17:00:39 <oomichi> #startmeeting qa
17:00:39 <openstack> Meeting started Thu Dec  8 17:00:39 2016 UTC and is due to finish in 60 minutes.  The chair is oomichi. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:40 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:00:43 <openstack> The meeting name has been set to 'qa'
17:00:50 <oomichi> Hello, who's here today?
17:00:58 <DavidPurcell> o/ hello
17:01:04 <jordanP> o/
17:01:10 <andreaf> o/
17:01:19 <rodrigods> o/
17:01:49 <oomichi> ok. lets get start
17:01:55 <oomichi> Today's agenda: https://wiki.openstack.org/wiki/Meetings/QATeamMeeting#Agenda_for_December_8th_2016_.281700_UTC.29
17:02:16 <oomichi> #topic Specs Reviews
17:02:29 <oomichi> #link https://review.openstack.org/#/q/status:open+project:openstack/qa-specs,n,z
17:02:51 <oomichi> The first one bug-notification-decorator is discussed on the patch
17:03:15 <oomichi> #link https://review.openstack.org/#/c/405781/
17:03:21 <oomichi> that seems good progress
17:03:30 <oomichi> happy if getting more feedback :)
17:03:59 <oomichi> The second one #link https://review.openstack.org/#/c/399618/ seems to get some feedback and need to update
17:04:06 <jordanP> yeah, I am not 100% convinced by the idea of that decorator
17:04:14 <jordanP> not sure it helps a lot
17:04:37 <jordanP> if someones notices a tempest test failure, then he should spend time and investigate the issue
17:04:44 <jordanP> git blame and such
17:05:13 <oomichi> jordanP: haha, yeah that is current way as the commit message
17:05:45 <oomichi> jordanP: but it is easy to imagine the situation they don't have git history on the cloud beacuse of installing it on package
17:06:24 <oomichi> the decorator is not so harmful for us, I feel
17:06:32 <jordanP> I guess they have a laptop and an internet connection
17:06:41 <jordanP> it adds code, more maintenant and such
17:06:49 <jordanP> I am not even sure it's going to be widely used
17:06:55 <jordanP> but okay we can trye
17:07:17 <oomichi> jordanP: thanks, it is great to get opinion on that
17:07:23 <andreaf> oomichi, jordanP: even if the code is not installed from git, the git log is available - I guess it would useful for users not familiar with git / dev tools
17:07:52 <jordanP> are these users "our" users ?
17:08:41 <andreaf> heh, no idea :P
17:09:06 <oomichi> andreaf: yeah, they can see. but it is easy to make difference between them
17:09:23 <oomichi> them means package one and the latest git
17:10:12 <jordanP> ok, let's move on maybe :)
17:10:27 <oomichi> ok, lets move on
17:10:42 <oomichi> someone have more on the other specs?
17:11:07 <oomichi> #topic Tempest
17:11:21 <oomichi> jordanP: maybe you have one point, I feel :)
17:11:26 <oomichi> for tempest
17:11:40 <jordanP> yes so we have 3 tempest patchs that try to bring cinder v3 teting
17:11:53 <jordanP> #link https://review.openstack.org/#/q/project:openstack/tempest+message:v3+message:cinder+status:open
17:12:13 <jordanP> those patches are not so good because they duplicate a lot of code from cinder v2
17:12:26 <andreaf> jordanP: yes I noticed that I asked the author to sync with each other in some review comment
17:12:30 <oomichi> jordanP: yeah, you are right. that is my concern also
17:12:43 <jordanP> in many aspects Cinder v3 and v2 are the same, to we need new clients ?
17:13:01 <jordanP> and do we even need new tests to begin with ?
17:13:12 <oomichi> jordanP: I feel not necessary because we are using the same clients for Nova case
17:13:15 <jordanP> I am fine adding new clients to test new features of cinder v3
17:13:31 <oomichi> jordanP: +1 for new features of v3
17:13:35 <andreaf> jordanP: well I don
17:13:45 <jordanP> yeah, maybe not
17:14:04 <jordanP> and we haven"t sorted micro version handlind with Cinder right ? (like we do for Nova)
17:14:05 <andreaf> tI don't know in terms of cinder code base
17:14:06 <andreaf> is it the same code shared between v2 and v3?
17:14:31 <jordanP> andreaf, yes it use for most cinder v3 'routes' (they subclass the cinder v2 constrollers)
17:14:52 <smcginnis> v3 and v2 are exactly the same. They only diverge with microversions of v3.
17:14:57 <jordanP> grr I meant, cinder v3 controllers are subclasses of cinder v2 controllers
17:15:06 <oomichi> andreaf: yeah, maybe right. Cinder team wanted to clarify the difference of V3 by different endpoint
17:15:26 <oomichi> then not v2.1, v3
17:15:54 <oomichi> smcginnis: thanks for your help :)
17:16:03 <smcginnis> oomichi: Any time. :)
17:16:19 <jordanP> so I think the first step is to properly support micro version for Cinder
17:16:25 <andreaf> smcginnis: does it mean that the base 3.0 will never diverge from 2.0?
