09:00:03 #startmeeting qa 09:00:04 Meeting started Thu Apr 6 09:00:03 2017 UTC and is due to finish in 60 minutes. The chair is gmann. Information about MeetBot at http://wiki.debian.org/MeetBot. 09:00:05 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 09:00:07 The meeting name has been set to 'qa' 09:00:13 hi, who all here today 09:00:19 o/ 09:00:35 \o/ 09:00:53 o/ 09:01:20 o/ 09:01:30 o/ 09:01:55 #link https://wiki.openstack.org/wiki/Meetings/QATeamMeeting#Agenda_for_April_6th_2017_.280900_UTC.29 09:02:02 ^^ today agenda 09:02:28 #topic Previous Meeting Action review 09:02:55 saw 3 sessions andreaf proposed on forum 09:02:56 I proposed three different sessions for the forum 09:03:04 andreaf: yea thanks 09:03:26 andreaf: whats next on that, TC will select those ? 09:04:32 #link http://forumtopics.openstack.org/cfp/details/111 http://forumtopics.openstack.org/cfp/details/112 http://forumtopics.openstack.org/cfp/details/113 09:04:44 gmann: yeah they're going to be reviewed - but how exactly I don't know tbh - I will check what the process is for sake of completion 09:05:04 gmann: I'm pretty sure we don't get to vote though 09:05:35 heh, yea, let's wait for final selection. hope we get max :) 09:05:49 #topic The Forum, Boston 09:06:02 #link https://etherpad.openstack.org/p/BOS-QA-onboarding 09:06:29 this is same, onboard sessions, put more ideas if anyone have 09:06:53 yeah there is still some time for that but we should be prepared 09:07:15 I guess everyone in the volunteer section below is going to be in Boston? 09:07:24 yea, and hope we have max strength as presence 09:07:44 * masayukig nods 09:07:51 hope so, i still need to get flight booking though 09:08:33 #link Gate Stability - status update 09:08:56 saw some pike on gate pipeline on neutron ssh one 09:09:23 #link https://goo.gl/ptPgEw 09:09:49 #link http://status.openstack.org/openstack-health/#/ 09:10:19 failure seems negative tests on o-h 09:10:26 there's one interesting patch being discussed atm on the ML 09:10:28 #link https://review.openstack.org/#/c/451492/ 09:10:59 would be great if you guys, especially andreaf, would have a look at it 09:11:04 jordanP: thanks 09:11:22 it's an important one, because it asks the question of "what distro/hybrid distro do we test" 09:11:59 jordanP: yes I've seen the ML thread - I will definitely review the patch 09:12:13 our libvirt version is old, newer version might have fewer memory corruption issues 09:12:28 the patch is about testing newer version of libvirt in the Gate 09:12:38 let's see libvirt 2.5.0 have some good fix there 09:12:42 yea 09:12:48 jordanP: possibly yes, the main idea is that we will get support from the libvirt community if we use a recent version :) 09:12:54 in other words, stop testing Ubuntu LTS, but test TLS + UCA packages 09:13:07 yes 09:13:18 jordanP: yeah the idea with LTS was to avoid impossible backports 09:13:36 jordanP: but UCA packages have been already tested by canonical in that sense 09:13:41 will UCA packages be stable always? 09:13:44 jordanP: so they should be a valid option 09:13:49 yeah, I agree we should use UCA 09:14:04 i remember when I used to "manage" a cloud, we used UCA back in the days 09:14:15 i have seen nova doing something like minimum libvirt version required, maybe we can try doing that in tempest too 09:14:54 Tempest is kinda black box testing, we don't even require libvirt 09:14:56 prateek: on nova side it feature dependence etc 09:14:57 prateek: I'm not sure there is a dependency to libvirt in Tempest? 09:15:13 yea, we should not make in tempsest 09:15:33 andreaf, no there is not but i get the idea that we are having problems with a particular version of libvirt 09:15:54 so we can have a constructor which allows us to skip tempest tests on that particular rogue versions 09:16:13 prateek yeah but tempest is there to detect those issues - it's not like we can say that there are specific tests that we can skip with wrong libvirt version 09:16:23 andreaf, ok 09:17:09 prateek: failures are kind of random - at least we don't have a clear reproduce now and I doubt we can pinpoint any specific test in tempest 09:17:19 andreaf: jordanP anything more we need to do on tests separation ? or we monitor gate status with lot of approaches going in 09:17:46 tests separation ? 