16:04:41 <jaypipes> #startmeeting
16:04:42 <openstack> Meeting started Wed Nov 16 16:04:41 2011 UTC.  The chair is jaypipes. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:04:43 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:05:16 <jaypipes> I'd like to start today discussing the status of getting openstack-integration-tests gating trunk
16:05:23 <jaypipes> #topic status of getting openstack-integration-tests gating trunk
16:05:31 <jaypipes> mtaylor: quick update, sir?
16:05:37 <jaypipes> or jeblair
16:05:41 <jeblair> i can take it
16:05:45 <mtaylor> yay
16:05:51 <jeblair> very close now:
16:06:31 <jeblair> i have code that wraps the process, launching a vm, running devstack on it, exercises.sh for now, and destroying the vm
16:07:11 <jaypipes> jeblair: have you attempted runnign openstack-integratin-tests yet, and if so, any result?
16:07:38 <jeblair> there's a jenkins job set up to trigger on changes to any of _several_ projects, currently nova, glance, keystone, to run that.
16:07:44 <jeblair> not yet.  do you think it's ready?
16:08:11 <jeblair> basically, i figured you would let me know when you thought it was ready
16:08:23 <jaypipes> jeblair: yes, I think it is. At least, basic stuff is done
16:08:24 <jeblair> so i could work on getting the surrounding infrastructure in place
16:08:26 <westmaas> I think it could use a few tweaks first.
16:08:37 <jeblair> is there a particular way we should run it?
16:08:40 <westmaas> mostly because I don't think you can choose specific tests to run in the new code.
16:08:44 <jaypipes> westmaas: well, seeing the results of a first run would be good :)
16:08:54 <westmaas> thats fine too, just not to gate :)
16:08:58 <jeblair> we talked about it having a "--gate" option or something to run really solid tests
16:09:03 <jaypipes> westmaas: no, I think you can... --nova --glance, etc
16:09:10 <jeblair> so another thing we have going on:
16:09:21 <jeblair> is the jenkins job i have set up is currently running in "silent mode"
16:09:22 <westmaas> not for the new approach everyone is focusing in, the storm directory
16:09:27 <westmaas> on*
16:09:33 <jaypipes> jeblair: right. basically, it's adding a @attr('gate') to test cases, IIRC
16:09:40 <jeblair> meaning that it is triggered by gerrit, but doesn't report back or vote
16:09:45 <jaypipes> westmaas: ah, ok, sorry for confusion
16:10:04 <jeblair> so at any point, if we want to start experimenting with a new way of running the tests, we can set up another 'silent' jenkins job to do it
16:10:26 <jaypipes> westmaas, jeblair: OK, so sounds like there's a bit more work to do. Shall we aim to have some job running openstack-integration-tests by end of week?
16:10:33 <jaypipes> not gating, just running
16:10:39 <jeblair> i think we can do that
16:10:42 <westmaas> yep
16:10:44 <jaypipes> coolio.
16:11:04 <jeblair> i'll continue working on exercise.sh, and maybe start gating on it, and have a parallel job running integration silently
16:11:24 <jaypipes> awesome
16:11:26 <dwalleck> ack, time change, sorry about that
16:11:29 <jeblair> (not working on exercise.sh, actually but rather continue using it as a placeholder in gating)
16:11:30 <jaypipes> OK, moving on to more specific stuff about the integration tests...
16:11:36 <jaypipes> dwalleck: no worries, perfect timing
16:11:43 <jaypipes> jeblair: understood
16:12:09 <jaypipes> #topic westmaas and dwalleck give status report on what functional integration tests are top priority for adding at this moment.
16:12:29 <westmaas> dwalleck: you want to take this? either way is fine.
16:12:58 * jaypipes looking to have get a list of priority tests that need written, so we can delegate/assign them to folks
16:12:59 <dwalleck> I think we can both talk about priority. As far as the suite itself, we've been busy :)
16:13:12 <jaypipes> dwalleck: yes, very :) great stuff so far
16:13:36 <westmaas> sounds like the priority should be a set of tests that we feel we can and should gate on.
16:13:38 <dwalleck> Sounds reasonable. We have a smoke priority list internally we've been working from that we could put somewhere public
16:14:07 <westmaas> jaypipes: any suggestions for making tha public? I suggested just making bugs on the project itself, is there a better place?
16:14:13 <dwalleck> But I'm also thinking from a very nova-centric world, so I would like some thoughts to make sure we're covering everything we need to
16:15:00 <dwalleck> I'm okay with bugs also if that works. The only other thing I could think of would be a google docs spreadsheet
16:15:39 <jaypipes> westmaas: no, that's exactly what I would propose. bugs (or blueprints for large collections of tests) on openstack-qa project on Launchpad.
16:15:53 <jaypipes> westmaas, dwalleck: that way, we can target to milestones same as we do for core projects
16:15:56 <westmaas> I like bugs because it makes it easier to take individula tests
16:16:01 <jaypipes> yuppers
16:16:12 <dwalleck> jaypipes: Sounds good. I'll make that my priority today
16:16:26 <jaypipes> dwalleck: awesomesauce.
