 05:29 |
| creiht | incredibly slow | 05:29 |
| winston-d | or, how should i interpret it? | 05:29 |
| winston-d | :) | 05:30 |
| creiht | winston-d: and next time I would recommend pasting to paste.openstack.org | 05:30 |
| winston-d | i see, thanks. | 05:30 |
| creiht | winston-d: Have you tuned any of the config settings? (like workers) | 05:31 |
| winston-d | no, everything is default | 05:31 |
| creiht | ok... default is only one worker | 05:31 |
| winston-d | oh, I just check 2 workers for accoutn/container/object | 05:32 |
| winston-d | how to tune proper number of workers? some ratio to # account/container/object nodes and replicates? | 05:34 |
| winston-d | maybe also the device per nodes? | 05:35 |
| creiht | winston-d: Would you mind if we worked through this tomorrow? | 05:36 |
| creiht | (It is time for me to get some sleep) | 05:36 |
| winston-d | of course. I know it is late for you. | 05:36 |
| creiht | :) | 05:36 |
| winston-d | thank you for your help, creiht. have a good nite | 05:37 |
| creiht | a decent starting point for good hardware is: | 05:37 |
| *** mdomsch has quit IRC | 05:37 | |
| creiht | proxy workers: 2x number of cores | 05:37 |
| creiht | acct/cont/obj workers: start with 1x number of cores | 05:37 |
| winston-d | good. let me do that. I'll re-do the test and discuss w/ u tomorrow. :) | 05:38 |
| anticw | creiht: what about HT 'cores' ? | 05:41 |
| anticw | creiht: wrt to obj workers ... why base that on cores? | 05:41 |
| anticw | it's not cpu bound, more IO bound | 05:41 |
| anticw | so something like n x spindles sure makes more sense? | 05:41 |
| creiht | anticw: yeah the obj workers are more complicated, but 1x core is just something to get him going :) | 05:42 |
| winston-d | :) | 05:42 |
| *** reldan has joined #openstack | 05:46 | |
| winston-d | anticw: i assume the configuration for rsyncd should be altered too? | 05:48 |
| anticw | no | 05:48 |
| anticw | i think the connections aren't capped | 05:49 |
| anticw | so inbound you'll get a lot but not insane numbers | 05:49 |
| winston-d | alrite | 05:49 |
| *** GeorgeSebastian has quit IRC | 05:50 | |
| *** sassyn has joined #openstack | 06:17 | |
| *** jeevan_ullas has quit IRC | 06:28 | |
| *** jeevan_ullas has joined #openstack | 06:28 | |
| *** sassyn has quit IRC | 06:29 | |
| *** reldan has quit IRC | 06:32 | |
| *** jbaker has joined #openstack | 06:36 | |
| *** jimbaker has quit IRC | 06:36 | |
| *** jbaker is now known as jimbaker | 06:36 | |
| *** ramkrsna has joined #openstack | 06:37 | |
| *** ramkrsna has joined #openstack | 06:37 | |
| *** sandywalsh has quit IRC | 06:43 | |
| *** sandywalsh has joined #openstack | 06:56 | |
| *** guigui1 has joined #openstack | 07:00 | |
| *** anotherjesse has quit IRC | 07:12 | |
| *** MikeZ1 has joined #openstack | 07:26 | |
| *** MikeZ has quit IRC | 07:28 | |
| *** rcc has joined #openstack | 07:38 | |
| *** rcc has joined #openstack | 07:39 | |
| *** ArdRigh has joined #openstack | 07:42 | |
| *** jimbaker has quit IRC | 07:50 | |
| *** allsystemsarego has joined #openstack | 07:54 | |
| *** allsystemsarego has joined #openstack | 07:54 | |
| *** brd_from_italy has joined #openstack | 07:58 | |
| *** MikeZ has joined #openstack | 08:01 | |
| *** gasbakid has joined #openstack | 08:02 | |
| *** MikeZ1 has quit IRC | 08:03 | |
| *** blpiatt has quit IRC | 08:09 | |
| *** fabiand_ has joined #openstack | 08:17 | |
| *** trin_cz has joined #openstack | 08:20 | |
| *** befreax has joined #openstack | 08:23 | |
| *** reldan has joined #openstack | 08:26 | |
| *** littleidea has quit IRC | 08:29 | |
| *** fabiand_ has quit IRC | 08:33 | |
| *** calavera has joined #openstack | 08:35 | |
| soren | vishy: My concern isn't just about euca2ools. It's about everything that consumes the EC2 API. | 08:40 |
| soren | vishy: I'm confident that there are many tools out there that expect instance ID's to not be 39 characters long. | 08:41 |
| *** calavera has quit IRC | 08:44 | |
| *** allsystemsarego has quit IRC | 08:44 | |
| *** kashyapc has quit IRC | 08:44 | |
| *** jtimberman has quit IRC | 08:44 | |
| *** perestrelka has quit IRC | 08:44 | |
| *** lool has quit IRC | 08:44 | |
| *** JordanRinke has quit IRC | 08:44 | |
| *** brainproxy has quit IRC | 08:44 | |
| *** _cerberus_ has quit IRC | 08:44 | |
| *** klumpie has quit IRC | 08:44 | |
| *** cory__ has quit IRC | 08:44 | |
| *** fitzdsl has quit IRC | 08:44 | |
| *** smoser` has quit IRC | 08:44 | |
| *** Dweezahr has quit IRC | 08:44 | |
| *** letterj_ has quit IRC | 08:44 | |
| *** devcamcar has quit IRC | 08:44 | |
| *** sunech has quit IRC | 08:44 | |
| *** deshantm has quit IRC | 08:44 | |
| *** Kdecherf has quit IRC | 08:44 | |
| *** tmarble has quit IRC | 08:44 | |
| *** deshantm_cosi has quit IRC | 08:44 | |
| *** annegentle has quit IRC | 08:44 | |
| *** cloud0_ has quit IRC | 08:44 | |
| *** ioso has quit IRC | 08:44 | |
| *** brd_from_italy has quit IRC | 08:44 | |
| *** gondoi has quit IRC | 08:44 | |
| *** karmabot has quit IRC | 08:44 | |
| *** _0x44 has quit IRC | 08:44 | |
| *** filler has quit IRC | 08:44 | |
| *** sophiap has quit IRC | 08:44 | |
| *** ksteward has quit IRC | 08:44 | |
| *** jeremyb has quit IRC | 08:44 | |
| *** MikeZ has quit IRC | 08:44 | |
| *** jbryce has quit IRC | 08:44 | |
| *** arcane has quit IRC | 08:44 | |
| *** cw has quit IRC | 08:44 | |
| *** clayg has quit IRC | 08:44 | |
| *** yosh has quit IRC | 08:44 | |
| *** reldan has quit IRC | 08:44 | |
| *** gholt has quit IRC | 08:44 | |
| *** Daviey has quit IRC | 08:44 | |
| *** zul has quit IRC | 08:44 | |
| *** dabo has quit IRC | 08:44 | |
| *** soosfarm_ has quit IRC | 08:44 | |
| *** mtaylor has quit IRC | 08:44 | |
| *** herki_ has quit IRC | 08:44 | |
| *** olivier_ has quit IRC | 08:44 | |
| *** styro has quit IRC | 08:44 | |
| *** mjmac has quit IRC | 08:44 | |
| *** termie has quit IRC | 08:44 | |
| *** michaeldreamhost has quit IRC | 08:44 | |
| *** zykes- has quit IRC | 08:44 | |
| *** creiht has quit IRC | 08:44 | |
| *** pandemicsyn has quit IRC | 08:44 | |
| *** Adri2000 has quit IRC | 08:44 | |
| *** patri0t has quit IRC | 08:44 | |
| *** xtoddx has quit IRC | 08:44 | |
| *** ChrisAM has quit IRC | 08:44 | |
| *** morfeas has quit IRC | 08:44 | |
| soren | vishy: So it's also not just about usability. | 08:44 |
| *** brd_from_italy has joined #openstack | 08:47 | |
| *** filler has joined #openstack | 08:47 | |
| *** sophiap has joined #openstack | 08:47 | |
| *** gondoi has joined #openstack | 08:47 | |
| *** karmabot has joined #openstack | 08:47 | |
| *** _0x44 has joined #openstack | 08:47 | |
| *** ioso has joined #openstack | 08:47 | |
| *** cloud0_ has joined #openstack | 08:47 | |
| *** annegentle has joined #openstack | 08:47 | |
| *** deshantm_cosi has joined #openstack | 08:47 | |
| *** tmarble has joined #openstack | 08:47 | |
| *** Kdecherf has joined #openstack | 08:47 | |
| *** deshantm has joined #openstack | 08:47 | |
| *** ksteward has joined #openstack | 08:47 | |
| *** jeremyb has joined #openstack | 08:47 | |
| *** zelazny.freenode.net sets mode: +v _0x44 | 08:47 | |
| *** MikeZ has joined #openstack | 08:47 | |
| *** jbryce has joined #openstack | 08:47 | |
| *** arcane has joined #openstack | 08:47 | |
| *** cw has joined #openstack | 08:47 | |
| *** clayg has joined #openstack | 08:47 | |
