| *** johnbergoon has joined #openstack | 00:06 | |
| *** mtaylor has quit IRC | 00:07 | |
| *** jakedahn has quit IRC | 00:07 | |
| *** iammartian_ has quit IRC | 00:12 | |
| *** daleolds has joined #openstack | 00:16 | |
| *** aliguori has quit IRC | 00:18 | |
| *** Sanjib has quit IRC | 00:18 | |
| *** iammartian_ has joined #openstack | 00:19 | |
| *** DubLo7 has joined #openstack | 00:22 | |
| *** mmalgeri has joined #openstack | 00:31 | |
| *** mmalgeri has quit IRC | 00:32 | |
| *** mmalgeri has joined #openstack | 00:35 | |
| *** rlucio has quit IRC | 00:35 | |
| *** mmalgeri has quit IRC | 00:35 | |
| *** mmalgeri has joined #openstack | 00:36 | |
| *** jakedahn has joined #openstack | 00:36 | |
| *** kevnfx has joined #openstack | 00:46 | |
| kevnfx | vishy: cool ... much appreciated ... | 00:47 |
|---|---|---|
| *** ar1 has joined #openstack | 00:47 | |
| *** kevnfx has left #openstack | 00:49 | |
| vishy | they aren't in trunk but you can grab the branch if you want | 00:49 |
| vishy | lp:~anso/nova/fix-context-deprecation | 00:50 |
| *** jc_smith has quit IRC | 00:51 | |
| *** mmalgeri has quit IRC | 00:57 | |
| *** daleolds has quit IRC | 01:11 | |
| *** kevnfx has joined #openstack | 01:15 | |
| *** KanGouLya has quit IRC | 01:15 | |
| *** mtaylor has joined #openstack | 01:18 | |
| *** ChanServ sets mode: +v mtaylor | 01:18 | |
| *** daleolds has joined #openstack | 01:24 | |
| *** KanGouLya has joined #openstack | 01:30 | |
| *** schisamo has quit IRC | 01:39 | |
| *** KanGouLya has quit IRC | 01:45 | |
| *** rnirmal has joined #openstack | 01:53 | |
| *** jsgotangco has joined #openstack | 01:59 | |
| *** jsgotangco has quit IRC | 02:03 | |
| *** joearnold has quit IRC | 02:17 | |
| *** joearnold has joined #openstack | 02:18 | |
| *** metoikos has joined #openstack | 02:19 | |
| *** joearnold has quit IRC | 02:22 | |
| *** greenisus has quit IRC | 02:26 | |
| *** AimanA is now known as HouseAway | 02:45 | |
| *** mmalgeri has joined #openstack | 02:47 | |
| *** mmalgeri has quit IRC | 03:05 | |
| *** mmalgeri has joined #openstack | 03:10 | |
| *** iammartian_ has quit IRC | 03:14 | |
| *** jakedahn has left #openstack | 03:16 | |
| *** kashyapc has joined #openstack | 03:38 | |
| *** DubLo7 has quit IRC | 03:39 | |
| *** howardroark has joined #openstack | 03:51 | |
| *** joearnold has joined #openstack | 03:53 | |
| *** dysinger has joined #openstack | 03:57 | |
| *** kevnfx has quit IRC | 04:07 | |
| *** ArdRigh has quit IRC | 04:11 | |
| *** dysinger has quit IRC | 04:14 | |
| *** miclorb has quit IRC | 04:17 | |
| *** silassewell has quit IRC | 04:23 | |
| *** rnirmal has quit IRC | 04:27 | |
| *** daleolds has quit IRC | 04:29 | |
| *** jakedahn has joined #openstack | 04:38 | |
| *** jakedahn has quit IRC | 04:39 | |
| *** jakedahn has joined #openstack | 04:41 | |
| *** petabit has joined #openstack | 04:42 | |
| *** jakedahn has quit IRC | 04:42 | |
| *** johnbergoon has quit IRC | 04:44 | |
| *** petabit has quit IRC | 04:44 | |
| *** jakedahn has joined #openstack | 04:46 | |
| *** petabit has joined #openstack | 04:47 | |
| *** petabit has quit IRC | 04:50 | |
| *** kashyapc has quit IRC | 04:51 | |
| *** kashyapc has joined #openstack | 04:51 | |
| *** kashyapc_ has joined #openstack | 04:52 | |
| *** gundlach has quit IRC | 04:52 | |
| *** abecc has quit IRC | 04:53 | |
| *** f4m8_ is now known as f4m8 | 04:53 | |
| *** Ryan_Lane_ has joined #openstack | 04:54 | |
| DesiJat | ? | 04:54 |
| *** joearnold has quit IRC | 04:59 | |
| *** Ryan_Lane_ has quit IRC | 05:02 | |
| *** mgriffin has quit IRC | 05:13 | |
| *** kainam has quit IRC | 05:20 | |
| *** khussein has joined #openstack | 05:43 | |
| *** guigui has joined #openstack | 05:45 | |
| *** arthurc has joined #openstack | 05:50 | |
| *** comstud has joined #openstack | 05:51 | |
| *** ChanServ sets mode: +v comstud | 05:51 | |
| *** Kdecherf has quit IRC | 05:57 | |
| *** Kdecherf has joined #openstack | 05:57 | |
| *** sirp1 has quit IRC | 06:13 | |
| *** sirp1 has joined #openstack | 06:15 | |
| *** ibarrera has joined #openstack | 06:33 | |
| *** ArdRigh has joined #openstack | 06:40 | |
| *** khussein has quit IRC | 06:47 | |
| *** miclorb has joined #openstack | 06:56 | |
| soren | vishy: Still around? | 07:08 |
| soren | vishy: Do you guys actually /use/ the default_kernel and default_ramdisk flags? | 07:08 |
| soren | vishy: I'm thinking about getting rid of them. | 07:08 |
| soren | vishy: So, for instance, the images Ubuntu publishes for EC2 don't have a ramdisk. | 07:08 |
| soren | vishy: EC2 and Eucalyptus both support booting without a ramdisk. | 07:09 |
| soren | vishy: ...but because Nova uses a default one in the absence of an explicitly requested one, that doesn't work. | 07:09 |
| vishy | soren: i think we still have some images that don't have it specified | 07:09 |
| vishy | soren: but just legacy | 07:10 |
| soren | vishy: ...as for the default kernel, I'm considering getting rid of it as an alternate means for implementing Justin's raw disk thing. | 07:10 |
| vishy | soren: i think that seems fine | 07:10 |
| soren | vishy: His raw-disk think makes you specify an explicit, magic kernel id that makes it do things differently. It seems like The Right Thing to simply do those things if no kernel is specified. | 07:11 |
| vishy | soren: images have ramdisk specfied if they have one or none if they are whole disk booted | 07:11 |
| vishy | sure | 07:11 |
| soren | vishy: The only problem is that it's hard to discover if your image actually does need a kernel, but you just forgot to specify one. | 07:11 |
| *** joearnold has joined #openstack | 07:11 | |
| vishy | yeah | 07:12 |
| *** calavera has joined #openstack | 07:14 | |
| *** kashyapc has quit IRC | 07:16 | |
| vishy | soren: do you use flat networking mode? | 07:23 |
| soren | vishy: I believe I do, yes. | 07:26 |
| vishy | just wondering if anyone has tried the new flatdhcp version | 07:27 |
| soren | Odd sort of answer, I suppose. | 07:27 |
| soren | I haven't, no. Not yet. Not had a chance. | 07:27 |
| soren | I'm hoping to do so today, though. | 07:27 |
| *** allsystemsarego has joined #openstack | 07:28 | |
| *** joearnold has quit IRC | 07:31 | |
| zykes- | alekibango: around ? | 07:45 |
| *** kashyapc has joined #openstack | 07:47 | |
| *** perestrelka has quit IRC | 07:59 | |
| zykes- | what's UML ? | 08:06 |
| soren | User Mode Linux. | 08:08 |
| zykes- | which does ? | 08:11 |
| soren | Runs linux. | 08:13 |
| soren | In userspace. | 08:13 |
| soren | It's a sort of a hypervisor | 08:13 |
| *** perestrelka has joined #openstack | 08:14 | |
| zykes- | so let's you test stuff then i guess without needing a phys. host | 08:19 |
| *** miclorb has quit IRC | 08:20 | |
| soren | Right. | 08:20 |
| *** miclorb_ has joined #openstack | 08:27 | |
| *** jonwood has joined #openstack | 08:51 | |
| alekibango | zykes-: ? | 08:57 |
| zykes- | so turbogears ? | 09:00 |
| alekibango | zykes-: why not that objectiveJ one? :) | 09:00 |
| alekibango | btw i have seen author on irc yesterday (greenisus) | 09:00 |
| zykes- | alekibango: thought a python one was wanted but ok | 09:01 |
| alekibango | zykes-: imho some system which would include some accounting would be nice... and one with adaptable billing method... | 09:01 |
| alekibango | and imho tg is good for that | 09:01 |
| zykes- | doesn't accounting and billing allready exist in nova ? | 09:02 |
| alekibango | most companies will want own billing system | 09:02 |
| alekibango | own set of SLAs | 09:02 |
| alekibango | zykes-: no, accounting which will suit generic SLAs is not there | 09:03 |
| zykes- | ah | 09:03 |
| alekibango | i think making generic one would be a hit | 09:03 |
| alekibango | but 'generic' is hard to define right now | 09:04 |
| zykes- | heh | 09:04 |
| zykes- | what to start making first then ? | 09:04 |
| jonwood | alekibango: Do you personally have a need for billing and SLAs? | 09:05 |
| alekibango | jonwood: not right now, but i will be | 09:06 |
| jonwood | Wait until you have that need before implementing it then. | 09:06 |
| alekibango | thats it :) | 09:06 |
| jonwood | There's not such thing as a generic SLA. | 09:06 |
| alekibango | heh, thats why i said hard to define | 09:06 |
| alekibango | as it must be in correlation with scheduler in nova | 09:07 |
| alekibango | (which i might need to change first) | 09:07 |
