17:01:09 <jaypipes> #startmeeting
17:01:10 <openstack> Meeting started Thu Mar 22 17:01:09 2012 UTC.  The chair is jaypipes. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:01:11 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:01:33 <jaypipes> dwalleck: done getting lunch?
17:01:38 <anotherjesse> dwalleck: that doesn't mean we can't try harder to have a default based on introspecting the environment
17:01:41 <dwalleck> yup! I'm back
17:01:57 <dwalleck> anotherjesse: Not a problem, that makes total sense.
17:02:06 <jaypipes> OK, so couple major items...
17:02:20 <jaypipes> 1) does anyone know what happened to Chris Fattarsi?
17:02:33 <dwalleck> not I
17:02:35 <jaypipes> He was working on the Keystone tests but seems to have vanished.
17:02:51 <jaypipes> and it's to the point now that I'm about to just take over the work...
17:03:02 <dwalleck> yeah, I halted our submission because he had started
17:03:17 <jaypipes> dwalleck: I think we need to just push ahead...
17:03:27 <jaypipes> Keystone is a critical functional test area..
17:03:30 <dwalleck> Probably not a bad idea
17:03:32 <jaypipes> davidkranz_: afternoon
17:03:37 <davidkranz_> Yo.
17:03:45 <jaypipes> dwalleck: OK, you have a lot on your plate already...
17:03:49 <dwalleck> davidkranz_: howdy
17:03:59 <Carlos_Swift> hello room
17:04:00 <jaypipes> dwalleck: do you have someone on your team who can do the keystone work?
17:04:04 <davidkranz_> I can't even get keystone to work!
17:04:05 <jaypipes> Carlos_Swift: heyo :)
17:04:15 <jaypipes> davidkranz_: well, that's another story ;_)
17:04:23 <dwalleck> jaypipes: That's an understatement :) I do. He had some code to merge, but chris beat him to the punch
17:04:57 <dwalleck> But I can either have him pick up those branches, submit his branches, or work with you to get them into a good state
17:05:08 <dwalleck> Whatever is the easiest and/or makes the most sense
17:05:33 <jaypipes> dwalleck: submit it ASAP in whatever state...
17:05:44 <jaypipes> dwalleck: if it gets in before chris' work, so be it.
17:05:51 <dwalleck> jaypipes: you got it boss
17:05:55 <jaypipes> dwalleck: but we need to start iterating on the keystone stuff ASAP.
17:05:57 <jaypipes> dwalleck: thx mate
17:06:03 <dwalleck> jaypipes: agreed
17:06:21 <jaypipes> dwalleck: as for the Quantum tests, you are working on unabandoning those, right?
17:06:36 <dwalleck> jaypipes: Right. I just need to make some minor fixes
17:06:41 <jaypipes> dwalleck: k.
17:07:07 <jaypipes> I will write an email to Ravi about Sapan's branches. They are long abandoned, even though they have review comments to address
17:07:24 <jaypipes> so hopefully Ravi can have someone else pick those up if Sapan cannot do them.
17:07:41 <jaypipes> davidkranz_: approved the updates to the stress tests from Adam G.
17:07:48 <jaypipes> davidkranz_: they should be merged in trunk now.
17:07:53 <davidkranz_> jaypipes: thanks.
17:08:00 <jaypipes> np
17:08:15 <davidkranz_> As far as I can tell, there are no complete instructions for keystone/essex  that work anywhere. Are there? Or at least some kind of code one can read other than devstack which uses this template catalog stuff.
17:08:35 <jaypipes> davidkranz_: yes, that is correct (but being worked on)
17:08:58 <jaypipes> davidkranz_: devstack is a source of info on setting it up, but as you say, it uses the templated catalog.
17:09:06 <jaypipes> whereas the ubuntu packages use the SQL backends
17:09:16 <davidkranz_> I am pretty sure I am only one or two bugs away from having it working bug debugging is tricky.
17:09:37 <anotherjesse> are you hoping to get coverage of both styles?