| *** yosh has joined #openstack | 08:47 | |
| *** reldan has joined #openstack | 08:48 | |
| *** gholt has joined #openstack | 08:48 | |
| *** Daviey has joined #openstack | 08:48 | |
| *** zul has joined #openstack | 08:48 | |
| *** dabo has joined #openstack | 08:48 | |
| *** soosfarm_ has joined #openstack | 08:48 | |
| *** mtaylor has joined #openstack | 08:48 | |
| *** herki_ has joined #openstack | 08:48 | |
| *** olivier_ has joined #openstack | 08:48 | |
| *** styro has joined #openstack | 08:48 | |
| *** mjmac has joined #openstack | 08:48 | |
| *** termie has joined #openstack | 08:48 | |
| *** michaeldreamhost has joined #openstack | 08:48 | |
| *** Adri2000 has joined #openstack | 08:48 | |
| *** zykes- has joined #openstack | 08:48 | |
| *** creiht has joined #openstack | 08:48 | |
| *** pandemicsyn has joined #openstack | 08:48 | |
| *** patri0t has joined #openstack | 08:48 | |
| *** xtoddx has joined #openstack | 08:48 | |
| *** ChrisAM has joined #openstack | 08:48 | |
| *** morfeas has joined #openstack | 08:48 | |
| *** zelazny.freenode.net sets mode: +vvv mtaylor creiht pandemicsyn | 08:48 | |
| *** localhost has quit IRC | 08:48 | |
| *** mtaylor has quit IRC | 08:48 | |
| *** mtaylor has joined #openstack | 08:48 | |
| *** kashyapc has joined #openstack | 08:48 | |
| *** calavera has joined #openstack | 08:48 | |
| *** allsystemsarego has joined #openstack | 08:48 | |
| *** jtimberman has joined #openstack | 08:48 | |
| *** perestrelka has joined #openstack | 08:48 | |
| *** lool has joined #openstack | 08:48 | |
| *** JordanRinke has joined #openstack | 08:48 | |
| *** Dweezahr has joined #openstack | 08:48 | |
| *** brainproxy has joined #openstack | 08:48 | |
| *** _cerberus_ has joined #openstack | 08:48 | |
| *** klumpie has joined #openstack | 08:48 | |
| *** cory__ has joined #openstack | 08:48 | |
| *** fitzdsl has joined #openstack | 08:48 | |
| *** smoser` has joined #openstack | 08:48 | |
| *** letterj_ has joined #openstack | 08:48 | |
| *** devcamcar has joined #openstack | 08:48 | |
| *** sunech has joined #openstack | 08:48 | |
| *** zelazny.freenode.net sets mode: +v _cerberus_ | 08:48 | |
| *** mtaylor has quit IRC | 08:49 | |
| *** mtaylor has joined #openstack | 08:49 | |
| *** localhost has joined #openstack | 08:50 | |
| *** ibarrera has joined #openstack | 08:53 | |
| *** mtaylor has quit IRC | 08:58 | |
| *** mtaylor has joined #openstack | 08:58 | |
| *** ChanServ sets mode: +v mtaylor | 08:58 | |
| *** fabiand_ has joined #openstack | 09:01 | |
| *** trin_cz has quit IRC | 09:01 | |
| *** rchavik has quit IRC | 09:07 | |
| *** reldan has quit IRC | 09:16 | |
| *** Jordandev has quit IRC | 09:20 | |
| *** rchavik has joined #openstack | 09:26 | |
| *** littleidea has joined #openstack | 09:29 | |
| *** sassyn has joined #openstack | 09:33 | |
| *** DigitalFlux has joined #openstack | 09:49 | |
| *** colinnich has joined #openstack | 09:55 | |
| *** aku1 has joined #openstack | 10:01 | |
| *** aku1 has quit IRC | 10:06 | |
| *** fabiand_ has left #openstack | 10:06 | |
| *** sassyn has quit IRC | 10:12 | |
| *** sassyn has joined #openstack | 10:14 | |
| *** MarkAtwood has quit IRC | 10:16 | |
| *** littleidea has quit IRC | 10:17 | |
| *** MarkAtwood has joined #openstack | 10:18 | |
| *** littleidea has joined #openstack | 10:18 | |
| *** gasbakid has quit IRC | 10:23 | |
| *** sassyn has quit IRC | 10:23 | |
| *** gasbakid has joined #openstack | 10:24 | |
| *** gasbakid has quit IRC | 10:27 | |
| *** gasbakid has joined #openstack | 10:29 | |
| *** gasbakid has quit IRC | 10:36 | |
| ttx | creiht: ok :) | 10:43 |
| *** littleidea has quit IRC | 10:49 | |
| *** _skrusty has quit IRC | 11:04 | |
| ttx | so 2011 so far seems to be the year of huge scrollbacks and hot technical discussions (UUIDs vs. readability, daemontools vs. what the world actually uses) | 11:09 |
| *** _skrusty has joined #openstack | 11:18 | |
| * soren missed the daemontools one | 11:28 | |
| *** ArdRigh has quit IRC | 11:29 | |
| * soren lunches | 11:30 | |
| *** rchavik has quit IRC | 11:32 | |
| *** gasbakid has joined #openstack | 11:36 | |
| *** schisamo has joined #openstack | 12:14 | |
| *** sandywalsh has quit IRC | 12:24 | |
| *** lool has quit IRC | 12:24 | |
| *** lool has joined #openstack | 12:25 | |
| *** MarkAtwood has quit IRC | 12:33 | |
| *** arthurc has joined #openstack | 12:35 | |
| *** sandywalsh has joined #openstack | 12:38 | |
| *** lool has quit IRC | 12:39 | |
| *** jbryce has quit IRC | 12:40 | |
| *** jbryce has joined #openstack | 12:41 | |
| *** arthurc has quit IRC | 12:42 | |
| *** arthurc has joined #openstack | 12:43 | |
| *** MarkAtwood has joined #openstack | 12:45 | |
| *** lool has joined #openstack | 12:49 | |
| *** arthurc has quit IRC | 12:50 | |
| *** arthurc has joined #openstack | 12:55 | |
| *** kashyapc has quit IRC | 13:01 | |
| *** b_erb has joined #openstack | 13:03 | |
| b_erb | hi | 13:05 |
| b_erb | i followed the SAIO instructions and additionally edited the default_cluster_url entry in auth-server.conf into a local network address (instead of 127.0.0.1). | 13:05 |
| b_erb | but after auth'ing, i still get back 127.0.0.1 URIs as X-Storage-Url | 13:05 |
| b_erb | i'm using 1.1.0 | 13:05 |
| b_erb | any idea how to get back the right url? | 13:06 |
| *** nelson__ has quit IRC | 13:10 | |
| *** nelson__ has joined #openstack | 13:11 | |
| nelson__ | ttx: there's a reason why daemontools exists: because what the world actually uses actually sucks. | 13:14 |
| soren | nelson__: I just don't really see daemontools solving any problems that I have. | 13:15 |
| nelson__ | and when people defend the suck by saying "but that's what everybody uses" I kinda am unimpressed by that argument, because suck is still suck even if everybody uses it. | 13:15 |
| ttx | nelson__: I don't defend the suck. But distributions have already chosen how they want to resolve the suck (upstart, for example). Consistency has value. | 13:16 |
| ttx | (personally, I think this needs to be solved at distro packaging level, not inside our core code) | 13:16 |
| nelson__ | soren: I understand that. Oftentimes, people don't realize that they have a problem until they see the solution. That's the role of the entrepreneur. But ahhh, I'll just put my lampshade on to continue talking about economics. :) | 13:17 |
| nelson__ | ttx: the problem is that people demand init.d, and the distros listen to them. But then to say that we should use init.d because the distros use them is circular reasoning. | 13:18 |
| ttx | nelson__: in case you have missed the memo, distros are moving away from init.d. And not towards daemontools. | 13:19 |
| nelson__ | ttx: and I agree with you that consistency has value, but not sucking also has value. | 13:19 |
| nelson__ | by distros you mean Ubuntu. | 13:19 |
| ttx | and Fedora. | 13:19 |
| nelson__ | what's Fedora using? | 13:20 |
| ttx | moving towards systemd, last time I looked. | 13:20 |
| alekibango | i would say lets support both... its possible, not much work, its nice to most ppl | 13:20 |
| nelson__ | Oh, I see, so now we can throw out consistency as a reason not to use daemontools. | 13:20 |
| ttx | Debian also moving away, though they still hesitate. daemontools is not part of their options. | 13:21 |
