09:07:05 <andreaf> #startmeeting qa
09:07:06 <openstack> Meeting started Thu Aug 25 09:07:05 2016 UTC and is due to finish in 60 minutes.  The chair is andreaf. Information about MeetBot at http://wiki.debian.org/MeetBot.
09:07:07 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
09:07:09 <openstack> The meeting name has been set to 'qa'
09:07:29 <dmellado> hi guys, who's around? ;)
09:07:33 <andreaf> good morning / evening everyone, sorry about the delayed start
09:08:00 <masayukig> andreaf : thanks for your hosting :)
09:08:27 <andreaf> I don't have an agenda ready for today, but we can re-use last week's
09:08:31 <andreaf> #link https://wiki.openstack.org/wiki/Meetings/QATeamMeeting#Agenda_for_August_18th_2016_.281700_UTC.29
09:08:34 <dmellado> +1
09:10:35 <andreaf> #topic newton priorities
09:10:52 <andreaf> #link https://etherpad.openstack.org/p/newton-qa-newton-priorities
09:11:32 <chandankumar> Hello everybody.
09:11:39 <andreaf> I don't have anything specific on this today
09:11:40 <andreaf> hello chandankumar
09:11:46 <gyx> hello
09:12:08 <chandankumar> This is my first qa meeting, looking forward to working with you guys upstream.
09:12:10 <dmellado> chandankumar: o/
09:12:15 <andreaf> We plan to address some of the remaining open items during the sprint
09:12:18 <chandankumar> dmellado: \o/
09:12:25 * masayukig is waving
09:12:43 * chandankumar waves back.
09:12:47 <masayukig> :)
09:13:56 <andreaf> Just a reminder that the qa/infra sprint is going to be 19th -21st of September in Walldorf, Germany: #link https://wiki.openstack.org/wiki/Sprints/QAInfraNewtonSprint
09:14:13 <andreaf> ok let's move on
09:14:25 <andreaf> #topic spec reviews
09:14:39 <andreaf> #link https://review.openstack.org/#/q/status:open+project:openstack/qa-specs,n,z
09:15:24 <jordanP> not much since last week, I didnt read the spec although I said I would...
09:15:30 <andreaf> Nothing really new here, just a note that we have two spec that could use some more reviews
09:15:35 <andreaf> jordanP: heh :)
09:16:18 <andreaf> I meant to re-spin my spec, but I didn't yet
09:16:48 <masayukig> I actually paid attention for the specs these days, but I'll do.
09:17:13 <dmellado> I'll re-read them, still wrapping up after PTO
09:17:30 <andreaf> dmellado: welcome back
09:17:39 <dmellado> andreaf: thanks! ;)
09:17:41 <andreaf> masayukig: thanks
09:17:47 <andreaf> ok, let's move on
09:17:48 <jordanP> I like the fixture spec but I had/have so many issue with resources not properly cleaned up (leak) and shared resources that triggers race conditions
09:17:58 <jordanP> that I am afraid to change everything now
09:18:30 <jordanP> I am not sure continuing the way we do now (addCleanup or cleanup_resource) is a dead end
09:19:15 <andreaf> jordanP: uhm good point
09:19:33 <andreaf> jordanP: part of the spec should be to track what's left around and log of fail at the end of the test run if we leave things around
09:19:42 <andreaf> jordanP: I can add something about that
09:20:09 <andreaf> jordanP: the idea is to make the code more readable and debuggable
09:20:09 <jordanP> I think tracking what's left could fit into DevStack
09:20:42 <andreaf> jordanP: well we could have commands for that in Tempest and trigger them in devstack
09:20:46 <jordanP> if people stopped reinventing the wheel (like how to create a volume) in each test, that would be great to begin with :)
09:20:57 <jordanP> andreaf, true
09:21:23 <dmellado> lol
09:21:42 <dmellado> I'll read that part of the spec, but couldn't the cleanup be used for that?