| jonwood | The schdeuler wouldn't be enough I don't think. You need monitoring infrastructure to do SLAs in any useful way. | 09:07 |
| alekibango | yes, yes | 09:07 |
| alekibango | brb | 09:08 |
| zykes- | but doing a webui in turbogears is ok ? | 09:10 |
| *** ibarrera has quit IRC | 09:11 | |
| jonwood | zykes-: There's one already being built, so unless there's a definite advantage to using Turbogears, I don't think there's much point duplicating effort. | 09:12 |
| zykes- | ok | 09:12 |
| zykes- | so much for wanting to do that :p | 09:12 |
| jonwood | Contribute to the existing one. Even if it's just because Cappucino is fun to work with. | 09:13 |
| zykes- | don't want to have to learn another new language as well so | 09:13 |
| *** mgriffin has joined #openstack | 09:14 | |
| zykes- | enough learning python for the time being | 09:15 |
| jonwood | Fair enough. | 09:15 |
| zykes- | are there any simpler tasks that can be helped with ? | 09:17 |
| jonwood | I'm sure there's plenty to be done in the bug tracker. | 09:17 |
| jonwood | https://launchpad.net/nova/+milestone/low-hanging-fruit | 09:18 |
| jonwood | I'm assuming that's the list of bugs that are quick/easy fixes. | 09:18 |
| *** mgriffin has quit IRC | 09:18 | |
| *** sophiap_ has joined #openstack | 09:23 | |
| *** ibarrera has joined #openstack | 09:24 | |
| *** sophiap has quit IRC | 09:27 | |
| *** sophiap_ is now known as sophiap | 09:27 | |
| *** sophiap_ has joined #openstack | 09:30 | |
| *** sophiap has quit IRC | 09:33 | |
| *** sophiap_ is now known as sophiap | 09:33 | |
| zykes- | but the objective-c one, doesn't seem to have been updated jonwood since august ? | 09:36 |
| *** ArdRigh has quit IRC | 09:36 | |
| *** miclorb_ has quit IRC | 09:38 | |
| jonwood | It may still be being worked on. You'd have to ask the author. | 09:42 |
| alekibango | zykes-: for me, turbogears is really rapid development which is allowing you to have freedom. its not pressing too many 'framework' methods on you | 09:55 |
| alekibango | zykes-: tg is just bunch of very good libraries packed together to provide you powerfull methods to deal with the problem... | 09:56 |
| alekibango | but, i must say that it might be better to help existing one, as jonwood says | 09:57 |
| alekibango | zykes-: what we need most now is to write nova manuals | 09:58 |
| alekibango | if you can help with this, you will also learn a lot | 09:58 |
| *** Ryan_Lane has quit IRC | 09:59 | |
| zykes- | alekibango: but i don't know a squat of objectiveJ and not wanting to either due to little time availability. But I could help with manuals / wiki | 10:01 |
| zykes- | any overview of the process ? | 10:01 |
| alekibango | zykes-: spector* is working on one | 10:01 |
| alekibango | eh, on overview of nova architecture | 10:01 |
| alekibango | doc process - see wiki recent changes http://wiki.openstack.org/RecentChanges or you might look how it started on http://etherpad.openstack.org/ep/pad/view/NovaDocsNotes/latest | 10:04 |
| zul | soren: get rid of the default kernel and ramdisks imho | 10:05 |
| *** ar1 has quit IRC | 10:06 | |
| *** miclorb_ has joined #openstack | 10:09 | |
| zykes- | alekibango: won't learn python there though :p | 10:13 |
| *** Ryan_Lane has joined #openstack | 10:21 | |
| alekibango | zykes-: :) how long are u with python? | 10:27 |
| alekibango | zykes-: one of learning methods is "reading code and understanding it" | 10:28 |
| zykes- | been reading the pep once and doing some tutorial stuff | 10:28 |
| alekibango | and its also good method to make docs :) | 10:28 |
| alekibango | zykes-: good. now u know python :) | 10:29 |
| zykes- | not that much :p | 10:29 |
| zykes- | the basics at least | 10:29 |
| alekibango | but its enough to READ code | 10:29 |
| alekibango | and change it here and there | 10:29 |
| alekibango | it takes some time to learn about some advanced stuff, but most of it is about using libraries | 10:30 |
| alekibango | and about style | 10:30 |
| alekibango | (how to do logging, tests, etc) | 10:30 |
| zykes- | import logging | 10:31 |
| alekibango | zykes-: there are something like config files you know :) - thats that style i mean -- configure logging in .conf files | 10:32 |
| alekibango | lol messed the sentence but you can dig it i hope | 10:32 |
| alekibango | and using config files.... | 10:33 |
| alekibango | using decorators | 10:33 |
| alekibango | etc | 10:33 |
| alekibango | zykes-: try some coding with decorators | 10:33 |
| alekibango | i once implemented smart caching with those, it was lots of fun, having pretty statistics | 10:34 |
| jonwood | zykes-: Do you know any other programming languages? | 10:34 |
| alekibango | zykes-: it takes some time before you will find out those little tricks. -- and its best done by reading good code | 10:36 |
| zykes- | jonwood: Perl.. | 10:36 |
| alekibango | :) | 10:36 |
| jonwood | Ok. Just checking you're not learning programming from scratch, rather then just a new language | 10:36 |
| alekibango | i also did perl in the past :) | 10:37 |
| alekibango | was fun intil the code scaled up to thousands lines | 10:37 |
| zykes- | js also, though a scripting language | 10:37 |
| *** miclorb_ has quit IRC | 10:43 | |
| *** coolhandluke has joined #openstack | 10:48 | |
| zykes- | but again alekibango, writing technical documentation how does that help on python skills ? not sure where to star.t. | 10:49 |
| *** miclorb has joined #openstack | 10:49 | |
| alekibango | zykes-: try to install nova on 4 servers :) | 10:50 |
| alekibango | try to understand the architecture - how it works | 10:50 |
| alekibango | and describe what you will find on wiki | 10:50 |
| zykes- | document the install process | 10:51 |
| zykes- | i guess ? | 10:51 |
| alekibango | yes | 10:52 |
| alekibango | and the architecture | 10:52 |
| alekibango | and limits | 10:52 |
| alekibango | and potential security threats | 10:52 |
| zykes- | ok | 10:52 |
| coolhandluke | hi.. we are trying to do a multi node nova installation but its failing with the following error : | 10:52 |
| coolhandluke | File "/usr/lib/pymodules/python2.6/nova/db/sqlalchemy/api.py", line 1296, in user_get | 10:52 |
| coolhandluke | 2010-10-15 16:15:10+0530 [-] raise exception.NotFound('No user for id %s' % id) | 10:52 |
| coolhandluke | 2010-10-15 16:15:10+0530 [-] NotFound: No user for id admin | 10:52 |
| zykes- | i'll try it when i get to my other job then, no access to play with stuff here | 10:52 |
| alekibango | coolhandluke: please paste longer texts on http://paste.openstack.org | 10:53 |
| *** miclorb has quit IRC | 10:53 | |
| jonwood | coolhandluke: Have you created a new user using nava-admin? | 10:53 |
| alekibango | nova-manage you mean? | 10:53 |
| coolhandluke | alekibango: noted. | 10:53 |
| jonwood | Yes, nova-manage | 10:53 |
| coolhandluke | we've already done that. nova-manage user admin admin | 10:53 |
| alekibango | coolhandluke: if you can write some short tutor about that, it would be great | 10:54 |
| alekibango | user create ? | 10:54 |
| jonwood | coolhandluke: Have you configured everything to use the right database? | 10:55 |
| coolhandluke | alekibango: will do. as soon as i get this to work :) . I've verified that the user was created by doing nova-manage user list | 10:55 |
| jonwood | The default is to use SQLite on the current node, which won't work for multi-node installs. | 10:55 |
| coolhandluke | jonwood: right | 10:55 |
| alekibango | coolhandluke: please do, it should be part of nova Deployment manual | 10:55 |
| coolhandluke | jonwood: we've setup myssql as suggested by you yesterday | 10:55 |
| jonwood | Ok. | 10:56 |
| alekibango | are the nodes configured to connect to it? | 10:56 |
| jonwood | When you verified that the user was created, were you doing so from the same node that's throwing that error? | 10:56 |
| coolhandluke | alekibango: yes. I've set the connection param in sql_connection.. only problem is that there are no records in the 'users' table in nova database. and the api seems to be expecting a user record there | 10:57 |
| coolhandluke | jonwood : yes | 10:58 |
| alekibango | that seems to be teh problem | 10:58 |
| alekibango | creating user using nova-manage will not make him in db? | 10:58 |
| coolhandluke | alekibango: thats what I thought. it should create the user in the db as well. but it seems the user is being registered only in ldap. | 11:00 |