| ttx | nelson__: no, I throw out consistency as a reason not to make any choice. | 13:21 |
| ttx | (from the upstream project perspective) | 13:22 |
| ttx | init scripts are distribution integration. Let the distributions handle it (for consistency with the rest of their stuff) in packaging. | 13:22 |
| nelson__ | urgh, I'm not following you | 13:22 |
| nelson__ | okay, I give. Clearly I'm not going to change the world on this one. | 13:23 |
| ttx | nelson__: I see no reason for an upstream project to provide init scripts. Or daemontools scripts. Or upstart scripts. | 13:23 |
| nelson__ | I much prefer daemontools to syslog /var/log init.d /var/run and that farrago. | 13:23 |
| nelson__ | ttx: you should go read Dan Bernstein's reasoning. | 13:24 |
| nelson__ | Unix fragmentation is, in his mind, a very bad thing, because it reproduces the FAIL that caused Unix to suffer twenty years ago. | 13:24 |
| alekibango | nelson__: he is right that is packaging issue... but we have at least one packager right here :) | 13:25 |
| nelson__ | But I also realize that a lot of people pushing the suck weren't alive then. | 13:25 |
| b_erb | what do i have to configure to get non-"127.0.0.1" storage URLs? already changed default_cluster_url entry in auth-server.conf | 13:25 |
| ttx | I'm not sure I want to go deeper into djb's mind :) | 13:25 |
| *** kashyapc has joined #openstack | 13:25 | |
| nelson__ | ttx: yeah, see, now that's where you're going irrational on me. | 13:25 |
| nelson__ | djb has a very bad reputation in the Unix community ... but that's because he was very pushy. But he was pushy because he was right. | 13:26 |
| nelson__ | http://lesswrong.com/lw/372/defecting_by_accident_a_flaw_common_to_analytical/ | 13:26 |
| ttx | nelson__: being right is often not enough. | 13:26 |
| nelson__ | yes, you have to be powerful. | 13:27 |
| ttx | being accepted by others is more important. | 13:27 |
| nelson__ | and by being offensive and radical, he gave up power. | 13:27 |
| nelson__ | ja. | 13:27 |
| nelson__ | and sometimes the best solution is too much of a change for people to accept. | 13:27 |
| *** fraggeln has joined #openstack | 13:29 | |
| *** aliguori has joined #openstack | 13:32 | |
| *** jfluhmann has quit IRC | 13:35 | |
| *** jfluhmann_ has joined #openstack | 13:35 | |
| *** jbryce has quit IRC | 13:35 | |
| *** localhost has quit IRC | 13:41 | |
| *** jbryce has joined #openstack | 13:50 | |
| *** hadrian has joined #openstack | 13:51 | |
| *** Podilarius has joined #openstack | 13:51 | |
| *** roamin9_ has joined #openstack | 13:54 | |
| *** Ryan_Lane|away is now known as Ryan_Lane | 13:54 | |
| *** hggdh has quit IRC | 13:58 | |
| *** hadrian has quit IRC | 14:00 | |
| *** hggdh has joined #openstack | 14:01 | |
| *** stewart has quit IRC | 14:04 | |
| *** roamin9_ has quit IRC | 14:05 | |
| *** stewart has joined #openstack | 14:07 | |
| *** ccustine has joined #openstack | 14:08 | |
| dabo | How do I update my database to match a new model.py that adds columns? Is there a programmatic way, or do I do it manually? | 14:09 |
| sandywalsh | I think sqlalchemy does it, but migrations are being done in nova currently (afaik) | 14:13 |
| sandywalsh | *aren't | 14:13 |
| *** Lcfseth has joined #openstack | 14:13 | |
| dabo | so I should just use ALTER TABLE? | 14:15 |
| sandywalsh | if it's not faster to just drop the db | 14:16 |
| *** b_erb has quit IRC | 14:16 | |
| *** mark has joined #openstack | 14:17 | |
| *** ppetraki has joined #openstack | 14:17 | |
| dabo | nah, alter table is quicker | 14:20 |
| *** kashyapc has quit IRC | 14:21 | |
| *** hadrian has joined #openstack | 14:27 | |
| soren | AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAARGH! | 14:29 |
| Ryan_Lane | soren: one of those days? | 14:30 |
| soren | I've spent absolutely ridiculous amounts of time trying to get the SOAP API to work. Turns out what I've been doing all along has been great, except my xmlns property on my FooResponse wasn't what the client expected. | 14:31 |
| soren | ...so it just segfaulted. | 14:31 |
| Ryan_Lane | :D | 14:31 |
| *** b_erb has joined #openstack | 14:31 | |
| soren | Now to unwind all my horrid debugging hacks. | 14:31 |
| ttx | soren: SOAP is bad for you. | 14:32 |
| *** jdarcy has joined #openstack | 14:32 | |
| soren | ttx: This is where Rick would usually drop a comment about Europeans and body odour. | 14:32 |
| ttx | soren: too bad he can't talk. | 14:33 |
| soren | Yeah. Too bad. | 14:37 |
| *** littleidea has joined #openstack | 14:43 | |
| *** f4m8 is now known as f4m8_ | 14:47 | |
| *** reldan has joined #openstack | 14:48 | |
| *** sophiap has quit IRC | 14:48 | |
| *** sophiap has joined #openstack | 14:57 | |
| *** ramkrsna has quit IRC | 14:59 | |
| ttx | Ryan_Lane: you should add your blog to http://planet.openstack.org/ | 15:02 |
| creiht | b_erb: The storage url is stored in the users record in the auth.db, so if you change it in the config, any users you created previously will still have the old url | 15:02 |
| creiht | you need to either edit the db, and update the rows, or just delete and recreate your users | 15:03 |
| creiht | nelson__: Nothing prevents you from using daemontools. If there are things we can do to make it easier for you to do so, please let us know | 15:04 |
| creiht | nelson__: And if our init script is the worst thing that someone complains about for swift, then I am a happy developer :) | 15:05 |
| nelson__ | creiht: I did a trial run of it last night and swift-auth-server seems to be exiting from time to time. Not sure why. | 15:05 |
| *** nijaba has quit IRC | 15:05 | |
| creiht | nelson__: did you try running it manually? | 15:06 |
| Ryan_Lane | ttx: ah. didn't realize I could. will do | 15:06 |
| creiht | and also just to make sure we are clear, the auth server is just made for dev purposes | 15:06 |
| *** smoser` is now known as smoser | 15:06 | |
| annegentle | Ryan_Lane: yep, instructions for adding your blog to the planet: http://wiki.openstack.org/AddingYourBlog | 15:07 |
| nelson__ | am doing now. right; I know that we'll need our own auth ... but of course almost nothing we do is private. | 15:07 |
| creiht | and by exiting, do you mean at start, or at some point after running? | 15:07 |
| *** nijaba has joined #openstack | 15:07 | |
| nelson__ | after about twenty seconds. but now, from the command line, it hasn't existed. | 15:08 |
| nelson__ | I'll look into it some more. | 15:08 |
| creiht | ok | 15:08 |
| creiht | when it was exiting, were you testing at any load? | 15:08 |
| nelson__ | no | 15:08 |
| creiht | hrm | 15:09 |
| creiht | nelson__: let me know if you find anything else | 15:13 |
| colinnich | creiht: while you are on the subject, do you know if SWAuth is ready for production? | 15:13 |
| nelson__ | creiht: will do. | 15:14 |
| creiht | colinnich: depends on how you define production :) | 15:14 |
| colinnich | creiht: does it work? :-) | 15:14 |
| gholt | It's not even merged to trunk yet. Heh. | 15:15 |
| creiht | it works for as much as we have tested it so far | 15:15 |
| creiht | oh heh | 15:15 |
| creiht | didn't realize it wasn't merged yet :) | 15:15 |
| nelson__ | Oh, creiht, in #7 of http://swift.openstack.org/howto_installmultinode.html#configure-the-proxy-node why do the second two names end in _1 ? | 15:15 |