09:22:32 <andreaf> jordanP: fixture can be created from existing methods quite easily, so it would not have to be such a massive code change
09:22:57 <jordanP> given that these methods are used everywhere/consistently
09:23:00 <andreaf> jordanP: but we do need to make those resource helpers a stable interface (credential providers & co) for plugins
09:23:41 <andreaf> jordanP: so since we need to refactor them anyways, the idea is to create a nice interface for plugins to use
09:23:48 <jordanP> andreaf, do you think that in the end, all fixtures should be in tempest-lib ?
09:23:52 <andreaf> jordanP: so we can avoid resource leaks from plugins as well
09:24:09 <andreaf> credential providers should be
09:24:14 <jordanP> for sure
09:24:15 <dmellado> that'd be a good idea
09:24:25 <andreaf> the mechanism to provision network resources for credentials as well
09:24:29 <jordanP> what about volume and server ? I am afraid tempest-lib will become fat
09:25:21 <andreaf> jordanP: validation resources may be another candidate, since ssh checks can be quite a common thing to do
09:25:39 <andreaf> jordanP: you mean helpers to provision server and volumes?
09:25:46 <jordanP> yup
09:26:01 <jordanP> are those fixtures according to you ?
09:26:43 <andreaf> jordanP: they are not written using the fixtures interface but many of them are kind of fixtures since they schedule the cleanup
09:27:55 <jordanP> I didn"t know fixtures had interface
09:28:28 <jordanP> anyway, I'll leave comments on the spec :)
09:29:05 <andreaf> jordanP: #link https://pypi.python.org/pypi/fixtures
09:29:42 <jordanP> ok
09:30:06 <andreaf> jordanP: for resources that are used by a test only, addCleanup the way to ensure cleanup and implement a fixture
09:30:40 <jordanP> aren't you missing a word on that last sentence ?
09:30:49 <jordanP> addCleanup *is* the way ?
09:31:27 <andreaf> jordanP: for resources that are used by several tests though, we rely right now on tearDownClass and on the enhanced setUpClass that I wrote to ensure tearDownClass is alwaays invoked (which is not true normally, as it won't be invoked in case of failure during setUpClass)
09:31:42 <andreaf> yes *is* the way
09:32:46 <andreaf> jordanP: so I thought we could introduce a addClassCleanup https://review.openstack.org/#/c/351871/ which does the same as addCleanup but at class level
09:33:00 <jordanP> imo, most of the time, sharing resources between tests is not worth it
09:33:16 <dmellado> jordanP: so you'd recreat them?
09:33:19 <dmellado> recreate
09:33:28 <jordanP> yes
09:34:05 <jordanP> if they are cheap like, volumes, servers (not booted from a volume), networks etc...
09:34:13 <andreaf> jordanP, dmellado: well, there are things that can be re-used such as credentials and their networks - I think those are worth re-using
09:34:42 <dmellado> andreaf: +1 on the credentials
09:34:50 <jordanP> maybe let's move on ?
09:34:51 <andreaf> jordanP, dmellado: there's value in re-using them as well as that may uncover extra issues
09:35:00 <jordanP> true
09:35:04 <dmellado> and I was also thinking about servers, not reusing them might lead to expand the test times at the gates
09:35:51 <andreaf> dmellado: well, it depends, if we break the class constraint and use a smart scheduler not necessarily - which is what lifeless is proposing with testresources
09:36:09 <andreaf> dmellado: however that may involve a lot of changes, I have a todo item to look into it
09:36:17 <jordanP> removing duplicate tests or make long tests run faster would bring a bigger benefit that using shared resources imo
09:37:12 <dmellado> andreaf: I see, I'll need to take a look at testresources then too
09:37:42 <masayukig> jordanP: +1
09:37:54 <dmellado> jordanP: I'm +1 on trying that, but maybe we can try both approaches
09:38:01 <dmellado> it'll be even better that way xD
09:38:04 <andreaf> jordanP. dmellado: the main priorities in any case are code readability, debuggability and stable interface for plugins - I don't think we have a big issue in term of gate time to tackle
09:38:23 <jordanP> +1
09:38:26 <dmellado> +1
09:38:45 <andreaf> we have 20min left, so probably we should move to the other topics
09:38:50 <dmellado> let's move on
09:39:17 <andreaf> please leave comments on the spec
09:39:28 <andreaf> #topic Tempest
09:39:38 <andreaf> #link https://review.openstack.org/#/q/project:openstack/tempest+status:open
09:39:47 <andreaf> anything to be discussed on Tempest side?