| jonwood | coolhandluke: One second. | 11:00 |
| jonwood | I think you have the wrong auth driver activated. | 11:00 |
| coolhandluke | interstingly, at the end of the stack trace I also see this error: http://paste.openstack.org/show/57/ | 11:02 |
| jonwood | coolhandluke: Try setting --auth_driver=nova.auth.dbdriver.DbDriver | 11:03 |
| jonwood | If I'm right that'll set it to do all authentication against the database, rather then LDAP. | 11:03 |
| jonwood | If I'm wrong, I'm out of ideas I'm afraid. | 11:03 |
| coolhandluke | ok. I'll give that a go and report back | 11:03 |
| *** ctennis has quit IRC | 11:06 | |
| *** DubLo7 has joined #openstack | 11:12 | |
| *** DubLo7 has quit IRC | 11:25 | |
| *** aimon has quit IRC | 11:33 | |
| *** aimon has joined #openstack | 11:33 | |
| *** mtaylor has quit IRC | 11:34 | |
| alekibango | coolhandluke: you are using ldap? | 11:44 |
| alekibango | ah, jonwood, maybe the user is in redis | 11:44 |
| alekibango | :) | 11:44 |
| alekibango | try monitoring network (wireshark) when you add user | 11:44 |
| coolhandluke | quick update.. jonwood: your suggestion helped. we have now got the node running fine. there were some issues with enabling KVM in bios. we've sorted out those now. currently thinking why we're not able to ssh into the VM.. perhaps it needs NAT + masquerading | 11:45 |
| jonwood | coolhandluke: Excellent. | 11:46 |
| jonwood | And it probably is an iptables issue. Do you specifically need your VMs on private IPs? | 11:46 |
| jonwood | I think you can set nova-network to use bridging instead, which entirely bypasses NAT problems. | 11:47 |
| coolhandluke | I'll try that. do you know how to configure nova-network.. does it also have a corresponding .conf flagfile | 11:48 |
| *** faccenda has joined #openstack | 11:49 | |
| jonwood | Probably... I've not done anything with it. | 11:49 |
| *** iammartian_ has joined #openstack | 11:50 | |
| jonwood | coolhandluke: It's not the friendliest --help interface, but running 'ack "flags." in the root of a nova checkout will give you all the command line flags defined. | 11:51 |
| *** BK_man has joined #openstack | 11:53 | |
| *** gustavomzw has joined #openstack | 11:54 | |
| coolhandluke | jonwood: thanks. I couldnt figure that out simply by looking at the nova-network script. I'll try the flags thing | 11:54 |
| jonwood | It looks like you need to set the network manager to FlatManager, but I'm not sure how you go about doing that. | 11:55 |
| jonwood | Aha... --network_manage=nova.network.manager.FlatManager might work. | 11:55 |
| jonwood | Sorry, --network-manager | 11:55 |
| coolhandluke | awesome :) trying that now | 11:55 |
| jonwood | Let me know how it goes - it's an educated guess ;) | 11:56 |
| uvirtbot | New bug: #661159 in nova "Does not allow booting without a ramdisk" [Undecided,New] https://launchpad.net/bugs/661159 | 12:02 |
| dendrobates | morning | 12:06 |
| dendrobates | er afternoon to some | 12:06 |
| coolhandluke | , and evening to another few :)) | 12:08 |
| dendrobates | coolhandluke: where are you located? | 12:09 |
| jonwood | afternoon :) | 12:09 |
| *** ctennis has joined #openstack | 12:11 | |
| *** ctennis has joined #openstack | 12:11 | |
| coolhandluke | jonwood: i tried setting --network_manager=nova.network.manager.FlatManager , on the compute node as well as the controller.. on the controller I got an error "FlatManager has no attribute allocate_network" | 12:11 |
| jonwood | Hmmm. I wonder if something is trying to create a private network for the project. | 12:12 |
| coolhandluke | jonwood: probably. I'll report back with findings soon. on a related note, our team will start creating documentation for the multi node install starting monday. Whats the best way to share that with everyone? | 12:15 |
| jonwood | Put it on the wiki. | 12:15 |
| zykes- | can i test stuff without requiring hypervisor enabled nodes ? | 12:15 |
| coolhandluke | jonwood: ok. I'll do that and notify you. thanks as usual for your help. | 12:16 |
| jonwood | Yup. There's a test driver for just that purpose. | 12:16 |
| zykes- | the uml jonwood ? | 12:16 |
| jonwood | No, not even that. | 12:17 |
| jonwood | It just pretends things worked or didn't, depending on what you've set it to. | 12:17 |
| *** sophiap has quit IRC | 12:17 | |
| jonwood | So databases and things like that get updated, but no actual VMs get started. | 12:17 |
| *** coolhandluke has quit IRC | 12:17 | |
| jonwood | zykes-: --connection=fake | 12:19 |
| dendrobates | uml works great. it's very fast | 12:19 |
| soren | dendrobates: So, about bug 661159 | 12:19 |
| uvirtbot | Launchpad bug 661159 in nova "Does not allow booting without a ramdisk" [Medium,In progress] https://launchpad.net/bugs/661159 | 12:19 |
| soren | dendrobates: Ubuntu's images don't have a ramdisk. | 12:20 |
| soren | dendrobates: It just seems silly not to support Ubuntu's images. | 12:20 |
| alekibango | jonwood: ack flags? ??? | 12:20 |
| dendrobates | soren: we couldn't create a fake ramdisk? | 12:21 |
| soren | dendrobates: Dunno. | 12:21 |
| jonwood | alekibango: Yes, it uses ack to find all instances of "flags." in or below the current directory. | 12:21 |
| alekibango | i do not have command ack :) | 12:21 |
| jonwood | Then you may need to install it first. | 12:21 |
| alekibango | kanji convertor? | 12:22 |
| dendrobates | basically, a workaround seems less risky at this point | 12:22 |
| alekibango | :) | 12:22 |
| jonwood | alekibango: What? | 12:22 |
| alekibango | http://packages.debian.org/lenny/ack | 12:23 |
| alekibango | jonwood: where did you get your ack? | 12:23 |
| jonwood | Try ack-grep | 12:23 |
| jonwood | brew install ack | 12:23 |
| alekibango | ah, ic | 12:23 |
| alekibango | ty | 12:23 |
| *** westmaas has joined #openstack | 12:23 | |
| jonwood | But then I'm using mac os | 12:23 |
| soren | dendrobates: I can't argue against that. | 12:25 |
| alekibango | jonwood: so then its ack-grep command for me, but rgrep would work on most systems | 12:25 |
| alekibango | but i must say nice colors | 12:25 |
| dendrobates | soren: or better yet let's let ubuntu make an image for us. I know the guy who wrote their virt-builder app. :) | 12:27 |
| soren | dendrobates: It's the sort of argument that can't be beaten, really. I can only go: "Bah, I don't think it'll break anything," and if it does I'm an idiot. If it doesn't it's just meh. | 12:27 |
| soren | I want to be a release manager when I grow up. Then I'll replace myself with a shell script that just says no. :) | 12:28 |
| alekibango | soren: ;-) | 12:28 |
| dendrobates | I don;t want to be a release manager. | 12:28 |
| soren | dendrobates: These are the kernel images, I'm afraid :) vmbuilder doesn't factor into it. | 12:29 |
| dendrobates | ah | 12:29 |
| alekibango | i think running disk only images (without kernel etc) would be sweet also | 12:29 |
| dendrobates | well we still know the guy who builds them | 12:29 |
| arthurc | you are talking about ramdisks.. are the nodes can boot and run diskless ? | 12:29 |
| soren | dendrobates: ...but they're not going to travel back in time and change this for Maverick. | 12:31 |
| alekibango | arthurc: nodes? i would not do that yet... | 12:31 |
| soren | dendrobates: This stuff is pretty fundamental. As such, it seems to me, that if it breaks something, it'll be quite obvious, quite quickly. | 12:32 |
| alekibango | it can be done but i requires some sort of unionfs or some sort of network fs or some sort of hackity hacks... | 12:32 |
| alekibango | and none of them is very safe for your health, arthurc | 12:32 |
| soren | dendrobates: I'm going to test it in a little bit. Right now, it's only been unit tested (you'll notice I didn't file an mp for it yet, for that very reason). | 12:32 |
| dendrobates | soren: I'd rather have you testing and fixing release critical bugs. | 12:33 |
| arthurc | yep. I used to manage my cloud without installing or configuring my nodes, just rack a lame, plug wires and I've got my new node (conf is managed centrally) | 12:33 |
| alekibango | arthurc: for start, it will work much better if you have such install method (fai or foreman or whatever) and configuration management tools | 12:34 |
| soren | dendrobates: Do you have a list of those? | 12:34 |
| alekibango | as network can get 100% used and then you are in all sorts of problems with your servers | 12:34 |
| alekibango | (when using nfs like system) | 12:34 |