17:09:40 <jaypipes> davidkranz_: that's good to hear.
17:09:58 <davidkranz_> Is there an easy way to watch the exchange between nova and keyhstone that happens when you do 'nova list'?
17:10:21 <jaypipes> anotherjesse: not really a matter of something that tempest controls :) we just point the API calls at an environment. If that environment is devstack, it will stress the template backend, if the env is ubuntu, it will stress the SQL backend :)
17:10:42 <adam_g> davidkranz_: other than tailing logs on either end, none that i can think of. someone mentioned using ngrep for this on the ML the other da
17:11:11 <jaypipes> davidkranz_: I have screen open on my devstack install box and run tempest from another machine. having a look at the n-api and k-api screens will show you the communication.
17:11:55 <jaypipes> OK, so let's move on to the next topic...
17:12:05 <jaypipes> #topic Jenkins job(s) firing Tempest
17:12:22 <jaypipes> OK, so I put in a first stab at this a couple days ago.
17:12:37 <jaypipes> jeblair gave me some heads up on the code review and I'm going to try again today.
17:12:52 <dwalleck> nice
17:13:18 * jeblair is very excited about this
17:13:45 <jaypipes> it involves changes to the devstack-gate subproject in the openstack-ci repo (https://github.com/openstack-ci/devstack-gate)
17:13:49 * anotherjesse too
17:14:00 <adam_g> is this going to be another pre-commit gating job?
17:14:12 <jaypipes> I kinda borked the original merge proposal but hope to have another one submitted today to jeblair
17:14:21 <anotherjesse> adam_g: I think this is a job that runs, not a gate?
17:14:33 <anotherjesse> at least at the begininng
17:14:35 <jaypipes> adam_g: eventually, perhaps a subset of Tempest will gate, yes, depending on how long the tests take
17:14:42 <jaypipes> anotherjesse: correct
17:14:52 <adam_g> i see
17:15:09 <jaypipes> adam_g: we are going to see how long the tests take, then try to get the performance better, and gradually include more and more of the tempest tests as a gate
17:15:14 <jaypipes> at least, that's the plan.
17:15:46 <adam_g> we're in the process of getting the same up and going here for post-commit testing we do in addition to devstack exercises - https://jenkins.qa.ubuntu.com/view/Precise%20OpenStack%20Testing/job/precise-openstack-essex-tempest/lastCompletedBuild/testReport/
17:15:51 <jaypipes> In other performance news, I ran tempest with --processes=8 on my "uberbox" and got the time down about 15%.
17:16:07 <dwalleck> That seems like a reasonable goal
17:16:19 <dprince> When I run Tempest... I just run the 'smoke' tests.
17:16:31 <jaypipes> Now, this isn't great, but it's a start, and the tests completed successfully, so we should be able to get some better parallelization with a little effort
17:16:32 <dprince> Running the entire suite takes *way* to long at this point.
17:16:41 <jaypipes> dprince: yes, agreed.
17:16:53 <dalang> jaypipes: thanks for looking review earlier, ill resubmit later today
17:17:01 <jaypipes> dalang: rock on.
17:17:09 <dwalleck> Well, I've had it running at 50 threads...it kind of screams :) But, that was using a different runner
17:17:17 <jaypipes> dalang: ah! you are Chris Fattarsi!
17:17:27 <dalang> indeed
17:17:34 <jaypipes> dwalleck: what different runner?
17:17:46 <jeblair> jaypipes: to make it easier to have two jobs, one that runs all of tempest, and one that won't (starting with not running tempest at all, and moving onto running part of tempest for gating)...
17:17:48 <dalang> i should probably make that easier to spot :)
17:18:05 <jeblair> jaypipes: i'd recommend adding an argument to the devstack-vm-gate.sh that controls whether/to what degree to run tempest
17:18:16 <jaypipes> jeblair: OK, will do.