17:16:34 <jordanP> it will, but with micro version bumps
17:16:44 <oomichi> jordanP: ++
17:16:52 <smcginnis> andreaf: Correct. That API is set. Changes will only be allowed as microversion bumps.
17:17:17 <jordanP> andreaf, but we don't want to test cinder v3.0, we want to test cinder v3
17:17:19 <andreaf> cool so basically we can test 2.0 and 3.0 microversion
17:17:22 <jordanP> (notice the 0)
17:17:41 <jordanP> andreaf, yes but what's the point ? they are the same :)
17:17:52 <oomichi> andreaf: yeah, by the same test implementation for both APIs
17:18:09 <andreaf> jordanP: I meant 2.0 and the mircroversions on top of 3
17:18:37 <oomichi> then we will need to implement v3 microversion specific tests as jordanP said
17:18:40 <jordanP> I am not sure I am following :)
17:18:43 <andreaf> the thing is, there may be a cloud where v2 is not available and v3 is, so we have to make things so that we can hit the v3 endpoint from the v2 tests or so
17:19:08 <jordanP> andreaf, true. I didnt think about it
17:19:30 <oomichi> andreaf: yeah, current v2 API tests should be useful for base microversion of v3 API
17:19:42 <oomichi> without copy&past from v2
17:20:18 <jordanP> so shortly we will run the same tests for Cinder v1,v2 and v3 ?
17:20:32 <jordanP> that looks suboptimal...
17:20:33 <oomichi> I am not sure about v1
17:21:05 <andreaf> jordanP, oomichi: I think having the v3 clients would be useful, so we can implement the microversions in them - there's no need to duplicate code just subclass should be fine I guess?
17:21:05 <oomichi> smcginnis: ^^^?  v1 is different from v2 and v3 base microversion?
17:21:42 <smcginnis> oomichi: Correct. There are breaking changes between v1 and v2/3.
17:21:42 <oomichi> andreaf: yeah, that is my thinking
17:21:43 <jordanP> andreaf, yeah, what ever we chose we shouldn"t duplicate code but subclass
17:22:05 <andreaf> for tests I think we should only run one set of tests, either against v2 or v3 - which one is the default is to be decided, I guess v2 for now?
17:22:17 <oomichi> smcginnis: cool, it is fine to have v1 and v2 specific clients on Tempest as current implementation
17:22:21 <smcginnis> andreaf: +1 for just extending the same client.
17:22:30 <andreaf> and then we will have extra tests only running if v3 is available for microversions or so
17:22:57 <oomichi> andreaf: +1
17:23:13 <oomichi> that way is the same as current Nova tests
17:23:52 <oomichi> we can reuse the Nova test code and maybe the test way will be consistent in Tempest
17:24:05 <jordanP> (you guys are going to Atlante in feb ?)
17:24:22 <jordanP> *Atlanta
17:24:23 <oomichi> jordanP: (maybe just first 3 days)
17:24:37 <oomichi> ok, do we have more topic about Tempest today?
17:24:48 <oomichi> jordanP: (how about you?)
17:24:51 <andreaf> I plan to be there 3 or 4 days not sure
17:24:58 <jordanP> (same here, maybe, first 3 days)
17:25:06 <oomichi> jordanP: (cool)
17:25:15 <oomichi> ok, lets move on :)
17:25:27 <oomichi> #topic DevStack + Grenade
17:26:03 <oomichi> I don't have topic about them today, anoyone have?
17:26:13 <jordanP> it's good to have sdague back ;)
17:26:35 <andreaf> oomichi: on Tempest I had the bug report
17:26:41 <andreaf> https://etherpad.openstack.org/p/tempest-weekly-bug-report
17:26:47 <andreaf> #link https://etherpad.openstack.org/p/tempest-weekly-bug-report
17:27:01 <oomichi> andreaf: oh, that is super important, thanks for working that
17:27:13 <andreaf> The open bugs are down to 2 which is good news
17:27:25 <oomichi> andreaf: yeah, I updated the graph: https://github.com/oomichi/bug-counter#current-graph
17:27:43 <andreaf> but there is still a rather large number of open and in progress ones
17:27:58 <oomichi> andreaf: yeah, thanks. The number is still huge
17:28:35 <oomichi> but in this week, the number becomes less as the graph
17:29:12 <oomichi> andreaf: do you have bug reports we need to discuss here? difficult ones or something?
17:29:43 <andreaf> I'm looking at a libvirt issue
17:30:10 <andreaf> #link https://bugs.launchpad.net/tempest/+bug/1646779
17:30:10 <openstack> Launchpad bug 1646779 in OpenStack Compute (nova) "Cannot connect to libvirt" [Undecided,Incomplete]
17:30:30 <jordanP> yes, I've seen it myself too
17:30:43 <andreaf> mriedem was looking at it as well, but I've got no clue what's going on there
17:31:24 <oomichi> andreaf: the failure happens many times on the gate?
17:31:26 <andreaf> it's not so terrible, but it's frequent enough to deserve attention I think, like it happens a few times a day
17:31:51 <jordanP> andreaf, you're e-r query was merged ?