| dendrobates | I keep trying, but people are fixing them as they report them. :) | 12:34 |
| alekibango | or your unionfs will find some nice bugs (like it does) | 12:34 |
| soren | dendrobates: Fire them. | 12:34 |
| soren | dendrobates: :) | 12:34 |
| alekibango | or your unflexible images will get old | 12:34 |
| dendrobates | that makes an up to date list difficult to make | 12:34 |
| arthurc | cobbler does the job for me alekibango | 12:35 |
| jonwood | alekibango: In the nicest possible way, what are you talking about? | 12:35 |
| alekibango | arthurc: i have seen cobbler... didnt like it much :) | 12:35 |
| soren | dendrobates: I have an rc bug. Maybe. I need to make sure it exists before I file it. If it does, I'll be a good boy and file it before I start working on it. | 12:35 |
| *** DubLo7 has joined #openstack | 12:35 | |
| alekibango | jonwood: trying to answer arthurc -- he is talking about diskless nodes | 12:35 |
| alekibango | but i think its too risky or unflexible | 12:36 |
| jonwood | Why? It's been done in production since the 80s | 12:36 |
| arthurc | it's very practical to be able to do that | 12:36 |
| alekibango | well might be, but risky :) | 12:36 |
| arthurc | why ? | 12:36 |
| *** sophiap has joined #openstack | 12:37 | |
| alekibango | will u use some sort of unionfs or aufs ? | 12:37 |
| *** jdarcy has joined #openstack | 12:37 | |
| alekibango | will you use network disks? | 12:37 |
| dendrobates | diskless vm's would require a ramdisk and in multi-tenant cloud environments memory is a scarce resource | 12:37 |
| alekibango | network bandwidth etc | 12:37 |
| alekibango | dendrobates: or network drive, and that is also scarce | 12:37 |
| alekibango | network WILL get used much | 12:38 |
| jonwood | alekibango: There's not a great shortage of bandwidth within a well run data centre. | 12:38 |
| arthurc | alekibango: vms use iscsi storage | 12:38 |
| alekibango | arthurc: price and complexity will go up, not down :) | 12:38 |
| arthurc | we hosts hundred of vms for a year with this system | 12:40 |
| alekibango | arthurc: can you draw the schema of the system for me please? | 12:40 |
| alekibango | :) | 12:41 |
| arthurc | https://fedorahosted.org/ovirt/wiki/Architecture | 12:42 |
| alekibango | ah, ty | 12:42 |
| alekibango | i have been reading about ovirt in the past | 12:42 |
| arthurc | well, I'm a maintainer of the project | 12:42 |
| arthurc | but I'm not from RH | 12:43 |
| alekibango | great | 12:43 |
| alekibango | the link looks inspirational. we need similar images for nova | 12:43 |
| arthurc | yep, I looked for them but I the best schema I found is there http://nova.openstack.org/architecture.html | 12:45 |
| alekibango | :) | 12:45 |
| alekibango | there are 2 more links to see about arch | 12:45 |
| alekibango | http://wiki.openstack.org/ArchitecturalOverview http://wiki.openstack.org/Overview | 12:45 |
| dendrobates | awesome | 12:46 |
| alekibango | arthurc: how fast your storage netwrok is? | 12:50 |
| arthurc | gigabyte network and vms with virtio | 12:51 |
| alekibango | VMS? | 12:52 |
| arthurc | virtual machines | 12:52 |
| alekibango | VAX/VMS? :) | 12:52 |
| alekibango | is that gigabyte network enough? | 12:52 |
| alekibango | do you somehow limit CPU or network or disk access for guests? | 12:52 |
| arthurc | well, we don't need extrem performances but we recently had to use virtio and we like to use jumbo frames but it needs some hacks | 12:53 |
| *** BK_man has quit IRC | 12:54 | |
| alekibango | virtio is good | 12:54 |
| arthurc | we allocate ram and cpu as we want | 12:54 |
| alekibango | never was into jumbos | 12:54 |
| alekibango | arthurc: as we want? what that means? | 12:54 |
| alekibango | as user wants? | 12:54 |
| arthurc | nop, as the admin want, as the client pays :) | 12:55 |
| alekibango | so no limits for disk or netwrok? | 12:56 |
| alekibango | (not talking about quota,. but about bandwidth) | 12:56 |
| arthurc | it's much more the work of kvm than ovirt the ressources allocation. storage is set up externally and network is not limited | 12:56 |
| alekibango | ic | 12:57 |
| arthurc | I don't known if kvm manage this feature | 12:57 |
| alekibango | kvm didnt even limit cpu power :) | 12:57 |
| alekibango | you must use normal linux tools for that | 12:57 |
| alekibango | this certainly is doable.... but methods differ | 12:58 |
| alekibango | you can limit network per user, and having cgroups + maybe some patches you can limit also disk access per user or process | 12:58 |
| *** mgriffin has joined #openstack | 12:59 | |
| alekibango | arthurc: i propose to move this talk into time after the release. i will try to help with this, but now most devels are busy with fixing and merging, and we should do docs | 13:00 |
| alekibango | after release there will be call for blueprints | 13:01 |
| alekibango | so we could come up with some systems to build bootable images of diskless nodes... be it debian or fedora system | 13:03 |
| alekibango | or others | 13:03 |
| *** mgriffin has quit IRC | 13:03 | |
| jonwood | alekibango: Why? There's already cobbler, and several other things, which do that quite happily. | 13:04 |
| alekibango | there is also debian live :) and others | 13:04 |
| alekibango | jonwood: it would be nice to document even more ways of doing it | 13:05 |
| alekibango | having common architecture | 13:05 |
| arthurc | alekibango: no problem, I don't have the time neither to test openstack but I'll bench the product soon | 13:06 |
| alekibango | we all will try ;) | 13:07 |
| arthurc | But it's a pleasure to see how you're all enthusiastic, it's a good sign :) | 13:08 |
| alekibango | ;-) | 13:11 |
| *** sirp1 has quit IRC | 13:14 | |
| *** khussein has joined #openstack | 13:16 | |
| uvirtbot | New bug: #661214 in nova "When a node dies, its instances should be marked !running" [Undecided,New] https://launchpad.net/bugs/661214 | 13:16 |
| *** BK_man has joined #openstack | 13:16 | |
| *** sirp1 has joined #openstack | 13:19 | |
| *** aliguori has joined #openstack | 13:25 | |
| dendrobates | soren: re: bug #661214 should they be marked unknown instead of !running? | 13:26 |
| uvirtbot | Launchpad bug 661214 in nova "When a node dies, its instances should be marked !running" [Undecided,New] https://launchpad.net/bugs/661214 | 13:26 |
| dendrobates | jaypipes: around? | 13:29 |
| soren | dendrobates: unknown is one of the !running states. It sounds like the right one. | 13:31 |
| dendrobates | soren: where is that code? | 13:34 |
| soren | dendrobates: Eh? | 13:35 |
| soren | which code? | 13:35 |
| dendrobates | where the !running states are defined? | 13:35 |
| soren | Oh. Not sure. | 13:36 |
| soren | I thought you knew when you suggested "unknown" :) | 13:36 |
| soren | lemme check | 13:36 |
| dendrobates | power_state.py perhaps | 13:37 |
| *** gundlach has joined #openstack | 13:38 | |
| dendrobates | models.py | 13:38 |
| dendrobates | soren as you know I am obsessed with state and how/where we keep it. | 13:41 |
| soren | It's an interesting problem! | 13:41 |
| soren | I found a delightful refresher of the CAP theorem the other day. | 13:42 |
| soren | http://codahale.com/you-cant-sacrifice-partition-tolerance/ | 13:42 |
| *** gundlach has quit IRC | 13:44 | |
| * annegentle loves to see the Monday Nova doc plans, we'll be doing something similar for Swift on Tuesday | 13:47 | |
| *** sirp1 has quit IRC | 13:47 | |
| *** f4m8 is now known as f4m8_ | 13:49 | |
| *** pharkmillups has joined #openstack | 13:49 | |
| *** abecc has joined #openstack | 13:52 | |
| *** sirp1 has joined #openstack | 13:53 | |
| *** abecc has joined #openstack | 13:54 | |
| dendrobates | annegentle: where are these plans? | 13:56 |
| annegentle | dendrobates: just saw coolhandluke say his team is creating doc starting Monday | 13:57 |
| dendrobates | oh, cool | 13:58 |
| dendrobates | I thought I missed something on the wiki | 13:58 |
| piken | Does anyone know of a good api for working with the ec2 services. I am time restricted and don't think I will have time to right an api my self. | 13:59 |
| piken | something in php, python, or c++ would work if any one knows one. | 13:59 |
| jonwood | piken: libcloud I think is Python. | 13:59 |
| soren | It is. | 13:59 |
| soren | So is boto. | 13:59 |
| *** pharkmillups has quit IRC | 14:00 | |
