17:01:05 <dkranz> #startmeeting qa
17:01:05 <openstack> Meeting started Thu Jun 18 17:01:05 2015 UTC and is due to finish in 60 minutes.  The chair is dkranz. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:01:06 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:01:09 <openstack> The meeting name has been set to 'qa'
17:01:23 <sdague> o/ (though I will have to cut early)
17:01:25 <dkranz> Hi. Who is here today?
17:01:35 <andreaf> o/
17:01:40 <jlanoux> hi
17:02:16 <dkranz> #link https://wiki.openstack.org/wiki/Meetings/QATeamMeeting#Proposed_Agenda_for_June_11th_2015_.281700_UTC.29
17:02:31 <dkranz> Forgot to change the date in the agenda
17:02:32 <anteaya> o/
17:02:36 <anteaya> sc68cal: ping
17:02:41 <sc68cal> hello
17:02:57 <dkranz> #topic Specs Reviews
17:03:09 * afazekas hi
17:03:26 <dkranz> I encourage every one to comment on the plugin spec
17:03:43 <dkranz> #link https://review.openstack.org/#/c/184992/
17:04:06 <dkranz> Does any one have any comments or another spec to discuss?
17:04:46 <edwarnicke> #info edleafe
17:04:50 <edwarnicke> #undo
17:04:53 <edwarnicke> #info edwarnicke
17:04:58 <dkranz> ok, how about any blueprints
17:05:00 <andreaf> oh I missed that, I will do that
17:05:53 <dkranz> ok, I put this item in: "Deprecation of user account options in tempest.conf"
17:06:27 <dkranz> Is there a reason to not transition to using test accounts and deprecate the demo/alt-demo stuff?
17:07:07 <afazekas> dkranz, It is simpler to use, I have no other idea
17:07:11 <andreaf> dkranz: well they are two different things I think
17:07:14 <dkranz> There is a bunch of ugly code to deal with that and it complicates cleanup, tempest config, etc.
17:07:38 <dkranz> andreaf: what is the use case for the tempest.conf user values?
17:07:57 <dkranz> andreaf: rather than using a single set of test accounts?
17:08:51 <andreaf> dkranz: so I think we need to support two use case stiil - tenant isolation, which requires admin credentials only (from configuration file, at least for now), and pre-provisioned accounts, which should use the yaml file only
17:09:11 <dkranz> andreaf: That's what I thought as well.
17:09:25 <afazekas> Generally, tempest has a several code path which is not used too frequently , and maybe noone really needs them. I am ok with droping those, and revert the drop if some pops up, he really needs the old code path ..
17:09:51 <dkranz> afazekas: yeah, we have the cases of "throw-away" cloud, test cloud, production cloud
17:10:27 <dkranz> But I don't see the need for the tempest conf  values for any of them
17:10:28 <andreaf> dkranz: so test jobs in the gate which run serially (ironic I think) should be migrated to use test-accounts, and we should also introduce some gate jobs which use test accounts in parallel
17:10:38 <dkranz> andreaf: yes, I agree
17:10:45 <afazekas> I would like to see, less but well working code path at the early phase of this cycle, then adding even more special cases when the main ones are close to `perfect`
17:10:57 <dkranz> andreaf: I just wanted to discuss a timeline for that
17:11:14 <dkranz> afazekas: what do you mean?
17:11:30 <dkranz> afazekas: I think all the code we need is already there
17:11:49 <dkranz> afazekas: and we just need to start using test accounts more aggressively
17:11:52 <andreaf> dkranz, afazekas: after we change the gate I think we could officially mark those options as deprecated - but I don't know what would be a fair time between deprecation and actually removing them
17:12:39 <dkranz> andreaf: To me it is most important when we can stop supporting them. For example, the cleanup case is hard for those values since we don't specify a network/router like we do in test accounts.
17:12:59 <dkranz> andreaf: So I would like to say we do not support cleanup unless you use test accounts
17:13:22 <afazekas> andreaf: I do not know do we really need to manage fair time, the problem is we do not really know what we are removing is really used by anyone.