| nelson__ | I'm guessing that's a typo. | 15:15 |
| colinnich | gholt: what's the best way to get it? | 15:15 |
| creiht | nelson__: yes, copy paste error | 15:16 |
| *** dendro-afk is now known as dendrobates | 15:17 | |
| gholt | colinnich: This is the latest: https://code.launchpad.net/~gholt/swift/swauth2/+merge/43312 I've tested the heck out of it on my own, but that's never enough. And it hasn't been used in a large prod environ. | 15:17 |
| creiht | colinnich: At some point the plan is to replace the dev_auth with it | 15:17 |
| colinnich | would I take the latest swift and merge in that branch? | 15:18 |
| gholt | You can just check that branch out directly. bzr branch lp:~gholt/swift/swauth2 | 15:19 |
| *** jdarcy has quit IRC | 15:19 | |
| *** WonTu has joined #openstack | 15:19 | |
| *** WonTu has left #openstack | 15:19 | |
| gholt | I do my best to keep it up to date with changes from trunk; it currently is up to date. | 15:19 |
| colinnich | gholt: that was my next question :-) | 15:20 |
| colinnich | gholt: thanks | 15:20 |
| Ryan_Lane | in nova, am I able to have the private IP network on one vlan, and the public floating IPs on another? | 15:20 |
| colinnich | better get my skates on though, as I see there is some competition over this side of the pond now :-) | 15:21 |
| creiht | lol | 15:21 |
| * creiht guesses that colinnich is referring to our release of cloud in the UK? :) | 15:22 | |
| colinnich | yup :-) | 15:22 |
| creiht | competition makes everything better right? :) | 15:24 |
| colinnich | can't complain since you are supplying the software...... | 15:24 |
| creiht | hehe | 15:24 |
| *** blpiatt has joined #openstack | 15:26 | |
| *** westmaas has joined #openstack | 15:26 | |
| *** reldan has quit IRC | 15:26 | |
| *** reldan has joined #openstack | 15:27 | |
| *** jimbaker has joined #openstack | 15:28 | |
| *** jimbaker has joined #openstack | 15:28 | |
| *** _skrusty is now known as sdkrusty | 15:41 | |
| *** sdkrusty is now known as skrusty | 15:41 | |
| *** reldan has quit IRC | 15:43 | |
| *** rnirmal has joined #openstack | 15:43 | |
| * soren breaks for dinner | 15:46 | |
| czajkowski | soren: no break for you! slides all done for your talk at FOSDEM :) | 15:47 |
| ttx | can you do mines as well ? | 15:49 |
| czajkowski | ttx: looking forward to going ? | 15:51 |
| ttx | czajkowski: I'm going. And presenting. | 15:51 |
| ttx | czajkowski: "Why Linux distributions hate Java", in the Java devroom. | 15:51 |
| czajkowski | ttx: excellent, are you giving a lightning talk or presentation? | 15:51 |
| czajkowski | ahhhh nice | 15:51 |
| ttx | should be lots of fun, if I survive it. | 15:54 |
| czajkowski | well if you surrive the Friday night you'll be fine :) | 16:00 |
| vishy | soren: the change I'm talking about is internal so it will work with all tools | 16:01 |
| *** johnpur has joined #openstack | 16:03 | |
| *** ChanServ sets mode: +v johnpur | 16:03 | |
| *** guigui1 has quit IRC | 16:05 | |
| *** littleidea has quit IRC | 16:08 | |
| *** mdomsch has joined #openstack | 16:08 | |
| *** opengeard has quit IRC | 16:12 | |
| *** crago has joined #openstack | 16:15 | |
| nelson__ | creiht: remember that sphinx build problem I had yesterday? It seems to be a sphinx version thing: my version of sphinx doesn't define css_files, so when _theme/layout.html attempts to append something to it, it whines because it doesn't exist. | 16:18 |
| nelson__ | The fix is to simply set, rather than to append to css_files. But the next question is: why am I getting this error under Ubuntu 10.04.1 ? | 16:18 |
| nelson__ | especially since it's a supported platform. | 16:19 |
| *** calavera has quit IRC | 16:19 | |
| *** dirakx has joined #openstack | 16:19 | |
| creiht | nelson__: Most of us have sphinx v1.0 | 16:20 |
| ttx | vishy: +1 | 16:21 |
| creiht | looks like lucid's default packaging is 0.6 | 16:21 |
| *** reldan has joined #openstack | 16:22 | |
| creiht | the docs use to build with the older version | 16:22 |
| creiht | I'll add that as a bug so that we can fix it | 16:22 |
| ttx | vishy: who at Anso works on https://blueprints.launchpad.net/nova/+spec/snapshot-instance , if anyone ? | 16:22 |
| creiht | nelson__: do you still have a copy of the traceback? | 16:22 |
| vishy | ttx: no one has it, but I think it should be a relatively easy addition to the cow code I'm working on | 16:23 |
| vishy | in fact I'm not really sure how to do it without cow | 16:23 |
| nelson__ | creiht: yes, but the problem isn't in the Python code. | 16:23 |
| nelson__ | it's in the mini-language that jinja2 implements. | 16:23 |
| ttx | vishy: you think that cow code you're working on may make it in time for the Bexar feature merge window ? | 16:24 |
| nelson__ | actually, I shouldn't say that. perhaps it has the concept of "add this to the end of that even if that doesn't exist yet" | 16:24 |
| creiht | hehe | 16:24 |
| creiht | nelson__: well whatever error message that was there :) | 16:24 |
| nelson__ | do you want me to paste it here? | 16:25 |
| ttx | vishy: is that https://blueprints.launchpad.net/nova/+spec/cow-instances ? | 16:25 |
| creiht | paste.openstack.org | 16:25 |
| nelson__ | (hehe) | 16:25 |
| creiht | please | 16:25 |
| nelson__ | http://paste.openstack.org/show/376/ | 16:26 |
| nelson__ | _theme/layout.html really does have a line adding tweaks to css_files, and I looked in the files that it includes, and nobody sets css_files in the first place. | 16:27 |
| creiht | My guess is that in 1.0 it is predefined | 16:27 |
| nelson__ | obviously, if it's working for you ... yeah. | 16:27 |
| vishy | ttx: yes, almost done with that one, trying to figure out key injection using cow | 16:32 |
| *** burris has quit IRC | 16:33 | |
| ttx | vishy: ok. For reference will assign snapshot-instance to you and consider it "Not started". Feel free to reassign if needed | 16:33 |
| *** MarkAtwood has quit IRC | 16:35 | |
| *** MarkAtwood has joined #openstack | 16:35 | |
| uvirtbot | New bug: #697288 in swift "Building docs with sphinx <1.0 results in an error" [Low,Triaged] https://launchpad.net/bugs/697288 | 16:37 |
| *** burris has joined #openstack | 16:38 | |
| *** DigitalFlux has quit IRC | 16:38 | |
| *** glen_ has joined #openstack | 16:41 | |
| *** Edwin__ is now known as EdwinGrubbs | 16:41 | |
| *** glen_ is now known as glenc | 16:41 | |
| nelson__ | hrm. As far as I can tell, _static/tweaks.css isn't being used by anything. | 16:42 |
| *** dirakx has quit IRC | 16:42 | |
| nelson__ | and neither is css_files. | 16:42 |
| creiht | hrm | 16:42 |
| nelson__ | and it's not referred-to by ANYTHING in my /usr/lib/pymodules | 16:43 |
| nelson__ | sounds like dead code to me. | 16:43 |
| creiht | The theme was changed a bit ago, so it might just be from the previous theme | 16:43 |
| creiht | nelson__: thanks | 16:43 |
| nelson__ | I'll boldly remove it, and if it breaks something that somebody cares about, they can revert. | 16:44 |
| *** befreax has quit IRC | 16:46 | |
| *** joearnold has joined #openstack | 16:46 | |
| *** trin_cz has joined #openstack | 16:47 | |
| *** evilrob has left #openstack | 16:47 | |
| *** ctennis has quit IRC | 16:48 | |
| *** kdsingh has joined #openstack | 16:48 | |
| *** joearnold has quit IRC | 16:51 | |
| *** eday has joined #openstack | 16:56 | |