| dendrobates | annegentle: our nova install instructions are only for the dev snapshot. I changed the heading to indicate that. | 14:02 |
| annegentle | dendrobates: yep, I saw that. Good call. | 14:04 |
| *** Podilarius has joined #openstack | 14:06 | |
| piken | hmm, libcloud might work for me. Thanks jon | 14:08 |
| *** kashyapc has quit IRC | 14:10 | |
| *** kevnfx has joined #openstack | 14:12 | |
| *** ko-iida has joined #openstack | 14:14 | |
| *** schisamo has joined #openstack | 14:16 | |
| *** kevnfx has quit IRC | 14:16 | |
| dendrobates | annegentle: we should expand the page with sections for installing on different linux distros when we have time. | 14:17 |
| annegentle | dendrobates: yeah I put a section at the bottom in anticipation | 14:17 |
| *** ko-iida has left #openstack | 14:24 | |
| *** dendrobates is now known as dendro-afk | 14:25 | |
| *** dendro-afk is now known as dendrobates | 14:26 | |
| zykes- | dendrobates: installing on centos etc you mean ? | 14:26 |
| dendrobates | I mean any distros that we can run on, probably ubuntu and fedora to start with | 14:27 |
| *** iammartian_ has quit IRC | 14:29 | |
| *** Cybodog has quit IRC | 14:30 | |
| *** Cybodog has joined #openstack | 14:31 | |
| *** Cybodog has joined #openstack | 14:33 | |
| *** ccustine has joined #openstack | 14:34 | |
| *** Cybodog has quit IRC | 14:34 | |
| *** khussein has quit IRC | 14:34 | |
| *** Cybodog has joined #openstack | 14:36 | |
| *** tonywolf has joined #openstack | 14:37 | |
| Daviey | piken: also look at txAWS | 14:39 |
| *** Cybodog has left #openstack | 14:40 | |
| *** Cybodog has joined #openstack | 14:41 | |
| *** Cybodog has quit IRC | 14:47 | |
| *** Cybodog has joined #openstack | 14:48 | |
| uvirtbot | New bug: #661260 in nova "nova-compute doesn't notice when a VM crashes" [Undecided,New] https://launchpad.net/bugs/661260 | 14:51 |
| uvirtbot | New bug: #661262 in nova "nova-compute doesn't update VM status on start" [Undecided,New] https://launchpad.net/bugs/661262 | 14:51 |
| *** ibarrera has quit IRC | 14:52 | |
| *** rnirmal has joined #openstack | 14:56 | |
| *** westmaas has quit IRC | 14:56 | |
| uvirtbot | New bug: #661267 in swift "document config options and defaults in sample configs" [Medium,New] https://launchpad.net/bugs/661267 | 14:56 |
| *** DubLo7 has quit IRC | 14:57 | |
| *** guigui has quit IRC | 15:00 | |
| *** Cybodog has quit IRC | 15:02 | |
| *** Cybodog has joined #openstack | 15:02 | |
| *** Cybo has joined #openstack | 15:02 | |
| jaypipes | dendrobates: yup. | 15:06 |
| *** silassewell has joined #openstack | 15:07 | |
| jaypipes | dendrobates: sorry, getting a late start today :( is there something you needed? | 15:10 |
| *** calavera has quit IRC | 15:19 | |
| jaypipes | annegentle: may be worth adding Ryan Lucio's answers/question to the docs on Nova networking... https://answers.launchpad.net/nova/+question/129521 | 15:24 |
| *** mmalgeri has quit IRC | 15:24 | |
| *** hisaki has joined #openstack | 15:31 | |
| annegentle | jaypipes: ah good one, will do | 15:32 |
| dendrobates | jaypipes: yeah could you or sirp1 write a blog post about glance? It is looking nice and I want to publicize it. | 15:39 |
| *** kevnfx has joined #openstack | 15:45 | |
| jaypipes | dendrobates: hmm, yes, I can do that next week? | 15:47 |
| *** silassewell has quit IRC | 15:49 | |
| *** DubLo7 has joined #openstack | 15:51 | |
| *** mdomsch has joined #openstack | 15:52 | |
| *** westmaas has joined #openstack | 15:54 | |
| annegentle | Can someone simply describe how you create flag files for nova? Where are they located/read from, what permissions do they need, etc? | 15:55 |
| annegentle | and if I need one flag for three services, can they be in one file? | 15:56 |
| *** perestrelka has quit IRC | 16:00 | |
| *** dendrobates is now known as dendro-afk | 16:00 | |
| *** ppetraki has joined #openstack | 16:01 | |
| annegentle | ah, is the flag file nova-network.conf and I just keep adding to that? | 16:04 |
| *** cloudmeat has joined #openstack | 16:04 | |
| *** mmalgeri has joined #openstack | 16:04 | |
| jonwood | annegentle: With a package install, each service gets it's own file. nova-network.conf for nova-network, and so forth. | 16:04 |
| jonwood | And yes, you just add any flags you want used to that file. | 16:04 |
| annegentle | jonwood: got it, thanks! | 16:05 |
| *** cloudmeat1 has quit IRC | 16:07 | |
| *** cloudmeat1 has joined #openstack | 16:07 | |
| *** Cybodog has quit IRC | 16:09 | |
| *** cloudmeat has quit IRC | 16:10 | |
| kevnfx | vishy: tried those changes ... still having same/similar problem (http://pastebin.com/p7DLKpkF) ... this, while invoking "./nova-manage network create 10.0.0.0/8 3 16" | 16:18 |
| vishy | kevinfx: ah | 16:20 |
| vishy | one more change missing | 16:20 |
| *** HouseAway is now known as AimanA | 16:22 | |
| *** dendro-afk is now known as dendrobates | 16:28 | |
| kevnfx | vishy: just in nova-manage correct? Made the change there and it's much better. | 16:34 |
| vishy | yes | 16:35 |
| vishy | create_networks on 410 | 16:35 |
| vishy | None should be context.get_admin_context() | 16:35 |
| *** metoikos has quit IRC | 16:35 | |
| vishy | i'll put in a patch | 16:35 |
| vishy | but that change should get you going | 16:35 |
| kevnfx | vishy: got it ... now having problems again with virDomainLookupByName ... I'll need to take a look at that (thought that was fixed earlier). | 16:37 |
| *** cloudmeat has joined #openstack | 16:40 | |
| vishy | that should be | 16:41 |
| vishy | perhaps a regression due to my backporting | 16:41 |
| vishy | stacktrace? | 16:42 |
| *** cloudmeat1 has quit IRC | 16:44 | |
| jdarcy | How would one get a concise list of the blueprints/topics/whatever being considered for summit? | 16:45 |
| *** ccustine has quit IRC | 16:48 | |
| kevnfx | vishy: well ... I'm seeing an instance created with id 2147483647 (-1) ... are there other instance_id changes "in the works"? | 16:49 |
| *** maplebed has joined #openstack | 16:49 | |
| kevnfx | s/instance/internal/ | 16:50 |
| vishy | hmm | 16:50 |
| vishy | packing 64 bit number into 32 bits? | 16:51 |
| vishy | kevnfx: I haven't really looked at that code, maybe gundlach has an idea what is going on there | 16:51 |
| *** hisaki has quit IRC | 16:52 | |
| annegentle | jdarcy: good question, hopefully dendrobates can answer? | 16:52 |
| vishy | kevnfx: i can't track down that bug at the moment, but if no one has tackled it by this evening I'll take a look | 16:52 |
| *** zaitcev has joined #openstack | 16:53 | |
| kevnfx | vishy: that's cool .... I'll try to dig into it as well (although I'll be considerably slower). | 16:53 |
| *** silassewell has joined #openstack | 16:54 | |
| vishy | kevnfx: if you come accross any more Exception('die') it is due to an empty context being passed to the db | 16:54 |
| vishy | I turned into a die instead of a deprecation to force us to clean them all up | 16:55 |
| vishy | so let me know if you hit any more | 16:55 |
| *** mtaylor has joined #openstack | 17:02 | |
| *** ChanServ sets mode: +v mtaylor | 17:02 | |
| *** schisamo has quit IRC | 17:02 | |
| *** westmaas has quit IRC | 17:03 | |
| *** BK_man has quit IRC | 17:05 | |
| zaitcev | https://blueprints.launchpad.net/swift says "Launchpad does not know how OpenStack Object Storage (swift) tracks feature planning or documentation. " -- nice | 17:06 |
| *** mmalgeri has quit IRC | 17:08 | |
| jdarcy | https://blueprints.launchpad.net/openstack seems to work and include Swift as a project. | 17:09 |
| *** DubLo7 has quit IRC | 17:09 | |
| notmyname | I think that's the default "there are no blueprints right now" page | 17:09 |
| jdarcy | Still trying to figure out what Glance and Teller are. | 17:09 |
| *** cdub has joined #openstack | 17:11 | |
| alekibango | jdarcy: (cold) storage and registry of images for nova in using swift | 17:11 |
| zaitcev | ok, I'm going with this, then https://blueprints.launchpad.net/swift/+spec/s3-hail | 17:11 |
| sirp1 | Glance is the super-project, it contains Parallax (the image registry) and Teller (the image delivery service) | 17:12 |
| alekibango | s/ in // in my line :) | 17:13 |
| *** rlucio has joined #openstack | 17:14 | |
| alekibango | jdarcy: swift is not for live (real-time) disk service, but it can store objects well. | 17:14 |
| jdarcy | Sounds like Teller (and Glance more generally) do some of the same things as the Image Warehouse (that Pete and I work on). Might be an opportunity for sharing code/ideas. | 17:15 |