17:13:55 <dkranz> afazekas: We don't have to remove it quickly but also do not have to make every new feature work with it
17:13:57 <afazekas> BTW. I know we have lot of requirements for more kind of ssh/network support
17:14:16 <andreaf> afazekas: yes but oslo.config introduced this concept of deprecation exactly because it's not fair to just drop options suddenly
17:14:39 <dkranz> I don't think we should drop it suddenly
17:14:52 <dkranz> Just suggesting that newer tools may not need to support that case
17:15:04 <andreaf> dkranz: sounds fair enough to me
17:15:18 <afazekas> andreaf: It is not fair, the question is do we consider tempest as stable api product, or just something which is testing at the ci system
17:15:41 <dkranz> afazekas: we are trying to stabilize it, but it is not stable quite yet
17:16:05 <andreaf> afazekas, dkranz: tempest aims to be run against as many cloud as possible
17:16:19 <dkranz> andreaf: yes
17:16:47 <dkranz> andreaf: I see this much like how we got rid of nose
17:16:54 <andreaf> afazekas, dkranz: I know we do not provide a stable interface (tempest-lib does) but still I believe we should provide a grace time - we can also ask the ML
17:17:10 <sdague> ++ to some deprecation time on config options
17:17:19 <dkranz> andreaf: sure
17:17:28 <sdague> given that people are given the whole config file and told that's their interface
17:17:38 <dkranz> ok, I'll send something to the mailing list
17:17:43 <dkranz> sdague: agreed
17:17:48 <afazekas> dkranz,andreaf: the question is can we do faster run, if we do thing not too politely now, and reaching a `product` which is more maintanble and consistent at the end of the cycle .
17:18:45 <dkranz> afazekas: The main thing I am proposing is that the new 'tempest cleanup' and 'tempest configure' not support the conf user names
17:19:03 <afazekas> sdague: With a rolling release component like timest, is 1 month deprecation time is enough ?
17:19:08 <dkranz> afazekas: but not remove those options except after quite some time
17:19:14 <sdague> afazekas: tempest has releases
17:19:42 <sdague> so it should deprecate across at least one release boundary
17:19:48 <dkranz> I suggest not removing these options until the beginning of M, even though we are not on that schedule
17:19:50 <sdague> mtreinish should clearly weigh in
17:19:52 <sdague> yep
17:20:02 <sdague> that seems pretty reasonable
17:20:12 <dkranz> sdague: yes, we should discuss with Matt for sure
17:20:20 <afazekas> I have doubts about we really need to have 6 month deprecation times at master.
17:20:48 <dkranz> afazekas: well now it is only 4 months :)
17:20:50 <andreaf> dkranz, afazekas: yes I think it's fair to say new tools won't support deprecated options - we will migrate credential providers to tempest-lib, but I won't migrate the support to configured credentials there for sure - it will stay in tempest as deprecated
17:21:05 <dkranz> andreaf: sounds good
17:21:44 <dkranz> I am going to try and create a test account job
17:21:51 <dkranz> If my bash is up to it
17:22:07 <dkranz> andreaf: have you done that yet, at all?
17:22:41 <dkranz> sdague: since you had to leave early, is there anything you wanted to discuss about devstack/grenade?
17:23:16 <andreaf> dkranz: we mostly used a patch on top to test the test-accounts code path until now - so I think you will have to create one
17:23:30 <dkranz> andreaf: ok
17:23:44 <sdague> dkranz: sure, the modular grenade bits are out there
17:24:04 <dkranz> sdague: that's great. Tempest is a little behind in that area.
17:24:05 <sdague> #info grenade out of tree plugin write up - http://lists.openstack.org/pipermail/openstack-dev/2015-June/066583.html
17:24:21 <sdague> heat works with it, but not all the heat code is landed yet
17:25:02 <sdague> I also started to execute some of the devstack cleanup agreed to at summit
17:25:13 <sdague> by deleting the non rabbit bits of devstack
17:25:14 <dkranz> sdague: I think there was a lot of noise about your email because many folks did not realize that "big tent" also meant "small tent" in terms of some things people were used to having and are now going to be ejected.
17:25:28 <sdague> #info Removing non Rabbit from Devstack - http://lists.openstack.org/pipermail/openstack-dev/2015-June/067111.html
17:25:37 <dkranz> sdague: I got similar comments in some of the tempest sessions.
17:25:47 <dkranz> sdague: but it is great that modular progress is being made
17:25:59 <sdague> so, yeh, that got more adventurous than I expected
17:26:03 <dkranz> sdague: :)
17:26:12 <dkranz> sdague: change is hard, I guess
17:26:24 <sdague> I wrote the zmq plugin this morning
17:26:32 <sdague> #info zmq out of tree devstack plugin - https://github.com/sdague/zmq-devstack
17:26:46 <dkranz> sdague: I think that will help demonstrate it is not such a big deal
17:26:52 <sdague> it took about 2 hours from blank canvas to fully tested
17:26:59 <afazekas> BTW, AMQP 1.0 proton thing (with routers instead of brokers) was promising, is anybody knows it's status ?