09:18:06 like separating scenario tests on separate jobs 09:18:12 adreaf, got it, thanks :) 09:18:33 ah, I see, hum nothing that comes into my mind right now 09:19:03 (as usual, please do take into consideration the overall runtime of our jobs when enabling new scenarios/adding new tests) 09:19:12 yea. 09:19:13 (1h20min is already very long) 09:19:45 one last patch that I'd like to see going through is https://review.openstack.org/#/c/450207/ 09:19:46 #link https://review.openstack.org/#/c/450207/ 09:19:47 i am sure we do lot of duplicate operation tests mainly on API tests 09:20:03 but that can be hard to fix due to lot of affort and mainly defcore 09:20:10 that patch is about disabling some not-so-needed Swift services in the Gate 09:20:27 thanks jordanP yeah that's a good one 09:20:49 yeah, we do have a lot of duplication/duplicated tests, so before approving any new test, we should make sure that this is not already tested somehow 09:20:52 yea make sense 09:22:05 thanks andreaf jordanP sdague, clarkb mtreinish ands everyone making so much effort on gate stability 09:22:12 #topic Specs Reviews 09:22:21 #link https://review.openstack.org/#/q/status:open+project:openstack/qa-specs,n,z 09:22:47 new one was #link https://review.openstack.org/#/c/449295/ 09:22:57 Upgrade testing toolset 09:23:22 castulo: sent mail also, i did not get chance to read yet 09:23:25 yea I think that one deserves some more reviews 09:24:25 yea 09:24:56 anything else on spec ? 09:25:27 #topic Tempest 09:25:31 #link https://review.openstack.org/#/q/project:openstack/tempest+status:open 09:25:39 ^^ current open reviews 09:26:22 I've submitted the goal ack for py35 09:26:22 andreaf: jordanP can you re check this - https://review.openstack.org/#/c/419776/ 09:26:36 use case is in bug 09:27:05 andreaf: thanks, so that is just for unit tests right. not as tempest on py35 09:27:12 I don't like that patch, too specific, but we can discuss it in #qa 09:27:25 sure 09:28:17 gmann: #link https://review.openstack.org/#/q/topic:qa_py35_ack - not even just adding py35 in setup.cfg 09:28:27 andreaf: and we have done for both goal, wsgi and py35 right? or any help you need on either side? 09:28:45 gmann: not really I just wanted to let you know 09:28:59 nice 09:29:46 how about py36? next step? 09:30:17 masayukig: still not on tc side right but good to do that 09:30:31 masayukig: heh I think we can do py36 if we have time, the goal for Pike in 3.5 and I think there may be more urgent things 09:30:31 ok 09:30:48 sure 09:31:11 on Bug Triage: 09:31:22 i had this week 09:31:46 #link https://etherpad.openstack.org/p/tempest-weekly-bug-report 09:32:04 few new triaged and cleanup on old one. 09:32:22 overall we have 4 new 09:33:14 for many of the old bugs i saw patches but not reviewed actively so they are stuck in-progress 09:33:42 we should start reviewing the old patches, some of them are nice to fix 09:34:48 next week we have chandankumar turn 09:34:49 #link https://etherpad.openstack.org/p/pike-qa-bug-triage 09:35:22 gmann : yeah, but I think most of old patches are out of date, so rebasing is required. 09:35:44 masayukig: yea and author does not seems to active for many 09:35:47 gmann: how much old patches they are ? 09:35:57 masayukig: i pick some imp one and do myself 09:35:59 if the patch is stuck 'in progress' for a long time, maybe it means that it's not really a bug or it's not that important to fix 09:36:32 gmann : sounds nice 09:36:32 chandankumar: well there are many, few cases with abandon patch and more with in-progress 09:36:47 jordanP: heh, maybe 09:36:59 jordanP: humm may be but sometime people give up :) 09:37:36 sure, they give up because it's not their priority, because of time etc. 09:37:47 mainly i want to cleanup the in-progress bugs 09:37:51 yeah 09:38:06 ++ 09:38:10 either we abandon those patches and open bugs for new assignee or merge them 09:38:15 gmann: i will take a look at in-progess ones this time 09:38:49 chandankumar: thanks. please do ask author if they do not want to work so that anyone else can do 09:38:58 #topic Patrole 09:39:00 gmann: sure, 09:39:57 any patrole guys here, sorry cannot find the irc name right now 09:41:01 ok, only thing i have on patrole is that still i do not know why admin needed on each tests 09:41:15 but i have to check framework more deeply 09:41:43 #topic DevStack 09:42:15 any more updates on devstack side apart jordanP pointed out about patches 09:42:47 I guess you have see the updates in the ML from sdague? 