08:59:19 #startmeeting qa 08:59:20 Meeting started Thu Sep 8 08:59:19 2016 UTC and is due to finish in 60 minutes. The chair is gmann. Information about MeetBot at http://wiki.debian.org/MeetBot. 08:59:21 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 08:59:24 The meeting name has been set to 'qa' 08:59:40 who's here for today? 08:59:50 o/ 09:00:04 \o 09:00:22 * masayukig was surprised at a little bit early starting :) 09:00:39 masayukig: hferenc: hi 09:00:47 masayukig: oh is it :) 09:00:52 hi guys 09:00:53 hi 09:01:06 * dmellado too was surprised, xD 09:01:19 dmellado : heh :) 09:01:45 its was already 09.00 UTC in my PC :) 09:02:29 andreaf: afazekas mkoderer : ping for meeting in case available 09:02:31 gmann : heh, you need a ntpd or something like that :) 09:02:49 masayukig: its from internet how it can be wrong :) 09:02:59 gmann: lol 09:03:11 anyways let's start the meeting 09:03:20 * masayukig nods 09:03:23 #link https://wiki.openstack.org/wiki/Meetings/QATeamMeeting#Agenda_for_September_8th_2016_.280900_UTC.29 09:03:34 Today agenda ^^ 09:04:14 #topic Newton Priorities 09:04:23 #link https://etherpad.openstack.org/p/newton-qa-newton-priorities 09:04:50 we will not go with each topic. 09:05:30 I think the highest priority cleanup one is around 40% done 09:05:48 cool 09:06:13 but there are lot of things still pending which we can target in code sprint 09:06:49 for other items also we will discuss in code sprint 09:07:02 gmann: I'd like to get some help on the test class hierarchy squash on the code sprint 09:07:03 oomichi might be picking up the priority one there 09:07:17 for the lib/migration I thing we're on pretty good track 09:07:18 dmellado: great, but you can always start before that also :) 09:07:46 dmellado: only service client but there are many things pending for migration 09:08:07 but that we covers one by one not so urgent 09:09:07 anyways let's move 09:09:13 in any case for the remaining items I guess we'll cover that in the sprint 09:09:15 yeah 09:09:24 dmellado: true 09:09:26 #topic QA Code Sprint 09:09:38 #link https://wiki.openstack.org/wiki/Sprints/QAInfraNewtonSprint 09:09:44 details here ^^ 09:09:54 there are still 5 seats available 09:10:28 i am surprised not to see andreaf name there 09:10:43 gmann: he was going to be on PTO, IIRC 09:10:45 so he couldn't make it 09:10:58 ahh 09:11:33 i remember to see his name there but do not know 09:11:44 anyways here is etherpad link for collecting the ideas 09:11:47 #link https://etherpad.openstack.org/p/qa-infra-newton-midcycle 09:12:08 write up your idea/items you are planning to work if not listed already 09:12:36 ok 09:13:29 if nothing on this can we move to next topic 09:13:50 let's move on 09:13:59 #topic Specs Reviews 09:14:09 #link https://review.openstack.org/#/q/status:open+project:openstack/qa-specs,n,z 09:14:43 not much new spec but we should review the exiting one 09:14:58 i did not get chance to look into those since 1 month seems 09:15:06 anyone has any updates or want to discuss 09:16:34 ok let's jump to next one 09:16:36 #topic Tempest 09:16:45 #link https://review.openstack.org/#/q/project:openstack/tempest+status:open 09:16:49 ^^ ongoing reviews 09:17:07 I could use some +2+A on the ones that already got a +2 from oomichi 09:17:10 https://review.openstack.org/#/q/status:open+project:openstack/tempest+branch:master+topic:bp/consistent-service-method-names 09:17:11 many has +2 one so review those 09:17:45 dmellado: yea :) you will for sure 09:18:00 gmann: so please do review them if you've the time ;) 09:18:16 dmellado: sure but may be tomorrow, i have to prepare the visa doc today 09:18:27 gmann: oh, no prob, thanks in any case! 09:18:33 dmellado: anyways nice progress in identity clients 09:18:44 we'll tackle volume ones now 09:19:00 hope we will finish all clients by code sprint 09:19:24 next is Bug triage things 09:19:36 oomichi: started a graph collection on those 09:19:39 #link https://github.com/oomichi/bug-counter#current-graph 09:20:13 #link https://github.com/oomichi/bug-counter/blob/master/tempest_bug_count.png 09:20:17 is this? 09:20:21 masayukig: how about including existing bugs data in dashboard may be critical/hogh priority one 09:20:49 masayukig: yea 09:21:11 that can give more eyes from project on critical open bugs? 09:21:44 but not sure masayukig and mtreinish can provide opinion on this 09:21:52 I think it's a nice idea 09:22:06 gmann : which dashboard? o-h? 09:22:13 masayukig: yea 09:22:56 hrm, it seems nice. 09:23:24 ok. anyways let's decide that later 09:23:37 sure :) 09:23:38 we stopped doing the regular bug traige and open bug count is high seems 09:24:10 i remember many people volunteers for that in last summit :) 09:24:24 gmann: masayukig how about adding a bug fixing section to the mid-cycle items 09:24:42 bug fixing is not something you do "one day" and be done with it 09:24:53 