| alekibango | Image Warehouse, how nice name. sounds like Galery for me :) | 17:16 |
| dendrobates | zaitcev: try it again. lp must have reset something during the update | 17:16 |
| zaitcev | dendrobates: Oooh, it works, thanks. | 17:16 |
| jdarcy | It's been through a few names. It might go through a few more. | 17:16 |
| alekibango | jdarcy: please tell us more... | 17:16 |
| jdarcy | alekibango: Basically image storage with a metadata DB (currently Mongo) plus multiple data back ends (currently filesystem/S3/Swift) with a way to define very flexible replication policies between them. | 17:19 |
| alekibango | sounds interesting to me. but i am not glance dev | 17:19 |
| jdarcy | alekibango: It's why you see me more here than in #gluster nowadays. ;) | 17:20 |
| alekibango | :) | 17:20 |
| alekibango | jdarcy: now you know what i wanted to use glusterfs for :) | 17:21 |
| dendrobates | jdarcy: how is mongodb working for you? | 17:21 |
| jdarcy | dendrobates: OK I guess. Efficient queries (not just ranges) are pretty important, so that narrows the field a lot. | 17:22 |
| jdarcy | dendrobates: In some ways I'd prefer CouchDB, Cassandra, or Riak/EnterpriseDS, but it's easy to change. | 17:22 |
| dendrobates | we will need to make those same decisions | 17:23 |
| *** daleolds has joined #openstack | 17:23 | |
| *** jonwood has quit IRC | 17:23 | |
| jdarcy | That sounds like a good thing to discuss over a beer or three in San Antonio. | 17:24 |
| alekibango | there is no beer like ours (cz) sometimes come here to taste it... :) | 17:24 |
| jdarcy | alekibango: If I ever find an excuse to visit our Brno office, I will. ;) | 17:25 |
| jaypipes | eday: hmm, looks like lp660818 introduced a bug: http://paste.openstack.org/show/58/ | 17:25 |
| alekibango | jdarcy: brq office is biggest, btw | 17:25 |
| alekibango | tell me if you will be here | 17:25 |
| alekibango | :) | 17:25 |
| jdarcy | Will do. | 17:25 |
| jaypipes | jdarcy: where can I see the code for Image Warehouse? | 17:27 |
| zaitcev | jaypipes: http://repo.or.cz/w/iwhd.git | 17:28 |
| jdarcy | Beat me to it. | 17:28 |
| alekibango | jdarcy: i am glad you started replicating process :) | 17:29 |
| jdarcy | Redundant Array of Inexpensive Engineers | 17:30 |
| zaitcev | Sadly I am very expensive. | 17:30 |
| alekibango | hehe | 17:30 |
| jdarcy | zaitcev: Yeah, I've been meaning to talk to you about that. | 17:30 |
| alekibango | :) | 17:32 |
| alekibango | anyway, i hope you can get enough help to do what is needed | 17:32 |
| jaypipes | jdarcy: you guys work for RedHat? | 17:32 |
| * jdarcy nods. | 17:32 | |
| jaypipes | jdarcy: when did this Image Warehouse project come to be? Recently? /me wondering why you went with C as opposed to Python... | 17:33 |
| jdarcy | A lot of people wonder that. | 17:33 |
| zaitcev | Good question. | 17:33 |
| * jaypipes also wonders what MHD is.... | 17:33 | |
| zaitcev | It's GNU microhttpd | 17:33 |
| jaypipes | ah. hadn't heard of it. | 17:34 |
| jdarcy | I proposed re-doing it in higher-level language (probably Python) a while back when it was still very young, but got a lukewarm response so we went with existing code. | 17:34 |
| jaypipes | understood. | 17:34 |
| *** jc_smith has joined #openstack | 17:34 | |
| jdarcy | I might do a version in Lua in my spare time, just to annoy *everyone*. | 17:35 |
| alekibango | lol | 17:35 |
| alekibango | jdarcy: postscript is more deadly for sure | 17:35 |
| alekibango | did you know you can program it in? :) | 17:35 |
| jaypipes | jdarcy: is this project just in internal use at RH? | 17:36 |
| alekibango | (dont try) | 17:36 |
| jdarcy | jaypipes: It's part of the Deltacloud project. | 17:36 |
| zaitcev | I think Hail was written in C because 1. "C is faster" fallacy, 2. jgarzik is a kernel graduate so it was faster for him to hack it together in C than to learn Python. | 17:36 |
| zaitcev | But I was surprised to find that iwhd was in C too | 17:37 |
| zaitcev | Thank heavens it's not Ruby at least | 17:37 |
| jdarcy | To be more precise, "Deltacloud" is now an Apache project, but there are some new higher-level components (Cloud Engine, Image Factory) that are still at Red Hat. We might need to change names. | 17:37 |
| jaypipes | jdarcy: I'd never heard of DeltaCloud...looking now... | 17:37 |
| alekibango | its cloud abstracting api | 17:38 |
| zaitcev | I always thought of DeltaCloud as API and a library that implements it, kinda like KDE's Phonon | 17:38 |
| zaitcev | But it grew into this huge chunk of cloud management everything | 17:38 |
| jdarcy | The main reason I started with C is that I knew where to find all the libraries I need (e.g. HTTP, S3, parsing) and I couldn't quickly find a framework in a better language that let me do in-memory forking of the data streams. | 17:39 |
| dendrobates | jaypipes: really? | 17:39 |
| alekibango | .... er ... lang? | 17:39 |
| jaypipes | dendrobates: yep, never heard of it. | 17:39 |
| jdarcy | The official Deltacloud line (I believe) is that it's not a cloud-management system itself, it's a layer above several cloud-management systems. Use whatever you want with a single interface. | 17:39 |
| jaypipes | jdarcy: I find it strange that the deltacloud API is GPL and the project is in Apache incubator. | 17:40 |
| alekibango | so, it can be used to manage nova, right? | 17:40 |
| alekibango | how good that could be? | 17:40 |
| jaypipes | dendrobates: remember, I come from a rdbms background, not a virt background :) | 17:41 |
| dendrobates | the main thing I questioned about deltacloud is the use of condor as a scheduler | 17:41 |
| jdarcy | alekibango: Yep. Not sure what the status is of supporting Nova as a provider (and probably couldn't say if I did know) but Swift is definitely a supported back end on the warehouse side. | 17:41 |
| alekibango | jdarcy: i am sure nova will be supported soon, as it will be important player :) | 17:41 |
| jdarcy | In fact, I'm more than half-way inclined to make Swift the *preferred* iwhd back end, but that's a political rather than technical matter. | 17:41 |
| zaitcev | How's that political? | 17:42 |
| zaitcev | I just don't have a Swift server running anywhere, so it's not very testable. | 17:42 |
| creiht | zaitcev: btw, i started a proof of concept s3 compatibility middleware for swift | 17:42 |
| jdarcy | zaitcev: market/community positioning, which companies/projects we want to align most closely with or avoid offending, all that kind of BS. | 17:43 |
| alekibango | creiht: yes, give them the link hehe | 17:43 |
| jaypipes | zaitcev: but it looks like deltacloud *is* Ruby: http://deltacloud.org/drivers.html#h1 | 17:43 |
| jdarcy | The Engine is Ruby. I think maybe the Factory too, but not sure on that one. | 17:43 |
| creiht | im on my phone right now, but can post later | 17:43 |
| jdarcy | It might be interesting to think about using iwhd as a Teller back end, or maybe even vice versa. | 17:45 |
| *** coredump|br has quit IRC | 17:47 | |
| *** maplebed has quit IRC | 17:55 | |
| jaypipes | vishy: dude, awesome writeup about Nova networking. annegentle, did you catch that on the mailing list? | 17:57 |
| eday | jaypipes: yeah, not seeing it locally, and can't imagine how it would cause the failure we see. wondering if hudson machine runs have a stale db or something... | 17:57 |
| rlucio | jaypipes: link? | 17:57 |
| jaypipes | eday: not sure. lemme pull your branch and try it | 17:57 |
| eday | jaypipes: yeah, termie's branch hit the same thing | 17:57 |
| jaypipes | rlucio: for the netowrking writeup? :) | 17:57 |
| rlucio | jaypipes: ya | 17:58 |
| jaypipes | rlucio: one sec | 17:58 |
| vishy | jaypipes: thanks | 17:58 |
| eday | jaypipes: I'll clean the nhudson workspace and try again | 17:58 |
| jaypipes | eday: sounds good | 17:58 |
| jaypipes | rlucio: https://lists.launchpad.net/nova/msg00299.html | 17:58 |
| *** schisamo has joined #openstack | 17:59 | |
| annegentle | jaypipes: sure did, and I'm totally rolling it in on the wiki. | 17:59 |
| * annegentle needs to get out some ninja graphics skillz | 17:59 | |
| rlucio | jaypipes: thanks | 18:00 |
| *** coredump|br has joined #openstack | 18:00 | |
| *** Pentheus has quit IRC | 18:02 | |
| *** Ryan_Lane is now known as Ryan_Lane|food | 18:03 | |
| *** metoikos has joined #openstack | 18:03 | |