17:18:23 <dwalleck> jaypipes: One of my devs has an experimental branch using py.test. It has some very slick, easy to configure multithreaded/distributed testing capabilities
17:18:41 <jaypipes> dwalleck: dalang == Chris Fattarsi. :) cancel the request about Keystone tests
17:18:48 <dwalleck> gotcha
17:18:50 <jaypipes> dwalleck: gotcha
17:18:54 <jaypipes> jinx.
17:18:58 <dwalleck> damnit
17:19:42 <dwalleck> Either way, performance is my next priority. I think we can pull a few easy tricks to make things pretty fast
17:19:49 <jaypipes> OK, so as far as tempest goes, today I'm just going to be doing that Jenkins job stuff and a blog article on running Tempest + devstack + RC1 Essex stuff to get folks to do more testing
17:19:56 <jaypipes> dwalleck: ++
17:20:09 <jaypipes> JoseSwiftQA: how's Swift tests going?
17:20:11 <dwalleck> Well performance + documentation + autoconfig + deep validation
17:20:43 <JoseSwiftQA> still a little ugle, need to update it to work better with auth 2.0
17:21:01 <jaypipes> JoseSwiftQA: gotcha. any ETA?
17:21:15 <JoseSwiftQA> that's a tough one
17:21:29 <jaypipes> JoseSwiftQA: any chance we can get a few tests added in for starters?
17:21:35 <jaypipes> JoseSwiftQA: and then build from there?
17:21:56 <JoseSwiftQA> I don't think that's a problem.  I had to throw away my old outdated branch, but that made me refactor a lot of bad code out
17:22:10 <JoseSwiftQA> so it's  alot closer to presentable now.
17:22:18 <jaypipes> k. well don't hesitate to grab us if you get into a rabbithole
17:22:24 <jaypipes> happy to assist
17:22:28 <JoseSwiftQA> will do, thanks
17:23:26 <jaypipes> OK folks, anything else to bring up today?
17:23:39 <jaypipes> #topic Open Discussion
17:24:07 <dwalleck> Design conference?
17:24:44 <jaypipes> dwalleck: ah..
17:24:52 <dwalleck> It'd be nice to sit together in a room and work through future plans
17:25:27 <jaypipes> I have one talk so far submitted for QA (from Ravi about test strategy)
17:26:01 <jaypipes> in the next week or two, I will submit 2 sessions -- one for the general overview of tempest and usage, and another for planning Folsom improvements.
17:26:17 <jaypipes> I encourage folks to submit talks on stuff that interests them.
17:26:25 <dwalleck> jaypipes: sounds great!
17:27:02 <dwalleck> I'll be finding something to talk about I'm sure. What is the process for submitting sessions?
17:27:16 * dwalleck still an OpenStack noob
17:27:17 <jaypipes> dwalleck: http://summit.openstack.org/
17:27:25 <jaypipes> dwalleck: should be a submit a session button.
17:27:44 <jaypipes> dwalleck: http://wiki.openstack.org/Summit
17:27:46 <jaypipes> #link http://wiki.openstack.org/Summit
17:27:58 <dwalleck> wow, that's quite a few sessions. Thanks!
17:28:32 <jaypipes> dwalleck: well there's close to 400 developers coming to the summit this time!
17:29:07 <dwalleck> It'll be interesting for sure. Looking forward to it
17:29:11 <jaypipes> ++
17:29:40 <jaypipes> OK y'all, I'm ready to end the meeting. Be prepared for me bugging people more and more for reviews and fixes over the next couple weeks :)
17:30:06 <jaypipes> I'd like our team to go into the design summit with a Jenkins job and some successful Tempest test runs under our belt.
17:30:16 <dwalleck> jaypipes: ++
17:30:26 <dwalleck> I'll get my nose back to the grindstone
17:30:31 <jaypipes> davidkranz_: would be good if you proposed a session about the stress tests at the DS (hint, hint) :)
17:30:44 <davidkranz_> jaypipes: Sure.
17:30:47 <jaypipes> alrighty, time to end. catch you all in the etherwebs.
17:30:51 <jaypipes> #endmeeting