we have to do it, every week 09:25:00 we can discuss this also in code sprint may be assigning a permanent or turn wise 09:25:09 +1 jordanP 09:25:16 jordanP: I know, but at least if we could get rid of some stuff by the code sprint it'd be nice 09:25:19 jordanP: yea but if we have someone looking up daily for those will be nice 09:25:19 I have this handy bookmark: https://bugs.launchpad.net/tempest/?field.searchtext=&orderby=-importance&field.status%3Alist=NEW 09:25:46 most of the bug reports are not real bug, most of the work is actually replyinq/asking some question 09:25:47 yea count is 30 09:26:09 weekly also works if we have some assignee there 09:26:12 +1 to adding bug triage/assignment to weekly meeting 09:26:38 yea, let's add that in code sprint agenda 09:26:54 #action gmann to add bug triage topic in code sprint etherpad 09:27:27 next one is from hferenc 09:27:33 hferenc: please go ahead 09:27:44 okay 09:28:08 first of all i'm aware of what a -2 means 09:28:21 and i sort of agree with oomichi 09:28:41 however i cannot find a way to implement this in nova 09:29:00 link ? 09:29:04 sry 09:29:13 #link https://review.openstack.org/#/c/315786/ 09:29:14 patch 315786 - tempest - Add test case to test the reset action of libvirt ... 09:29:16 hferenc: about that, I might've some related stuff 09:29:17 #link https://review.openstack.org/#/c/315786/ 09:29:18 patch 315786 - tempest - Add test case to test the reset action of libvirt ... 09:29:33 hferenc: you mean to add test in nova? 09:29:34 there was some stuff in nova that we wanted to tackle but we needed to interact directly with libvirt 09:29:35 * andreaf o/ 09:29:36 and I was proposing this 09:29:47 https://review.openstack.org/#/c/366091/ 09:29:47 patch 366091 - openstack-infra/project-config - Adds a new project: nova-tempest-staging 09:29:52 I was going to bring this on the open discussion 09:30:01 hey andreaf o/ 09:30:21 * masayukig waves 09:30:28 hferenc: could you please summarize the issue you were tackling? 09:30:59 dmellado: any example on that category which cannot go in tempest or nova 09:31:08 hello everyone, sorry I'm late 09:31:15 yea that will be nice to understand background 09:31:19 andreaf: hi, 09:31:30 the test would verify that everything happens if hw:watchdog_action extra spec is set 09:31:38 gmann: in my case, basically from what I was told, things that should interact directly with libvirt 09:31:47 i was able to verify tho whole process in tempest 09:32:18 however the solution is kind of on the edge regarding the tempest design priciples 09:32:27 I kinda agree this is going to be hard to maintain in Tempest 09:32:41 it's a very small feature that I doubt a lot of people use 09:33:30 but the libvirt driver is mocked in nova functional test so i can only check whether the flavor has a specific flag 09:33:50 hferenc: yea that is fake driver nova use 09:33:57 yepp 09:34:18 so i cannot check what happen when that flag is set 09:34:47 hmm, should we have a new tempest test repository for like these niche cases..? 09:35:07 masayukig: I think that'd be great 09:35:18 I think if it's a feature valuable to hferenc then it's the way to go, imo 09:35:19 humm, am not sure why we cannot do in tempest 09:35:41 gmann: could we add such feature in tempest? 09:35:47 interacting with libvirt in such way?¿ 09:35:55 only issue i see about the waiting on rebooting of instance by watchdog 09:36:09 hferenc, dmellado, masayukig, gmann: the test does not involve much more than nova really, so it sounds like something that should be tested in nova 09:36:13 libvirt only tests are not really nice to Xen/VMWare etc... 09:36:35 yea something we can skip for others driver 09:36:49 and we would have to maintain image_ref_watchdog, a specific glance image 09:36:49 andreaf: actually in nova, tests use fake driver only 09:37:14 hferenc, gmann: I understand that nova functional tests use the fake driver today, but that doesn't mean that it can't be changed if there is a use case 09:37:50 it may be possible to extend the fake driver to cover what you need, or to have a different job that uses a real hypervisor 09:38:01 andreaf: hferenc yea, i remember there was plan to build up full stack for nova testing in nova 09:38:15 and this requires CONF.validation.run_validation which we don't have enabled in the gate jobs by default 09:38:25 but i think in Atlanta summit 09:38:45 meah, overall atm, imo, it's a bad idea to support this inside Tempest itself 09:38:46 jordanP: thats should be ok to run on ssh job only 09:39:05 jordanP: the watchdog image is the same as the original images, except that i added one extra kernel module for the watchdog device 09:39:09 but yea driver specific things yes we should be carefull 09:39:17 which ssh-jobs ? The flaky ones, non voting ? The jobs no one is taking care of ? 