| eday | jaypipes: hmm, so tarmac must do it's own thing, no files in workspace | 18:03 |
| *** ppetraki has quit IRC | 18:05 | |
| *** kevnfx has quit IRC | 18:05 | |
| *** sophiap has quit IRC | 18:07 | |
| *** mtaylor has quit IRC | 18:08 | |
| eday | jaypipes: you have access to the tarmac machine? | 18:08 |
| *** jc_smith has quit IRC | 18:11 | |
| *** mtaylor has joined #openstack | 18:11 | |
| *** ChanServ sets mode: +v mtaylor | 18:11 | |
| *** dendrobates is now known as dendro-afk | 18:14 | |
| *** johnbergoon has joined #openstack | 18:15 | |
| eday | mtaylor: it appears test runs are failing with nova tarmac, is there a workspace/tmp files that can be cleaned out? | 18:26 |
| mtaylor | eday: jay had me add a nosetest run - lemme look real quick | 18:27 |
| eday | mtaylor: also, you in boston? :) (I'm on the train from NYC to Boston right now) | 18:27 |
| mtaylor | eday: can you point me at a link? | 18:28 |
| mtaylor | eday: I'm on the plane on my way to boston | 18:28 |
| mtaylor | eday: should be there in about 2 hours-ish | 18:28 |
| eday | https://code.edge.launchpad.net/~eday/nova/lp660818/+merge/38471 | 18:28 |
| eday | ahh, cool. we get there at 5 or so (Wendy is with me) | 18:28 |
| mtaylor | cool | 18:29 |
| eday | mtaylor: another unrelated branch by termie failed with the same thing | 18:29 |
| eday | so I'm assuming it's a broken test setup | 18:29 |
| mtaylor | so - I don't know enough about what the test cases do - is there something I should look at cleaning up in /tmp or something? (hudson blows away the branch dir each time) | 18:30 |
| mtaylor | and if there is - that should be fixed | 18:30 |
| mtaylor | eday: there area fuck-ton of test-* and tmp* dirs in /tmp | 18:30 |
| * mtaylor thinks tests are being ugly and are not cleaning up after themselves | 18:30 | |
| mtaylor | and for that matter, honestly shouldn't be making things in /tmp - but /me is a fairly pedantic person | 18:31 |
| * eday agrees :) | 18:32 | |
| mtaylor | eday: ok - I deleted the over a thousand temp dirs | 18:32 |
| mtaylor | there is still a /tmp/pip-QlLJp9-build and a /tmp/nova | 18:33 |
| *** jc_smith has joined #openstack | 18:35 | |
| eday | mtaylor: I'm guessing there are leftover sqlite files somewhere? not really sure | 18:36 |
| *** jc_smith_ has joined #openstack | 18:37 | |
| *** jc_smith has quit IRC | 18:37 | |
| *** jc_smith_ is now known as jc_smith | 18:37 | |
| mtaylor | eday: launch another one | 18:37 |
| *** dendro-afk is now known as dendrobates | 18:39 | |
| *** sophiap has joined #openstack | 18:44 | |
| soren | eday, mtaylor: What's in Boston? | 18:45 |
| *** westmaas has joined #openstack | 18:46 | |
| mtaylor | soren: opensqlcamp | 18:47 |
| soren | mtaylor: Ah. | 18:51 |
| soren | mtaylor, eday: Have fun :) | 18:51 |
| soren | I'm going to look at the get_console_ouput test case that keeps failing. | 18:52 |
| eday | mtaylor, soren: trying again now | 18:53 |
| soren | eday: Which one? | 18:53 |
| * soren keeps an eye on it | 18:53 | |
| mtaylor | stewart: awake? | 18:54 |
| eday | (lost connection for a while, in middle of nowhere Connecticut) | 18:54 |
| mtaylor | gah. wrong channel | 18:54 |
| eday | mtaylor: he's here too :) | 18:55 |
| mtaylor | indeed | 18:55 |
| soren | eday: failed. | 18:56 |
| soren | eday: I'll look into it. | 18:56 |
| soren | it's my unit test after all :) | 18:56 |
| eday | soren: ahh, ok :) | 18:56 |
| eday | soren: I was about to dig deeper, but it passes fine on my local box, and I don't have tarmac machine access, so can't reproduce | 18:57 |
| soren | eday: I just hope it reproduces easily on there. | 18:57 |
| dendrobates | is tarmac running on the same server as hudson? | 18:58 |
| soren | Yes. | 18:58 |
| soren | mtaylor: Which directory do the unit tests run in? | 18:59 |
| mtaylor | soren: /home/hudson/src/nova/hudson | 18:59 |
| mtaylor | soren: sorry - | 18:59 |
| soren | \o/ reproduced. | 18:59 |
| * soren starts digging. | 18:59 | |
| mtaylor | soren: ~hudson/src/nova/hudson | 18:59 |
| soren | mtaylor: Ah. ta. | 18:59 |
| mtaylor | ~hudson == /var/lib/hudson :) | 18:59 |
| soren | mtaylor: Gotcha. | 19:00 |
| *** dysinger has joined #openstack | 19:10 | |
| *** silassewell has quit IRC | 19:10 | |
| *** mgriffin has joined #openstack | 19:14 | |
| *** hazmat has quit IRC | 19:14 | |
| soren | Oh! | 19:15 |
| *** mgriffin has quit IRC | 19:18 | |
| *** mtaylor has quit IRC | 19:22 | |
| soren | I have a hideous, wrong patch that totally fixes it :) | 19:26 |
| eday | soren: why did this start breaking with unrelated changes? | 19:28 |
| eday | can someone with more LP access switch https://blueprints.edge.launchpad.net/nova/+spec/datastore-abstraction to implemented? | 19:29 |
| jaypipes | eday: ya, one sec | 19:29 |
| jaypipes | eday: oops, no I can't...thought I could..but no. | 19:30 |
| jaypipes | dendrobates: see eday's request ^^ | 19:30 |
| jaypipes | eday: did monty get that instance id error fixed on hudson? | 19:30 |
| eday | jaypipes: I think soren almost has it fixed | 19:30 |
| jaypipes | eday: ah, k | 19:31 |
| *** dendrobates is now known as dendro-afk | 19:36 | |
| *** dendro-afk is now known as dendrobates | 19:36 | |
| soren | eday: It's a race. | 19:37 |
| soren | Not sure how to fix it correctly. | 19:38 |
| soren | :( | 19:38 |
| *** mtaylor has joined #openstack | 19:38 | |
| *** ChanServ sets mode: +v mtaylor | 19:38 | |
| soren | Screw it, I'll hack it, add a TODO so we can get on with it. | 19:38 |
| *** schisamo has quit IRC | 19:39 | |
| * soren tries the tests a couple of times to make sure. | 19:40 | |
| soren | Oh! | 19:41 |
| eday | soren: ahh. strange it's started showing up more since the last push, hasn't passed since | 19:41 |
| soren | There are two problems, as it turns out. | 19:41 |
| soren | One is the get_console_output test. | 19:41 |
| soren | It's a race. | 19:41 |
| soren | It happens 90% of the times, roughly. | 19:42 |
| soren | The other thing is due to datastore.py being removed, but the .pyc file still being there. | 19:42 |
| soren | ..so a file that imported it still worked until someone cleaned out the .pyc files. | 19:42 |
| soren | Both are my fault :) | 19:42 |
| eday | hehe | 19:42 |
| jaypipes | soren: damn you soren :) | 19:43 |
| *** Pentheus has joined #openstack | 19:43 | |
| *** silassewell has joined #openstack | 19:45 | |
| soren | Here I am, trying to give you guys an excuse to call it a week early, but no no :) | 19:45 |
| eday | soren: about 30 mins left on the train :) | 19:46 |
| soren | I guess I need to actually fix them both in the same merge proposal.. | 19:46 |
| soren | Even though they're completely unrelated. | 19:46 |
| soren | Bah. | 19:46 |
| soren | branch and merge proposal coming up. | 19:50 |
| *** dysinger has quit IRC | 19:50 | |
| *** hazmat has joined #openstack | 19:51 | |
| soren | eday, jaypipes: https://code.launchpad.net/~soren/nova/fix-unit-tests/+merge/38587 if you please. | 19:52 |
| soren | (had to lump them together in one branch, otherwise tarmac would reject them due to the other unit test still failing) | 19:53 |
| jaypipes | soren: understood. looking now. | 19:54 |
| soren | mtaylor: We probably ought to "bzr clean-tree --unknown --ignored --force" to the run_tests.sh thing on tarmac. | 19:54 |
| Ryan_Lane|food | on lucid, nova-manage doesn't seem to have a network option. has that been removed? | 19:55 |
| soren | Ryan_Lane|food: Not been added yet, rather. | 19:55 |
| *** Ryan_Lane|food is now known as Ryan_Lane | 19:55 | |
| Ryan_Lane | ah | 19:55 |
| Ryan_Lane | should I be targeting maverick instead? | 19:55 |
| soren | Both should be fine. | 19:55 |
| Ryan_Lane | do I need to add the networks manually for now? | 19:56 |
| mtaylor | soren: well, tarmac is _supposed_ to be blowing the dir away and starting fresh each time... | 19:56 |
| mtaylor | soren: lemme check on that | 19:56 |
| *** Podilarius has quit IRC | 19:58 | |
| soren | mtaylor: Oh, really? | 19:58 |
| eday | soren: ick, I don't like the sleep, but it'll do for now | 19:58 |
| soren | eday: The alternatives are: | 19:59 |
| soren | 1) fix eventlet+rpc in unit tests | 19:59 |
| soren | 2) Ignore the failure to find the instance in the network manager. | 19:59 |
| soren | 1) would be ideal, but I'm afraid to do something like that at this point in the cycle. | 20:00 |
| soren | I'm not sure about 2). | 20:00 |
| soren | It was about to say it'd never happen in the real world, but I guess it could. | 20:00 |