09:39:53 <jordanP> I have this review https://review.openstack.org/#/c/331764/
09:40:12 <jordanP> I'd like another +2 :)
09:41:16 <andreaf> jordanP: look reasonable to me, I'll check later
09:41:28 <masayukig> jordanP : I'll do it if it'll remain :)
09:42:17 <andreaf> One thing on my side is that I think it would be good to make a release for tempest, since the new interface for service clients in out
09:42:53 <andreaf> not all service clients are stable yet, but I'm not sure we should wait for that
09:43:04 <dmellado> andreaf: that's what we were chatting about yesterday
09:43:17 <dmellado> +1 from my side, only I'd put out another tag when the clients are done too
09:43:36 <andreaf> dmellado: yes that would make sense
09:45:05 <andreaf> ok let's move on
09:45:20 <andreaf> #topic DevStack + Grenade
09:45:47 <andreaf> Anything to be discussed on these two?
09:46:04 <andreaf> #link https://review.openstack.org/#/q/project:openstack-dev/devstack+status:open
09:46:12 <andreaf> #link https://review.openstack.org/#/q/project:openstack-dev/grenade+status:open
09:46:40 * andreaf waits a few more seconds...
09:47:02 <andreaf> #topic OpenStack-Health
09:47:09 <andreaf> #link https://review.openstack.org/#/q/project:openstack/openstack-health+status:open
09:47:16 <andreaf> anything on o-h?
09:47:26 <masayukig> I made 2 patches #link https://review.openstack.org/359059
09:47:28 <masayukig> and
09:47:36 <masayukig> #link https://review.openstack.org/353274
09:47:52 <masayukig> for changing the graphs.
09:48:23 <masayukig> Please have a look if you have time :)
09:48:46 <masayukig> That's all from me.
09:49:39 <andreaf> look nice, thanks
09:50:21 <andreaf> on subunit2sql, two patches got merged, that fix the option to get test attributes into the test_metadata table
09:50:41 <andreaf> and also introduce a new config option to drop the prefix before injection
09:50:57 <andreaf> https://review.openstack.org/#/q/status:merged+project:openstack-infra/subunit2sql+branch:master+topic:story/2000698
09:51:17 <andreaf> I wonder if we could make a subunit2sql release for those?
09:51:22 <andreaf> masayukig?
09:51:45 <masayukig> I think it's nice.
09:52:02 <masayukig> Who can do that? mtreinish or oomichi?
09:52:35 <andreaf> masayukig: not sure, I guess both could, but neither is here now :)
09:52:42 <andreaf> I will ping them later today
09:52:48 <masayukig> thanks
09:53:16 <dmellado> +1 to grant andreaf or someone in EMEA 'power' to do that too xD
09:53:52 <andreaf> dmellado: heh I'm not sure we need releases so urgently that they cannot wait a few more hours
09:54:08 <andreaf> dmellado: but worth discussing :)
09:54:12 <dmellado> andreaf: totally, but it'd be a nice-to
09:54:55 <andreaf> ok maybe for next week's meeting
09:55:17 <andreaf> #topic Critical Reviews
09:55:34 <andreaf> Any review that needs urgent attention?
09:56:59 <andreaf> apart from those mentioned earlier ? :)
09:57:17 <andreaf> #topic Open Discussion
09:57:37 <andreaf> there are 3 minutes left, for any topic to be discussed
09:58:33 <jordanP> nothing from my side, have a good QA week :)
09:58:43 <dmellado> same here!
09:58:49 <andreaf> same to you!
09:58:55 <masayukig> yay!
09:59:09 <andreaf> thanks everyone for attending and contributing today
09:59:15 <andreaf> #endmeeting