| *** ChanServ sets mode: +v eday | 16:56 | |
| *** ctennis has joined #openstack | 17:00 | |
| *** daleolds has joined #openstack | 17:00 | |
| *** singh_ has joined #openstack | 17:02 | |
| *** jsgotangco has joined #openstack | 17:04 | |
| ttx | jbryce: do you have the next POC meeting scheduled ? | 17:05 |
| *** ibarrera has quit IRC | 17:08 | |
| *** jsgotangco has quit IRC | 17:08 | |
| *** reldan has left #openstack | 17:10 | |
| Ryan_Lane | is it possible to have nova create and connect multiple NICs connected to multiple VLANs? | 17:11 |
| Ryan_Lane | or multiple bridges? | 17:12 |
| *** dirakx has joined #openstack | 17:12 | |
| *** brd_from_italy has quit IRC | 17:13 | |
| *** dubsquared has joined #openstack | 17:16 | |
| *** cynb has joined #openstack | 17:16 | |
| *** rcc has quit IRC | 17:17 | |
| *** jdurgin has joined #openstack | 17:18 | |
| sandywalsh | Ryan_Lane, currently nova instances can only have 1 NIC | 17:21 |
| Ryan_Lane | :( | 17:22 |
| Ryan_Lane | that's what I was assuming | 17:22 |
| sandywalsh | there is a bp for multi-nic that I was working on, but it was pushed to cactus | 17:22 |
| Ryan_Lane | so that means all IPs must be on a public IP range | 17:22 |
| sandywalsh | currently, yes | 17:22 |
| * Ryan_Lane gets ready for some dirty hacks | 17:22 | |
| sandywalsh | 'friad so | 17:23 |
| Ryan_Lane | when's ipv6 support slated? | 17:23 |
| sandywalsh | hmm, good question. I'd have to search the bp's | 17:23 |
| Ryan_Lane | (we are running out of public IPs, and can't get any more :( ) | 17:23 |
| * creiht is stashing IPs under his matress for the IPocalypse | 17:24 | |
| dendrobates | Ryan_Lane: it is planned for bexar | 17:25 |
| Ryan_Lane | ipv6 support? great | 17:25 |
| dendrobates | NTT is working on it and should be submitting a merge request soon | 17:26 |
| sandywalsh | Ryan_Lane, https://blueprints.launchpad.net/nova/+spec/ipv6-support | 17:26 |
| dendrobates | http://wiki.openstack.org/BexarIpv6supportReadme | 17:26 |
| *** rlucio has joined #openstack | 17:29 | |
| *** singh_ has left #openstack | 17:30 | |
| *** schisamo has quit IRC | 17:34 | |
| *** hggdh has quit IRC | 17:35 | |
| *** maplebed has joined #openstack | 17:35 | |
| *** hggdh has joined #openstack | 17:39 | |
| *** schisamo has joined #openstack | 17:39 | |
| zaitcev | It was quite educational in the NTT IPv6 preso that we need a replacement for NAT due to cloud, no matter if global addresses are used (incl. IPV4 global addresses). | 17:49 |
| *** joearnold has joined #openstack | 17:51 | |
| termie | vishy: are you at anso? | 17:56 |
| termie | vishy: i'm about to get a sandwich at jump start and i would like to head over | 17:56 |
| termie | vishy: also i need to get a key | 17:56 |
| *** Mr_T has joined #openstack | 17:58 | |
| *** Mr_T has left #openstack | 17:58 | |
| *** colinnich has quit IRC | 17:58 | |
| *** Ryan_Lane is now known as Ryan_Lane|food | 17:58 | |
| b_erb | creiht: zhx | 18:09 |
| b_erb | creiht: thx | 18:09 |
| xtoddx | can I bum some reviews from nova-core? https://code.launchpad.net/~anso/nova/newlog2/+merge/45100 | 18:15 |
| xtoddx | Logging is ready for merge, the rebranch was to fix the deleting and re-adding of the cloudpipe directory | 18:16 |
| *** rlucio has quit IRC | 18:18 | |
| *** jamfl__ has joined #openstack | 18:45 | |
| *** glenc has quit IRC | 18:55 | |
| *** glenc has joined #openstack | 18:55 | |
| *** glenc has quit IRC | 18:58 | |
| *** dendrobates is now known as dendro-afk | 18:58 | |
| *** dendro-afk is now known as dendrobates | 18:58 | |
| *** jamfl__ has quit IRC | 19:02 | |
| *** glenc has joined #openstack | 19:03 | |
| *** glenc has left #openstack | 19:04 | |
| *** glenc has joined #openstack | 19:04 | |
| *** glenc has joined #openstack | 19:05 | |
| *** glenc has quit IRC | 19:05 | |
| *** Ryan_Lane|food is now known as Ryan_Lane | 19:06 | |
| *** MikeZ has quit IRC | 19:14 | |
| soren | vishy: Er... Whcih change is this? | 19:17 |
| vishy | shorter ec2 references | 19:17 |
| soren | czajkowski: Slides? Whuh? | 19:17 |
| vishy | soren: so the idea is we can have full uuids but allow users to refer to them using only enough to make them unique | 19:18 |
| soren | vishy: Once again, your ideas are fascinating to me, and I still wish to subscribe to your newsletter. | 19:18 |
| vishy | :) | 19:18 |
| eday | vishy: how can we guarantee this? we still have the same problem as if we have a smaller, secondary key | 19:18 |
| soren | eday: There's a problem with the ipv6 key thing, by the way. | 19:19 |
| eday | soren: I hadn't thought much about it, was an idea from pvo :) | 19:19 |
| soren | eday: While it indeed is 128 bit, we don't control the whole thing. Maybe we just have a /64 and then we're down to a 64 bit key space. | 19:19 |
| vishy | soren, eday: say your uuid is 234af34fdc2343bd34342 for example | 19:19 |
| *** glenc has joined #openstack | 19:20 | |
| vishy | you can do: euca-describe-instances ami-234 | 19:20 |
| soren | Oh. | 19:20 |
| soren | That doesn't really help. | 19:20 |
| vishy | the ec2_api can do a fuzzy search | 19:20 |
| soren | Well, a bit. | 19:20 |
| eday | vishy: what if another instance is created with that prefix before the next req? | 19:20 |
| soren | It doesn't solve the problem of clients suddenly receiving ID's massively longer than they expect. | 19:21 |
| vishy | it just gives an error, duplicate | 19:21 |
| vishy | soren: true, but it helps a lot with usability | 19:21 |
| soren | vishy: True. | 19:21 |
| vishy | alternatively, I really like the idea of nicknaming making display_name unique per project | 19:22 |
| vishy | and also using it for reverse dns | 19:22 |
| vishy | unfortunately it would require some kind of availability zone hack to allow for it on run | 19:22 |
| vishy | euca-run-instances -z east1:blue | 19:22 |
| soren | I still believe 128 bit keys are overkill. | 19:22 |
| vishy | euca-describe-instances ami-blue | 19:23 |
| vishy | er | 19:23 |
| vishy | euca-describe-instances i-blue | 19:23 |
| glenc | d | 19:23 |
| vishy | don't know why I was using ami above, which is the prefix for images not instances | 19:23 |
| * nelson__ has his own class C. | 19:23 | |
| glenc | excuse me | 19:23 |
| *** Zlasher has joined #openstack | 19:24 | |
| soren | vishy: 40 bit keys still lets us have a million physical hosts with a million VM's on each. | 19:25 |
| soren | That's about as far as my ambitions go. | 19:25 |
| soren | For this lifetime. | 19:26 |
| vishy | soren: just looking at your email, compatibility concerns are a bit annoying | 19:26 |
| soren | They are. | 19:26 |
| soren | Most things we're working on are :) | 19:26 |
| Zlasher | Is't possible to run game servers like Minecraft, Battlefield and such in a cloud enviorment? | 19:27 |
| vishy | soren: can we use a uuid and just use a portion of it for the ec2api? | 19:27 |
| vishy | soren: so internally openstack can support a much larger space? | 19:27 |
| eday | vishy: we could, but it depends on the UUID algorithm we use | 19:27 |
| soren | vishy: There'll be a miniscule chance of collisions. | 19:27 |
| eday | vishy: some bits are more static than others for different algos | 19:28 |
| soren | One thing we haven't really anaylised though: | 19:28 |
| eday | (ie, mac address of generating host) | 19:28 |