09:43:35 #link http://lists.openstack.org/pipermail/openstack-dev/2017-April/115039.html devstack systemd 09:43:39 yea that was nice actually 09:43:42 systemd 09:44:40 nice one #link https://dague.net/2017/03/30/in-praise-of-systemd/ 09:44:59 #link https://dague.net/2017/03/30/in-praise-of-systemd/ nice one 09:45:24 yea 09:46:14 i can get rid of moving myself over lot of screen 09:46:40 ok. anything else or we move next 09:46:55 #topic OpenStack-Health, Stackviz 09:46:58 masayukig: ^^ 09:47:07 k 09:47:50 jordanP: did you look any further into getting more statsd data into stackviz? 09:47:59 stackviz, I updated one patch for dstat graph improvement 09:48:03 #link https://review.openstack.org/442253 09:48:08 andreaf, we have enough data, now we need to graph them 09:48:33 jordanP: heh yeah that's what I meant 09:48:48 yeah, so, no I'havent tried 09:49:14 and regarding to openstack-health, 09:49:16 I don't know Javascript and I am bad at any frontend stuff 09:49:26 heh 09:49:34 we have masayukig :) 09:49:58 I'm bad, too 09:49:59 masayukig[m] so is stackviz installed from source in the gate? if I do a patch in tempest with depends on will I be able to see the changes? 09:50:39 andreaf: yeah, it sould be 09:50:55 like this http://logs.openstack.org/28/384528/9/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial/920f2e7/logs/stackviz/#/stdin/timeline 09:51:18 a 09:51:40 ah, sorry 09:51:47 I'm not sure that 09:52:14 but I think it's worth to try :) 09:52:18 i think we do from source but not 100% sure 09:52:22 humm I don't see whhat/where is stackviz installed 09:53:06 sudo pip install -U $stackviz_path 09:53:11 it's only written in frontend javascript 09:53:34 uhm pip install won't work 09:53:49 unless that's a file system path :) 09:53:53 jordanP: heh, I actually didn't use pip install 09:53:57 yeah, it's a path 09:54:01 yea #link http://git.openstack.org/cgit/openstack-infra/devstack-gate/tree/functions.sh#n584 09:54:10 #link https://review.openstack.org/#/c/454078/ testing stackviz 09:54:55 ok 09:55:02 5 min left 09:55:05 so, regarding to openstack-health, mtreinish and I approved two openstack-health api server patches, but the api server has nothing to change. 09:55:30 does anyone know the mechanizm of that? 09:56:12 did not get you approved patches but did not reflect on api server side? 09:56:30 don't you need to redeploy something ? 09:56:47 jordanP: it's automatically 09:56:53 automagic :) 09:56:55 masayukig[m]: puppet should pick it up automagically - it's in continuous deploy afaik 09:57:08 andreaf: yeah, I read https://docs.openstack.org/infra/system-config/openstack-health.html 09:57:26 masayukig[m]: so look into the puppet module and ping infra if for checking into logs maybe something is going wrong 09:57:31 a dependency update or so 09:57:35 So, I'd like to see the logs 09:57:47 or it may need a nudge 09:57:58 EmilienM can help may be 09:58:03 masayukig[m]: on subunit2sql I think https://review.openstack.org/#/c/334556 could use more reviews 09:58:16 #link https://review.openstack.org/#/c/334556 for subunit2sql please reviews :) 09:58:40 ok. lets move next 09:58:42 #topic Destructive Testing 09:58:48 andreaf: ah, yeah, 09:59:00 did not see spec updated from samP 09:59:05 andreaf: I'll do it tomorrow :) 09:59:10 #topic Critical Reviews 09:59:31 i have this one, did rebase many times - https://review.openstack.org/#/c/438006/ 09:59:43 #link https://review.openstack.org/#/c/451769/ not so critical but it could use more opinions 09:59:54 sure 10:00:08 we are out of time, let's move to qa if more discussion needed 10:00:14 thanks everyone for joining 10:00:18 #endmeeting