16:16:31 <westmaas> I have not followed up on what I was supposed to last week which was to make a poll to name this, which would then let us make a launchpad project.
16:16:44 <westmaas> will do it today, in progress now.
16:17:04 <jaypipes> westmaas: well, we can always make a  new LP project (under openstack-qa project or separately) and re-assign any bugs to that new project
16:17:15 <westmaas> jaypipes: sounds good
16:17:17 <jaypipes> westmaas: cheers, and lemme know if you need assistance. happy to help
16:17:35 <westmaas> jaypipes: can you make me less lazy?
16:17:40 <jaypipes> lol
16:17:43 <westmaas> would really help in a lot of ways
16:17:52 <jaypipes> OK, so nati isn't around, so we'll skip unit testing for now...
16:17:53 <dwalleck> For tracking purposes, should I also add bugs for test cases we've already finished?
16:18:05 <jaypipes> dwalleck: up to you.
16:18:07 <dwalleck> Or is this just a priority list going forward?
16:18:11 <dwalleck> gotcha
16:18:16 <westmaas> dwalleck: I'd rather you spend energy on other things, personally, but like jay said up to you
16:18:17 <jaypipes> dwalleck: priority is priority list going forward ;P
16:18:32 <dwalleck> I don't sleep. No problem :)
16:18:35 <westmaas> hah
16:18:36 <jaypipes> :)
16:18:47 <jaypipes> #topic HP QA cluster...
16:19:21 <jaypipes> Ravikumar_hp: OK, so looks like jeblair and mtaylor are making good progress on getting the devstack-based deployment jobs going in Jenkins
16:19:43 <mtaylor> yup
16:19:56 <jaypipes> Ravikumar_hp: after those are finished, I need to work with you on mapping out a plan to get a similar job, but using chef, going on the Jenkins server inside the HP QA cluster
16:20:15 <jaypipes> for those interested, HP has allocated 4 96GB servers for QA purposes
16:20:19 <Ravikumar_hp> our infrastructre team is in the process of movins server to public internet . I will set up OS
16:20:24 <jaypipes> right
16:20:35 <dwalleck> nice!
16:20:46 <Ravikumar_hp> There is a delay , but work is in progress
16:21:03 <mtaylor> great. so...
16:21:17 <jaypipes> anyway, Ravikumar_hp, the next logical step is going to be figuring out what is the latest chef deployment stuff. I think dprince is the man for that
16:21:26 <Ravikumar_hp> Nachi is working chef scripts for KVM based deploymen
16:21:40 <jaypipes> Ravikumar_hp: ah, good to know. Is that work on github?
16:21:47 <jeblair> there could be an opportunity to collaborate on openstack-chef
16:21:48 <mtaylor> we have a bare-metal deploy setup worked out that you can use to drive the chef
16:21:59 <jaypipes> Ravikumar_hp: also, is it based on github.com/openstack/openstack-chef
16:22:26 <jeblair> dprince is mostly driving the openstack-chef modules, but i'd love to see nachi and ravi collaborate on those if possible
16:22:28 <jaypipes> jeblair: ++ exactly my point :) want to de-duplicate effort as much as possible
16:22:37 <mtaylor> ++
16:22:49 <mtaylor> and they are managed in gerrit - so the process should be about the same as everything else
16:22:52 <Ravikumar_hp> ok. id dprince is working on that, we will sync with him
16:23:13 <jaypipes> Ravikumar_hp: OK, well you, me, and Gigi have a followup meeting. We can further discuss the chef stuff then, ok? Just wanted to let everyone else know what the plan was...
16:23:23 <Ravikumar_hp> sure
16:23:27 <jeblair> #link http://ci.openstack.org/jenkins.html#integration-testing
16:23:33 <mtaylor> jaypipes: lets circle up to on if we can use that ^^
16:23:46 <jaypipes> awesome
16:23:48 <mtaylor> to do all of the bits that get us to running chef on the hosts
16:23:58 <jaypipes> yep'
16:24:09 <jaypipes> dwalleck: OK, I have a special request for you :)
16:24:16 <dwalleck> uh oh :)
16:24:51 <jaypipes> dwalleck: so... to increase the number of folks writing integration/functional tests in the openstack-integration-tests project, it would be awesome to have a brief tutorial on writing test cases
16:24:52 <dwalleck> learn to not use parenths in python?
16:24:53 <jaypipes> dwalleck: I
16:25:15 <Ravikumar_hp> jaypipes: +1
16:25:20 <jaypipes> dwalleck: I'm happy to help with this, and I've (finally) gotten qa.openstack.org in the DNS system
16:25:21 <dwalleck> Of course! That sounds like a great idea. Where would you want me to put this?
16:25:33 <Ravikumar_hp> that can jump start my team contribution to functional test
16:25:51 <jaypipes> dwalleck: qa.openstack.org is going to be just like ci.openstack.org, which is RST docs that get built and pushed automagically to qa.openstack.org
16:25:58 <jaypipes> Ravikumar_hp: yep:)
16:26:20 <jaypipes> dwalleck: so, my goal is to have a number of RST docs that show folks what we're up to and how to contribute more tests
16:26:36 <jaypipes> dwalleck: shall the two of us set aside some time tomorrow to work on that?