| soren | What will happen if there's a collision? | 19:28 |
| eday | soren: thats what I asked above, vish said respond with an error | 19:28 |
| soren | Oh, er... no. | 19:28 |
| soren | That's not what I meant. | 19:28 |
| eday | and then you'd need to relist your instances, and get a different ID set | 19:28 |
| eday | which could be a problem | 19:28 |
| eday | if you store your instance list and it changes underneath you (although it's the same instances with longer IDs to make them unique) | 19:29 |
| soren | I mean.. Not a collision after we truncate the full uuid to the ec2 id. | 19:29 |
| soren | But an actual collision. | 19:29 |
| soren | If two hosts generated the same uuid, independently. | 19:29 |
| eday | soren: we can guarantee we don't with UUIDs, some implementations give you that | 19:29 |
| eday | unless you use the same mac address for two hosts, in which case you should be shot | 19:29 |
| soren | Yes, yes, but. | 19:29 |
| soren | This is not the point. | 19:30 |
| *** troytoman has joined #openstack | 19:30 | |
| soren | I'm trying to determine, in the theoretical case where you accidentally assign the same uuid to two instances, what happens? | 19:30 |
| soren | What's the effect? | 19:30 |
| soren | Let's forget uuid's for a second. | 19:30 |
| soren | If we had a 4 bit keyspace. | 19:31 |
| soren | Yes, four. | 19:31 |
| soren | Collisions are guaranteed. :) | 19:31 |
| soren | What happens when you get a collision? | 19:31 |
| eday | well, you could detect and log it for admin resolution, but, if you use a proper UUID lib where this never happens, it's a non-issue :) | 19:31 |
| soren | No, no. | 19:31 |
| soren | I'm.. | 19:31 |
| soren | Erm.. | 19:31 |
| soren | Ok, let's try it like this: | 19:32 |
| eday | I mean, why consider cases that we can guarantee don't happen? | 19:32 |
| soren | I just think it's interesting to understand exactly what we're trying to avoid. | 19:32 |
| eday | if we did | 19:33 |
| soren | If the only effect is that one instance every 28 years across the entire infrastructure fails to start... | 19:33 |
| eday | any request to the system would return an internall error | 19:33 |
| soren | I don't care much. | 19:33 |
| eday | and an admin notification is sent (email, log, ...) to resolve it | 19:33 |
| eday | would be the easy way | 19:33 |
| soren | If it means that every once in a while, people can steal each other's data, I care. | 19:33 |
| eday | right now the DB looks have first() appended to the sqlalchemy queries, so you would just get the first one according to insert order | 19:34 |
| eday | which would be very bad | 19:34 |
| soren | Well.. | 19:34 |
| eday | we would want to make sure only one was returned | 19:34 |
| eday | for a given key | 19:34 |
| soren | Right. | 19:34 |
| eday | and if > 1, report errors | 19:34 |
| eday | but that, to me, is an unacceptable case for a distributed system :) | 19:35 |
| soren | Let's start over.. | 19:35 |
| soren | Are we in agreement that 128 bit keys brings along some problems? | 19:35 |
| eday | If we care about ec2 api, it may have some problems, yes :) | 19:35 |
| soren | Ok. | 19:36 |
| soren | There are a number of ways to attack this: | 19:36 |
| soren | 1) Ignore it. | 19:36 |
| soren | 2) Keep the 128 internally (and in the OS api) and find a reaonsably good hashing algorithm to turn 128 bit keys into 40 bit ones for EC2. | 19:37 |
| soren | 3) Use smaller keys. | 19:37 |
| vishy | http://www.jackofallclouds.com/2009/09/anatomy-of-an-amazon-ec2-resource-id/ | 19:37 |
| soren | 4) Change the EC2 api in curious ways (that people also don't expect). | 19:37 |
| eday | I see 2 and 3 as the same really | 19:38 |
| eday | what matters is how unique the smallest key is | 19:39 |
| eday | and if you can't guarantee it, what you do to resolve | 19:39 |
| ttx | You could make 1-3 configurable | 19:39 |
| eday | so hashing a larger key to make a smaller one is really just using a smaller one | 19:39 |
| *** glenc has left #openstack | 19:40 | |
| termie | i am late to the conversation but what examples do we have of people not being able to fit our keys? | 19:40 |
| ttx | so that people can tune their EC2-lookalikeness level | 19:40 |
| soren | scalr. | 19:40 |
| soren | For one. | 19:40 |
| soren | It was the first I looked at. | 19:40 |
| termie | soren: varchar(20) you said? | 19:40 |
| eday | ttx: I'm afraid options for something so fundamental will lead to extra complexity and bugs :/ | 19:40 |
| soren | termie: Yeah. | 19:40 |
| termie | soren: 128 bits fit in that as base-36 don't they? | 19:40 |
| soren | termie: Nope. | 19:40 |
| soren | ln(2^128)/ln(36) ~24.75857966301066155973 | 19:41 |
| soren | 25 chars. | 19:41 |
| soren | plus "i-" | 19:41 |
| soren | 27. | 19:41 |
| *** Lcfseth has quit IRC | 19:41 | |
| *** MikeZ has joined #openstack | 19:41 | |
| *** MikeZ1 has joined #openstack | 19:41 | |
| soren | I still rather like the idea of handing out ID's in a great big bulks. | 19:42 |
| soren | Once a new server is started that needs to assign ID's for something, it requests a great big pile of them. | 19:43 |
| vishy | you and your natural logs | 19:43 |
| soren | They're handed out from some sort of central authority, ensureing uniqueness. | 19:43 |
| termie | hmm, amazon claims that their implementation details for their ids are subject to change | 19:43 |
| vishy | or 30 for "snap-" | 19:43 |
| soren | termie: They certainly do. | 19:43 |
| soren | termie: ...but they haven't, so clients don't expect it. | 19:44 |
| termie | i don't know whether 1 example is enough | 19:44 |
| termie | does euca care? | 19:44 |
| soren | They generate ID's of similar size to EC2. | 19:44 |
| termie | do their tools care | 19:44 |
| termie | does boto | 19:44 |
| soren | No. | 19:44 |
| soren | Python rarely cares about such things. | 19:45 |
| termie | my gut instinct would say it probably isn't actually a very widespread problem | 19:45 |
| soren | C and Java do. | 19:45 |
| soren | Anything that stores anything in a database does. | 19:45 |
| *** MikeZ has quit IRC | 19:46 | |
| soren | The EC2 API isn't an open standard. Their API reference can say whatever it pleases. If the actual service behaves differently, that's what we need to do, too. | 19:46 |
| soren | We're not trying to be better at implementing their API according to their spec than they are. | 19:46 |
| vishy | so do we just keep a separate ec2 id? | 19:46 |
| termie | not claiming we are, just trying to get some examples of tools that care | 19:47 |
| vishy | perhaps even in the compatibility layer? | 19:47 |
| termie | for example, they claim things should just take strings | 19:47 |
| termie | so base-64 is fine | 19:47 |
| termie | as an encoding format, however plenty of things probably aren't escaping their output properly | 19:48 |
| vishy | base 64 only saves a couple chars | 19:48 |
| termie | because they don't expect fancy chars | 19:48 |
| soren | Right. | 19:49 |
| termie | we are using base-36 right now from what i gather, which appears already different from ec2 | 19:49 |
| *** b_erb has left #openstack | 19:49 | |
| soren | How so? | 19:49 |
| termie | they are using hex :/ | 19:50 |
| soren | I thought that's what they used. | 19:50 |
| vishy | they use base-16 i think | 19:50 |
| soren | Oh. Wow. | 19:50 |