| soren | If a user starts an instance and terminates it before the network manager gets a chance to assign it an ip, this would happen. | 20:01 |
| *** ppetraki has joined #openstack | 20:01 | |
| soren | He'd have to be pretty darn quick (or have discovered a pattern in the instance id's being handed out). | 20:01 |
| *** Cybo has quit IRC | 20:03 | |
| dendrobates | soren: hmm it seems like #1 is out. | 20:03 |
| soren | dendrobates: I already spent a whole day trying to do that. | 20:04 |
| soren | dendrobates: last week some time. I have no expectation to get it done today, at least. | 20:04 |
| *** Cybodog has joined #openstack | 20:04 | |
| dendrobates | If this is impossible or rare enough, we can ship with it and fix it in bexar | 20:05 |
| dendrobates | I don't completely understand yet | 20:05 |
| soren | If it were to happen in the real world, it wouldn't matter much. | 20:05 |
| soren | An exception would be logged. Nothing else. | 20:05 |
| soren | It's not like it's a security vulnerability. | 20:06 |
| mtaylor | soren: I have made a config change for tarmac | 20:06 |
| mtaylor | soren: now it should be creating a new tmpdir each time and doing everything in there | 20:07 |
| dendrobates | then I say we punt on it. We should include it as a caveat in the release notes | 20:07 |
| soren | Noone will ever see it. | 20:08 |
| soren | If someone sees it, I'll buy them beer. | 20:08 |
| * mtaylor sees it | 20:08 | |
| * soren calls mtaylor's bluff | 20:09 | |
| dendrobates | I'll buy you both a beer. | 20:10 |
| eday | (back, damn tunnels) | 20:11 |
| eday | soren: option 1 is the obivous one, but yes, I agree not for now/austin | 20:12 |
| *** joearnold has joined #openstack | 20:12 | |
| *** ctennis has quit IRC | 20:15 | |
| *** joearnold has quit IRC | 20:15 | |
| *** joearnold has joined #openstack | 20:16 | |
| soren | Does anyone know why nova-api manages to survive starting up without rabbit being around? | 20:16 |
| eday | soren: probably because it doesn't touch rabbit until it needs to call/cast | 20:21 |
| soren | eday: But it does. | 20:21 |
| eday | soren: oh? for what? | 20:21 |
| soren | eday: First thing it does it to regist... Oh, hang on. | 20:21 |
| soren | So why does everything else fail if rabbit isn't around... | 20:22 |
| soren | Oh, api doesn't have a consumer. | 20:23 |
| soren | Hahah! | 20:23 |
| soren | Ok. | 20:23 |
| soren | Hm.. | 20:23 |
| eday | soren: it only has temp consumers for rpc.call | 20:23 |
| soren | eday: Right. That's totally it. | 20:23 |
| soren | Ok. | 20:23 |
| soren | Hmm.. | 20:23 |
| soren | eday: Thanks for the hint. | 20:24 |
| soren | I guess now isn't a good time to replace carrot with txamqp... :) | 20:26 |
| *** dendrobates is now known as dendro-afk | 20:28 | |
| *** ctennis has joined #openstack | 20:31 | |
| eday | soren: txamqp won't work well with eventlet :) | 20:32 |
| mtaylor | wow. I sure did just fly into northeast clouds... | 20:32 |
| mtaylor | and it was pretty out west... | 20:33 |
| soren | eday: I don't see why not, really. | 20:33 |
| *** BK_man has joined #openstack | 20:37 | |
| *** mtaylor has quit IRC | 20:39 | |
| eday | soren: but plus I'm still not convinced the workers really need the complexities of twisted... but that's a summit discussion :) | 20:39 |
| eday | soren: perhaps txamqp will, I'm just judging from issues I've seen with event+twisted | 20:41 |
| *** gustavomzw has quit IRC | 20:41 | |
| pquerna | i think you mean, rewrite all the twisted stuff in node.js | 20:42 |
| eday | pquerna: umm, no. that would make even less sense | 20:42 |
| eday | :) | 20:42 |
| pquerna | programing language and frameowrk choice is not about "sense" or "logic" | 20:43 |
| pquerna | its about feeling | 20:43 |
| pquerna | and hot damn node.js is sexy, get with it | 20:43 |
| soren | We've got our own kool-aid :) | 20:44 |
| annegentle | Alrighty, wrote up step-by-step of Nova on Maverick on http://wiki.openstack.org/NovaInstall. Feel free to edit, remix, whatnot, and thank you Soren for the walkthrough on your blog. | 20:45 |
| soren | annegentle: My pleasure. | 20:45 |
| eday | pquerna: haha | 20:46 |
| jaypipes | annegentle: nice job :) | 20:48 |
| *** Cybodog has quit IRC | 20:52 | |
| *** westmaas has quit IRC | 20:56 | |
| soren | bug 661472 <--- Darn it, was hoping to fix this today | 21:04 |
| uvirtbot | Launchpad bug 661472 in nova "Fails if rabbitmq isn't around or drops connection" [Undecided,New] https://launchpad.net/bugs/661472 | 21:04 |
| pquerna | soren: is that using txamqp? | 21:06 |
| *** jc_smith_ has joined #openstack | 21:08 | |
| *** DubLo7 has joined #openstack | 21:08 | |
| *** jc_smith has quit IRC | 21:09 | |
| *** jc_smith_ is now known as jc_smith | 21:09 | |
| *** kevnfx has joined #openstack | 21:10 | |
| uvirtbot | New bug: #661472 in nova "Fails if rabbitmq isn't around or drops connection" [Undecided,New] https://launchpad.net/bugs/661472 | 21:12 |
| soren | pquerna: No. | 21:12 |
| soren | pquerna: python-carrot, wrapper in various ways. | 21:12 |
| pquerna | soren: hmm. i need to dig out of my hole and contribute some stuff to openstack; but cloudkick has a pile of code for both txamqp and carrot to do reconnects/reliable/multi-master stuff against rabbit. | 21:16 |
| *** daleolds has quit IRC | 21:16 | |
| *** daleolds has joined #openstack | 21:20 | |
| *** jdarcy has quit IRC | 21:21 | |
| *** kevnfx has quit IRC | 21:21 | |
| *** rnirmal has quit IRC | 21:22 | |
| soren | pquerna: That would be totally awesome. | 21:28 |
| soren | Ok, one more bug for today, then I'm out. | 21:28 |
| vishy | is the build fixed again so we can merge? | 21:29 |
| soren | vishy: Yes. | 21:29 |
| vishy | soren: rockin | 21:29 |
| *** allsystemsarego has quit IRC | 21:30 | |
| *** kevnfx has joined #openstack | 21:33 | |
| *** dendro-afk is now known as dendrobates | 21:35 | |
| *** hazmat has quit IRC | 21:35 | |
| *** joearnold has quit IRC | 21:46 | |
| Ryan_Lane | are the packages for maverick more up to date then lucid? | 21:46 |
| Ryan_Lane | and for bug reporting purposes, should I be running something specific? | 21:47 |
| dendrobates | Ryan_Lane : you should be running from trunk | 21:47 |
| dendrobates | use vish's script for ease | 21:47 |
| soren | Ryan_Lane: For any reasonable project, daily built packages should be fine. We're not quite that reasonable yet :) | 21:48 |
| Ryan_Lane | heh | 21:48 |
| Ryan_Lane | I just don't want to report on solved problems :) | 21:48 |
| dendrobates | we are far less reasonable | 21:48 |
| Ryan_Lane | at some point I am going to need to be running a stable version though. | 21:48 |
| dendrobates | that is why I love spinning up cloud instances to test openstack. It is very quick to blow it away and start over | 21:49 |
| dendrobates | which I do 15 times a day | 21:50 |
| Ryan_Lane | heh | 21:50 |
| dendrobates | plus it is free for me | 21:50 |
| soren | Ryan_Lane: Sure, we just haven't made a stable release yet. Next week, mate. Next week :) | 21:50 |
| Ryan_Lane | soren: I'm not rushing you guys ;) | 21:50 |
| Ryan_Lane | just want to know the proper way to help right now | 21:50 |
| *** joearnold has joined #openstack | 21:54 | |
| *** mdomsch has quit IRC | 21:58 | |
| *** metoikos has quit IRC | 22:02 | |
| *** metoikos has joined #openstack | 22:03 | |
| *** iammartian has joined #openstack | 22:06 | |
| * soren calls it a week | 22:10 | |
| *** RamD has quit IRC | 22:13 | |
| *** abecc has quit IRC | 22:14 | |
| *** iammartian_ has joined #openstack | 22:22 | |
| *** dendrobates is now known as dendro-afk | 22:24 | |
| *** ppetraki has quit IRC | 22:28 | |
| *** AimanA is now known as HouseAway | 22:31 | |
| *** Cybodog has joined #openstack | 22:49 | |
| *** jakedahn has quit IRC | 22:51 | |
| *** jakedahn has joined #openstack | 22:51 | |
| *** aliguori has quit IRC | 22:59 | |
| *** mmalgeri has joined #openstack | 23:04 | |
| *** mmalgeri has quit IRC | 23:06 | |
| *** kevnfx has quit IRC | 23:08 | |
| *** jakedahn has quit IRC | 23:16 | |
| *** faccenda has quit IRC | 23:18 | |
| *** faccenda has joined #openstack | 23:18 | |
| *** jonwood has joined #openstack | 23:20 | |
| *** jonwood has quit IRC | 23:34 | |
| *** jc_smith has quit IRC | 23:48 | |
| *** tonywolf has left #openstack | 23:48 | |
| *** rlucio has quit IRC | 23:55 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!