16:26:39 <dwalleck> Sure, I definitely can do that
16:26:46 <dwalleck> Sounds like a good idea
16:26:51 <jaypipes> awesome, I'll email you a meeting request
16:27:10 * jaypipes apologizes for the long delay on getting qa.openstack.org up...
16:27:20 * mtaylor punches jaypipes in the face
16:27:28 <jaypipes> OK, so, let's wrap up with some open discussion
16:27:32 <jaypipes> #topic open discussion
16:27:42 <jaypipes> Anybody have stuff to bring up?
16:27:49 <westmaas> last call for test suite names!
16:28:04 <dwalleck> MightyMorphinTestinSuite!
16:28:06 <wwkeyboard> HurricaneHunter!
16:28:16 <mtaylor> let's call it jaypipes!
16:28:16 * jaypipes votes for "hurricane"
16:28:16 <westmaas> sorry guys you missed the window
16:28:24 <jaypipes> hehe
16:28:43 <jaypipes> dwalleck: just noticed your comment above about parenths in Python :) rofl
16:29:08 <dwalleck> I kick myself everything I see that. :) It's just second nature at this point
16:29:29 <jaypipes> dwalleck: *totally* understoood.
16:29:48 <jaypipes> OK, sounds like we're done. westmaas, I'll send the mailing list an update this week
16:29:54 <dwalleck> So one quick thing from me: so right now <fill_in the blank test suite> is using dictionaries to represent responses
16:29:57 <jaypipes> westmaas: so you can focus on the priority test list ;)
16:30:06 <westmaas> :)
16:30:28 <dwalleck> My team has wanted to change that to domain objects for equality and other functional purposes. Would anyone have any strong opposition to that idea?
16:30:35 <jaypipes> dwalleck: responses from the test case methods?
16:30:44 <jaypipes> dwalleck: not sure I'm following ya
16:30:57 <dwalleck> jaypipes: Sorry, responses to REST requests
16:31:11 <jaypipes> dwalleck: oh, you mean JSON
16:31:22 <jaypipes> or... ?
16:31:25 <dwalleck> So right now we just take the json response, convert it back to a dict and go on our merry way
16:31:34 <jaypipes> dwalleck: ah, yes
16:31:50 <dwalleck> We're thinking instead of passing that dict to a constructor to build an object representation of the response
16:32:05 <jaypipes> dwalleck: hmmm...
16:32:24 <jaypipes> dwalleck: so, my hesitation on that is that it may introduce some subtle bugs
16:32:25 <dwalleck> The work is technically done, but it's a big change. I wanted to see what the hivemind thought
16:33:09 <dwalleck> I agree. My condition would be that we would need to create some unit tests to verify that everything is working as expected
16:33:16 <jaypipes> dwalleck: but, that said, just propose it so we can see the code and comment... might be just what the doctor ordered :)
16:33:34 <dwalleck> Which we technically we need for the service methods too (one mountain at a time)
16:33:37 <dwalleck> Will do :)
16:33:39 <jaypipes> dwalleck: ya, sounds interesting. please do propose :)
16:33:46 <annegentle> o/
16:33:52 <jaypipes> annegentle: hi!
16:34:15 <annegentle> I've had a couple of people ask why they can't log bugs... and I can't remember what I had to do to be able to log bugs in Launchpad... is it permissions-based?
16:34:41 * annegentle gets lotsa OpenStack questions
16:34:53 <jaypipes> annegentle: anyone should be able to file a bug on any project
16:34:59 <jaypipes> annegentle: if they can't, that's a bug :)
16:35:31 <jaypipes> annegentle: now... to *target* a bug to a milestone or assign it to someone, you need to be a member of the driver and/or admin group, respectively
16:35:45 <annegentle> ok, literally they don't see the "Report a bug" with an arrow...
16:35:59 <jaypipes> annegentle: they DO need to be logged in to LP :)
16:36:33 <jaypipes> i.e. there's no support for anonymous bug filing
16:38:07 <annegentle> jaypipes: yep, he is... ok figured it out
16:38:29 <jaypipes> annegentle: cool :)
16:38:41 <jaypipes> OK, sounds like we're done with the meeting...
16:38:51 <annegentle> ok just for the record he kept going to groups instead of projects :)
16:38:58 <jaypipes> patelna: no worries, I'll be sending an update email out to the ML shortly.
16:39:00 <annegentle> can't log bugs against groups :)
16:39:05 <jaypipes> annegentle: aha!
16:39:09 <patelna> thanks
16:39:26 <mtaylor> we should be able to log bugs against groups :)
16:39:31 <patelna> just joined ...lost my time zone after I came back from India/DST
16:39:33 <jaypipes> patelna: any time! the daylight savings has caught lots of folks :)
16:39:47 <patelna> :)
16:39:56 <jaypipes> OK, till next time all!
16:39:59 <jaypipes> #endmeeting