| soren | bc | 19:50 |
| soren | whoops | 19:50 |
| termie | they don't claim to use hex, but it is what they are doing | 19:50 |
| soren | Ok, so they're actually using 32 bit keys. Wow. | 19:51 |
| vishy | agreed | 19:51 |
| termie | soren: based on that link though their keys are also rather silo'd | 19:51 |
| soren | termie: Yup. | 19:51 |
| soren | I think that's less of an issue than the length, though. | 19:51 |
| *** glenc has joined #openstack | 19:51 | |
| termie | which means, amongst other things, that they are probably using batches assigned by different master counters | 19:52 |
| soren | I doubt very many consumers of the api are taking their time to strip off the "i-" prefix and converting to an int and converting back when they need it. | 19:52 |
| soren | The length of the thing is much more likely to be a problem. | 19:52 |
| termie | but, again, a problem for who | 19:53 |
| termie | do we expect a website that is built against amazon to be able to switch to openstack seamlessly? | 19:53 |
| soren | The fact that there are seams isn't must of a justification to add more. | 19:53 |
| soren | I'd hope they could do it seamlessly. | 19:54 |
| termie | no, but it is important to decide where our goals are | 19:54 |
| *** joearnold has quit IRC | 19:54 | |
| termie | we could possibly go through a lot of effort to make a system that is that compatible without actually needing to | 19:54 |
| soren | Mine is quite short: I want to let people currently using EC2 to be able to point their stuff at nova and not be surprised. | 19:54 |
| pvo | I don't think our goals are to provide seamless compatability | 19:55 |
| soren | Who are "we"? | 19:55 |
| termie | if all of our users are built for openstack and can leverage most existing api libraries i think we are pretty much there | 19:55 |
| soren | If we don't try to make it seamless, why do we try at all? | 19:55 |
| termie | soren: us? myself, you, vishy... | 19:55 |
| pvo | i interpreted we as "openstack community" | 19:55 |
| soren | Well, I just said it was one of my goals. pvo said it wasn't one of "ours". I wonder who "we" are. | 19:56 |
| vishy | soren: there isn't really a good solution to this | 19:56 |
| vishy | I don't want to adopt amazons keysize particularly | 19:56 |
| soren | vishy: That's really not been established. | 19:56 |
| soren | vishy: Hang on, let me run some numbers real quick. | 19:56 |
| vishy | soren: so it seems like a compatibility layer is the "best" option | 19:56 |
| soren | How about this: | 19:58 |
| soren | The EC2 API grows a new flag: | 19:58 |
| ttx | --use_short_ec2like_ids_that_increase_collisions ? | 19:59 |
| soren | ec2_id_key_bits | 19:59 |
| soren | Pretty much, yes. | 19:59 |
| soren | If someone actually has a collision, I'll buy them a beer. | 19:59 |
| soren | Seriously, if they run a million instances, there's a 1% chance they'll have had 1 (one) collision. | 19:59 |
| soren | Ever. | 20:00 |
| soren | With 40 bit keys, that is. | 20:00 |
| * ttx makes mental note. I generated a collision on a eucalyptus setup, but that might just be their rng | 20:00 | |
| sandywalsh | how about our access keys with project tacked on the end? | 20:00 |
| soren | ttx: Maybe they copied the RNG from openssl. | 20:00 |
| * soren runs | 20:00 | |
| * soren needs to tend to other business for a while | 20:01 | |
| sandywalsh | (speaking of not exactly compatible) | 20:01 |
| vishy | I'm happy with the flag, does it just fill the rest of the internal_id with zeros? | 20:02 |
| sleepson- | soren - isn't there a 50% chance of collision at 1M? | 20:04 |
| *** anotherjesse has joined #openstack | 20:05 | |
| *** trin_cz has quit IRC | 20:05 | |
| *** glenc_ has joined #openstack | 20:20 | |
| *** glenc has left #openstack | 20:20 | |
| vishy | is anyone else getting TypeError: argument must be 9-item sequence, not datetime.datetime in trunk? | 20:25 |
| vishy | ah nm, my bad | 20:26 |
| ttx | Team meeting in 35 minutes in #openstack-meeting ! | 20:27 |
| *** joearnold has joined #openstack | 20:27 | |
| *** glenc_ has quit IRC | 20:27 | |
| *** troytoman has left #openstack | 20:27 | |
| *** troytoman has joined #openstack | 20:27 | |
| *** glenc has joined #openstack | 20:28 | |
| *** glenc_rackspace has joined #openstack | 20:28 | |
| *** glenc_rackspace has quit IRC | 20:28 | |
| *** pothos_ has joined #openstack | 20:29 | |
| *** pothos has quit IRC | 20:30 | |
| *** pothos_ is now known as pothos | 20:30 | |
| Ryan_Lane | should uec images work now that raw image support is in trunk? | 20:32 |
| dubsquared | +1 | 20:33 |
| *** trin_cz has joined #openstack | 20:36 | |
| ttx | Ryan_Lane: they did last time I checked | 20:38 |
| Ryan_Lane | hmm | 20:38 |
| Ryan_Lane | I must have some other problem | 20:38 |
| ttx | Ryan_Lane: did you use uec-publish-tarball to register them ? | 20:39 |
| Ryan_Lane | yep | 20:39 |
| ttx | Ryan_Lane: what are the symptoms ? | 20:39 |
| Ryan_Lane | sec | 20:40 |
| soren | sleepson-: How do you figure that? | 20:40 |
| Ryan_Lane | ttx: http://pastebin.com/NQCvqtKn | 20:40 |
| Ryan_Lane | in nova-compute log | 20:40 |
| soren | sleepson-: Darn it, I closed the terminal where I had run my numbers. | 20:42 |
| ttx | doesn't ring a bell. Do the ttylinux images work any better ? | 20:42 |
| ttx | Ryan_Lane: ^ | 20:42 |
| Ryan_Lane | I'm using a flatnetwork | 20:43 |
| Ryan_Lane | ttylinux images don't work | 20:43 |
| soren | Ryan_Lane: Look in /var/log/libvirt/qemu/instance-1315349274.log | 20:43 |
| Ryan_Lane | unless that's been changed recently | 20:43 |
| zul | devcamcar: ping where is the web ui stuff, i thought it was proposed already? | 20:43 |
| Ryan_Lane | soren: not much there: http://pastebin.com/bsnK69Ng | 20:44 |
| eday | soren: birthday paradox | 20:44 |
| Ryan_Lane | I wonder if apparmor is causing problems | 20:44 |
| soren | Ryan_Lane: Plenty of space left? | 20:44 |
| Ryan_Lane | ah. lemme check that | 20:44 |
| Ryan_Lane | bingo | 20:45 |
| Ryan_Lane | that may be it | 20:45 |
| Ryan_Lane | well. maybe, maybe not. 3GB left | 20:45 |
| Ryan_Lane | but to be sure, clearing that out and trying again | 20:45 |
| soren | eday: Sure, I understand the maths. :) | 20:45 |
| soren | eday: I just suspect there's a bracket that got misplaced, because now that I try again, I get ~50%. | 20:46 |
| *** ctennis has quit IRC | 20:46 | |
| eday | soren: ahh, ok | 20:48 |
| ttx | Team meeting in 10 minutes in #openstack-meeting ! | 20:49 |
| soren | eday: Oh, heh. | 20:51 |
| soren | eday: Nope, I just missed a 0 in my million :) | 20:51 |
| soren | For 100,000 instances, it comes out to about 1%. | 20:51 |
| *** arthurc has quit IRC | 20:53 | |
| soren | sleepson-: For 1,000,000 it's probably around 50%, yeah. My pessimistic estimate says 59%, so 50% is probably more accurate. | 20:53 |
| *** arthurc has joined #openstack | 20:54 | |
| Ryan_Lane | :( it isn't a space issue | 20:54 |
| soren | Ryan_Lane: Is it reproducable? | 20:54 |
| Ryan_Lane | it happens every time i try to run an instance :) | 20:55 |
| Ryan_Lane | using version 2011.1~bzr515-0ubuntu0ppa1~lucid1 | 20:55 |
| *** colinnich has joined #openstack | 20:56 | |
| soren | Ryan_Lane: Ok, pastebin the output of "file /var/lib/nova/instances/*/*" | 20:56 |
| Ryan_Lane | soren: http://pastebin.com/L3WjhC5E | 20:57 |