09:39:35 jordanP: :), yea we will make voting that soon 09:40:22 I heard that a long time ago already. I am all on favor of making it voting but we need to do it step by step. The more tests we throw that required ssh_validation, the harder it will be to stabilize the job 09:40:41 gmann, jordanP: still I don't think this belongs into tempest - it can be written in tempest format but it doesn't have to live in the tempest tree 09:40:57 andreaf: that was as well my idea 09:41:09 my concern is about where putting up those kind of tests 09:41:17 should we try to 'unify' those 09:41:28 tempest-plugin 09:41:35 I'm afraid of upcoming 'tempest-like' style repos coming up 09:41:46 tempest like repo is not good idea 09:41:50 jordanP: yep, that was the idea on here 09:41:50 plugin suits then 09:41:54 https://review.openstack.org/#/c/366091/ 09:41:54 patch 366091 - openstack-infra/project-config - Adds a new project: nova-tempest-staging 09:42:02 that's the point of tempest/lib, to have a lot of tempest like repo, no ? 09:42:11 gmann: andreaf jordanP would you mind commenting there as well when you get the chance? 09:42:32 if that's acceptable, I'd like to have some ack from your side there 09:42:44 dmellado: ok 09:42:58 ok 09:43:21 but this case can be in nova side with full stack testing framework or plugin 09:43:54 anyways let's comment in reviews 09:44:16 i need to think more on this if can not be in tempest 09:45:04 15 mins left 09:45:07 anything more on Tempest side or we move on ? 09:45:11 masayukig: yea 09:45:39 if nothing on devstack and grenade let's jump to dashboard one 09:45:45 we can move on i think 09:45:54 #topic openstack-health 09:46:02 masayukig: andreaf anything on this 09:47:04 I was looking forward to a new release of subunit2sql - but it may be already there I didn't check yet 09:47:16 so, timothyb89 are now trying to improve/reimplement the some graphs on it. 09:47:56 And I've heard mtreinish has some ideas to add graphs. But it is not started yet. 09:47:56 ok 09:48:19 And I put one small patch for improving the top page loading. 09:48:28 more graphs does not make it more heavy :) 09:48:29 Error: I haven't seen graphs. 09:48:48 https://review.openstack.org/#/c/367172/ 09:48:48 patch 367172 - openstack-health - Add caching on get_all_runs_time_series_by_key for... 09:48:59 why ERROR by patchbot 09:49:11 lol 09:49:16 "Recently, the home page loading time is so long and I'm really annoyed every time." <<--- yes 09:49:44 we should really do something about it, slow loading time makes o-h not super usable 09:50:08 yea that true. may be more caching thing can help 09:50:15 jordanP : yeah, I think there are some reason. one is api side, the other is rendering graphs. 09:50:42 my patch tries to improve api-db side. 09:51:03 masayukig: nice 09:51:36 That's all from me 09:51:40 ok let's wait for masayukig patch to see fast loading 09:51:42 masayukig: Thanks 09:51:54 #topic Critical Reviews 09:52:03 any critical reviews from anyone 09:52:53 ok, then i have 3 not so critical but need 1 more +2 09:53:02 #link https://review.openstack.org/#/c/365859/ 09:53:02 patch 365859 - tempest - Define separate inherited_roles_client for inherit... 09:53:04 #link https://review.openstack.org/#/c/366611/ 09:53:04 patch 366611 - tempest - Move v3 roles_client to library interface 09:53:11 #link https://review.openstack.org/#/c/366641/ 09:53:11 patch 366641 - tempest - Move v3 inherited_roles_client to library interface 09:53:19 please review 09:53:38 gmann: I'll do, and that'll be also great to have merged asap as I've some patches dependant on those ;) 09:53:41 I already +2 later two patches :) 09:53:53 thanks, masayukig 09:53:54 masayukig: yea thanks 09:54:05 dmellado: yes, we will finish identity v3 ones 09:54:11 5 min left 09:54:13 #topic Open Discussion 09:54:22 anything else to discuss 09:54:43 masayukig: if you could also have a look at the patches that already have +2 here, that'd be great 09:54:46 https://review.openstack.org/#/q/status:open+project:openstack/tempest+branch:master+topic:bp/consistent-service-method-names 09:55:15 dmellado : sure, will do tomorrow :) 09:55:39 This patch: https://review.openstack.org/#/c/344184 Python3 relaated 09:55:40 patch 344184 - tempest - py3: Miscellaneous fixes 09:55:40 * masayukig puts it in his todo list 09:55:46 i just wanted to say thanks for the ideas about the watchdog think, i'd appreciate some comments i you have any further ones 09:55:59 we should really make Python3 happen, are everybody else is going to wait for us 09:56:03 hferenc: thanks to you, I'll comment in there 09:56:09 jordanP: +1 09:56:25 jordanP: yea, ll do tomorrow for sure, love those kind of patch 09:56:34 hferenc: sure, thanks 09:57:13 hferenc : yeah, I'll have a look 09:57:13 ok, if nothing more, we close the meeting 09:57:14 3... 09:57:21 2... 09:57:24 :) 09:57:27 1. 09:57:28 xD 09:57:31 Thanks all 09:57:33 \o/ 09:57:36 #endmeeting