17:32:05 <andreaf> nope
17:32:12 <andreaf> #link https://review.openstack.org/#/c/406278/
17:33:01 <jordanP> mriedem, ^^
17:33:08 <oomichi> mriedem: coud you look at ^^
17:33:36 <andreaf> today I checked dstat logs but I didn't find any usage peak around the time I see the libvirt socket issue...
17:34:09 <jordanP> well I think it's more of a nova/libvirt/kvm issue, I am not sure we can do a lot about it
17:34:30 <oomichi> it is fine to use the E-R and check the more samples
17:35:14 <andreaf> on better news the personality tests are passing again now, so the -ssh check job is back to green
17:35:41 <andreaf> so I guess next I'll try and get the pre-prov cred job back to passing
17:35:49 <oomichi> andreaf: ah, thanks again. That was helped by your devstack patch, right?
17:36:21 <andreaf> yeah a devstack patch, a nova patch and a project-config one to switch the job to xenial
17:36:30 <jordanP> yeah, about the -ssh job, we should all watch over it and if it's stable enough, make it voting
17:37:03 <andreaf> jordanP: I'm not sure whether to make that job voting or to enable validation in one of the existing jobs?
17:37:40 <andreaf> but perhaps it's easier just to turn voting on, and then we can consolidate later
17:37:47 <jordanP> the latter approach would be better but how do we keep track which job uses which settings ?
17:38:04 <andreaf> only that would mean one extra job in the gate
17:38:47 <andreaf> eventually I would like to have validation on everywhere :D so we don't to track that anymore
17:39:38 <jordanP> the job is already running in the check queue so running it in the merge queue is just a tiny bit more work on the infra
17:39:45 <oomichi> if the difference between jobs is clearly, it is fine to make it voting I feel
17:40:26 <oomichi> yeah, just small changes
17:40:45 <jordanP> ok, let's decide next week
17:41:00 <oomichi> do we have more topic about tempest, devstack and grenade?
17:41:32 <oomichi> ok, lets move on
17:41:41 <oomichi> #topic OpenStack-health
17:42:17 <oomichi> as -dev discussion, 100% cue jobs are stopped. but that is still on the dashboard
17:42:31 <oomichi> #link http://status.openstack.org/openstack-health/#/
17:42:43 <oomichi> s/100%/100% failing/
17:43:08 <oomichi> andreaf: when can we avoid seeing that on the dashboard?
17:43:30 <oomichi> andreaf: that means how old data on the dashboard?
17:43:58 <timothyb89> oomichi: by default it shows the last 30 days
17:44:04 <oomichi> the job stopped 12/02/2016, but still on that
17:44:25 <oomichi> timothyb89: thanks :)  OK, we will avoid the next year
17:45:15 <timothyb89> for now, if we change the period in the drop down it disappears: http://status.openstack.org/openstack-health/#/?duration=P1D
17:45:52 <oomichi> timothyb89: cool, yeah I can avoid seeing that :)
17:46:00 <oomichi> timothyb89: btw, do you have topics/patches we need to pick up here today?
17:46:20 <timothyb89> nothing too crazy, but I did put up a patch to improve charts some more: https://review.openstack.org/#/c/406663/
17:46:39 <oomichi> #link https://review.openstack.org/#/c/406663/
17:47:00 <oomichi> ok, I will try it later
17:47:18 <oomichi> do we have more topics about o-h?
17:47:36 <oomichi> ok, lets move on
17:47:39 <oomichi> #topic Critical Reviews
17:47:56 <oomichi> do we have patches we need to review toay ?
17:48:36 <oomichi> (I don't have today)
17:49:23 <oomichi> ok, lets move on
17:49:30 <oomichi> #topic Open discussion
17:50:22 <oomichi> We will be close to PTG soon, and we need to get ideas for QA sessions
17:50:39 <andreaf> heh that was going to be my question :)
17:50:52 <oomichi> I will prepare the etherpad and please give us ideas
17:51:17 <oomichi> andreaf: haha, yeah jordanP picked up good topic in the meeting
17:51:38 <oomichi> andreaf: you are idea man always, I believe you have many :)
17:51:39 <jordanP> :)
17:52:11 <jordanP> cinder v3 and maybe refactor the scenarios manager
17:52:29 <jordanP> we don"t need topics to discuss but more topics to work on I think
17:52:42 <oomichi> jordanP: yeah, they are good working items
17:53:14 <oomichi> I don't want to wait for PTG for them
17:53:37 <oomichi> nice to implement patches for them in Ocata
17:54:29 <jordanP> yeah, we can try
17:55:06 <oomichi> btw, dmellado is the next big triager as just reminder
17:55:15 <oomichi> s/big/bug/
17:55:58 <oomichi> ok, do we have more topics today? or we can implement/review patches :)
17:56:04 <andreaf> the big bug triage
17:56:13 <oomichi> andreaf: haha, you did
17:56:33 <oomichi> super decreasing now
17:57:10 <oomichi> 233 -> 168
17:57:24 <oomichi> in the 4 monthes
17:57:49 <oomichi> ok, lets close the meeting
17:57:51 <oomichi> thanks all
17:57:54 <oomichi> #endmeeting