| *** arthurc has quit IRC | 20:59 | |
| soren | Ryan_Lane: No ramdisk? | 20:59 |
| soren | Ryan_Lane: is that intentional? | 20:59 |
| Ryan_Lane | soren: that's how it comes from uec | 20:59 |
| *** enigma has joined #openstack | 20:59 | |
| ttx | Team meeting NOW in #openstack-meeting. | 21:00 |
| *** DigitalFlux has joined #openstack | 21:07 | |
| *** littleidea has joined #openstack | 21:12 | |
| soren | sleepson-: Ok, 36%. That's my final offer :) | 21:16 |
| *** blpiatt has quit IRC | 21:20 | |
| sleepson- | soren - heh :) I usually just ballpark bday problems with sqrt(n) for 50% collision probability, but it is closer to 1.18*sqrt(n), so 36% sounds right | 21:23 |
| *** EdwinGrubbs has quit IRC | 21:25 | |
| soren | sleepson-: Oh. I did 100*(1-((2^40)!/((2^40)^1000000*((2^40)-1000000)!)). | 21:25 |
| soren | sleepson-: ...with Stirling approximation for the large factorials. | 21:26 |
| soren | Whoops, missed a bracket at the end. | 21:27 |
| *** rlucio has joined #openstack | 21:28 | |
| *** dfg_ has joined #openstack | 21:29 | |
| *** kdsingh has quit IRC | 21:30 | |
| *** MarkAtwood has quit IRC | 21:34 | |
| anotherjesse | littleidea / DigitalFlux - sorry - had a call - want to continue talk about puppet stuff here? | 21:35 |
| *** EdwinGrubbs has joined #openstack | 21:36 | |
| *** EdwinGrubbs has joined #openstack | 21:36 | |
| *** DigitalFlux has quit IRC | 21:37 | |
| *** bfschott has joined #openstack | 21:38 | |
| *** colinnich has quit IRC | 21:39 | |
| soren | sleepson-: http://paste.openstack.org/show/382/ if you care. | 21:39 |
| *** spectorclan has joined #openstack | 21:42 | |
| * ttx calls it a day | 21:42 | |
| littleidea | anotherjesse: let's take to email, I have to do some stuff for a meeting in 20 minutes. | 21:42 |
| anotherjesse | littleidea: sounds good | 21:43 |
| *** bfschott has left #openstack | 21:49 | |
| *** ArdRigh has joined #openstack | 21:51 | |
| *** masumotok_ has joined #openstack | 21:51 | |
| *** DigitalFlux has joined #openstack | 21:58 | |
| *** sophiap_ has joined #openstack | 22:01 | |
| *** sophiap has quit IRC | 22:01 | |
| *** sophiap_ is now known as sophiap | 22:01 | |
| spectorclan | Sorry I missed meeting today, had another meeting. If there are any issues for me - feel free to let me know | 22:03 |
| *** miclorb has joined #openstack | 22:03 | |
| *** MarkAtwood has joined #openstack | 22:06 | |
| *** blpiatt has joined #openstack | 22:06 | |
| *** MarkAtwood has quit IRC | 22:07 | |
| anotherjesse | DigitalFlux: I can talk about the existing puppet stuff ... | 22:10 |
| sleepson- | soren - cool - need to brush up on my math a bit though :) | 22:10 |
| *** DigitalFlux has quit IRC | 22:11 | |
| *** crago has left #openstack | 22:15 | |
| soren | sleepson-: :) | 22:17 |
| * soren heads bedwards | 22:17 | |
| dubsquared | Ryan_Lane: Did you ever find out the issue with this error —> libvir: QEMU error : Domain not found: no domain with matching name 'instance-407726969' | 22:18 |
| Ryan_Lane | dubsquared: nope | 22:18 |
| dubsquared | I just ran into it... | 22:18 |
| dubsquared | :| | 22:18 |
| dubsquared | 10.04 i get that error…10.10…i do not | 22:18 |
| dubsquared | 10.10 i am getting this in the nova-compute.log —> NotFound: No instance for id 3 | 22:19 |
| dubsquared | but…it still boots | 22:19 |
| *** westmaas has quit IRC | 22:20 | |
| rlucio | you should turn on verbose/debug messaging | 22:21 |
| rlucio | the domain errors are just the surface error message | 22:21 |
| rlucio | basically saying that a query to libvirt for a particular VM instance id failed | 22:22 |
| dubsquared | oh wow…i didn't know that isn't default | 22:22 |
| dubsquared | lemme do...thx.. | 22:22 |
| rlucio | np | 22:23 |
| rlucio | if things were working before, it could be something silly like the partitioning failed because your drive is out of space ro something | 22:24 |
| rlucio | its happened to me a few times :) | 22:24 |
| dubsquared | this is my first run with images w/o a ramdisk | 22:24 |
| dubsquared | the UEC images... | 22:24 |
| rlucio | ah | 22:24 |
| *** MikeZ1 has quit IRC | 22:25 | |
| rlucio | regardless though, it should show some sort of relevant trace in the compute log | 22:26 |
| *** dendrobates is now known as dendro-afk | 22:27 | |
| dubsquared | and you just use the public key to login i presume? | 22:28 |
| rlucio | huh? you mean to log into the vm? | 22:29 |
| eday | vishy: https://code.launchpad.net/~eday/nova/internal-api-cleanup/+merge/45152 could use some lovin' when you have a moment :) | 22:29 |
| *** opengeard has joined #openstack | 22:29 | |
| *** johnpur has quit IRC | 22:31 | |
| *** MikeZ has joined #openstack | 22:31 | |
| dubsquared | Yeah | 22:34 |
| *** Podilarius has quit IRC | 22:37 | |
| *** MarkAtwood has joined #openstack | 22:41 | |
| *** Jordandev has joined #openstack | 22:44 | |
| rlucio | dubsquared: i honestly dont know how to login to new uec vms, i'd have to look it up. im sure there are docs on it though on ubuntu's site | 22:46 |
| *** dendro-afk is now known as dendrobates | 22:46 | |
| dubsquared | rlucio: what? documentation? shenanigans! | 22:46 |
| rlucio | https://help.ubuntu.com/community/UEC | 22:48 |
| rlucio | looks like there is some web url you are supposed to use to create an account | 22:48 |
| rlucio | /shrug | 22:49 |
| *** Zlasher has quit IRC | 22:51 | |
| *** kevnfx has joined #openstack | 22:53 | |
| dubsquared | interesting… | 22:53 |
| dubsquared | and not in a good way | 22:53 |
| rlucio | well, if you can connect to the vm via serial console while it boots apparently it will print the password somewhere along the line | 22:55 |
| rlucio | and once you set up the vm the way you like it, you can always bundle it up as a new images so you have a built-in acct | 22:55 |
| *** MikeZ1 has joined #openstack | 23:03 | |
| *** nelson__ has quit IRC | 23:04 | |
| dubsquared | Yeah, I was just reading about that | 23:04 |
| dubsquared | great info.. | 23:04 |
| *** nelson__ has joined #openstack | 23:04 | |
| *** mdomsch has quit IRC | 23:05 | |
| *** anotherjesse has quit IRC | 23:06 | |
| *** MikeZ has quit IRC | 23:06 | |
| *** enigma has left #openstack | 23:07 | |
| *** rnirmal has quit IRC | 23:12 | |
| *** ppetraki has quit IRC | 23:12 | |
| annegentle | just fyi to all - for all the nova docs, I've changed Copyright 2010 to 2010-2011. I'll do the same for swift next. I haven't touched .py files though. | 23:15 |
| annegentle | this is the branch where the 2010-2011 statements are: https://code.launchpad.net/~annegentle/nova/fixnewscript/+merge/45085 | 23:15 |
| *** gasbakid has quit IRC | 23:25 | |
| *** ctennis has joined #openstack | 23:26 | |
| *** ctennis has joined #openstack | 23:26 | |
| *** daleolds has quit IRC | 23:32 | |
| *** blpiatt has quit IRC | 23:35 | |
| *** dendrobates is now known as dendro-afk | 23:39 | |
| *** spectorclan has quit IRC | 23:41 | |
| *** dendro-afk is now known as dendrobates | 23:46 | |
| *** daleolds has joined #openstack | 23:47 | |
| *** troytoman has left #openstack | 23:49 | |
| *** gondoi has quit IRC | 23:54 | |
| nelson__ | http://twitter.com/#!/search/Swift%20%26%20Jake%20Gyllenhaal | 23:57 |
| nelson__ | apparently a trending topic on twitter. I didn't know Jake was a geek. | 23:57 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!