17:27:00 <sdague> dkranz: yeh, I hope so
17:27:26 <sdague> afazekas: no idea, and honestly, from a devstack perspective, they should be doing that as a plugin like the zmq one
17:27:58 <dkranz> yup
17:28:03 <sdague> so anyway, that's where those things stand
17:28:15 <dkranz> sdague: ok, thanks for the update and the great example
17:28:44 <dkranz> I think bugs is next up.
17:29:01 <dkranz> Any one have any bugs to discuss?
17:29:26 <afazekas> I am not big fun of having a lot of small plugin repos, instead of occasionally merging something what you may not fully understand to the main repo
17:29:32 <dkranz> The bug triage rotation is still quite empty: https://etherpad.openstack.org/p/qa-bug-triage-rotation
17:30:03 <dkranz> Is there a reason folks are not signing up for a few slots?
17:30:29 <dkranz> afazekas: I think it is really going to be up to the devstack team to decide that
17:31:49 <andreaf> dkranz: it's really hard for me to spend time on bug triage, I'd rather use the time I have on reviews on blueprints
17:31:57 <afazekas> afazekas: I see, but et the end will end up with a lot of repos
17:31:57 <andreaf> dkranz: how much time does it take typically?
17:32:22 <afazekas> andreaf: 2-4h
17:32:27 <dkranz> andreaf: I would say about 1/2 to 1 hour for a week duty
17:32:39 <dkranz> andreaf: of course it depends on incoming bugs
17:33:02 <andreaf> dkranz, afazekas: ok I'll give it a go
17:33:23 <dkranz> andreaf: It is mostly changing bugs from New to Invalid or Confirmed/Triaged
17:33:28 <dkranz> andreaf: thanks!
17:33:47 <dkranz> andreaf: we get a fair number of invalid bug reports
17:34:12 <dkranz> andreaf: but usually not more than one or two per day
17:34:32 <dkranz> ok, any reviews that should get attention?
17:34:32 <andreaf> dkranz: ok
17:34:46 <andreaf> dkranz: yes - let me get the link :)
17:35:03 <dkranz> andreaf: anyway, none of us really wants to do it which is why we have the rotation :)
17:35:12 <afazekas> andreaf: some cases there are bugs for surprising things, and you may need to do some study before you can response in a good way.
17:35:29 <dkranz> afazekas: true
17:35:51 <andreaf> dkranz: #link https://review.openstack.org/#/c/151601/
17:36:04 <dkranz> cleanup: https://review.openstack.org/#/c/191978/
17:36:10 <andreaf> afazekas: yes sure
17:36:38 <jlanoux> dkranz: andreaf sdague SSL experimental #link https://review.openstack.org/#/c/180968/
17:37:22 <dkranz> sorry, I have to leave for two minutes
17:39:56 <dkranz> any one have anything else to discuss?
17:40:43 <dkranz> I will send an email about the tempest conf thing
17:41:12 <andreaf> dkranz: thanks
17:41:33 <dkranz> ok, last call for discussion...
17:42:19 <dkranz> Thanks every one.
17:42:21 <afazekas> andreaf, With ssh stuff, I would like to move fast. So if the change does not looks dangerous to the gate stability we should merge it and polish later. Looks like it will not be done ever if we nit picking..
17:42:24 <sdague> jlanoux: so my concern on that is we've got this whole issue with server catalog reflection
17:42:45 <sdague> which means a huge amount of config changes are needed, which we need to fix
17:42:59 <sdague> someone else was working through changes related to that and promissed an ML thread which I don't think I've seen yet
17:44:06 <sdague> anyway, can take that back to channel
17:44:26 <jlanoux> sdague: ok - I'll watch the server catalog reflection
17:44:42 <dkranz> afazekas: Is there a review you comment is referring to?
17:44:52 <jlanoux> sdague: I'll ping you later on - I need to take care of the kids now
17:45:08 <afazekas> dkranz: "dkranz: #link https://review.openstack.org/#/c/151601/"
17:45:42 <dkranz> afazekas: It just awaits your +A :)
17:46:46 <afazekas> dkranz: :) tomorrow, (I do not wan't +A when I am in a pub)
17:46:58 <dkranz> afazekas: ok :)
17:47:15 <dkranz> afazekas: there was also a request to look at the patches depending on this to make sure they run
17:48:00 <jlanoux> dkranz: afazekas yes - there was a last issue with rebasing. that shoud be good soon
17:48:21 <dkranz> jlanoux: thanks
17:48:29 <dkranz> ok, if there is nothing else, we can end a few minutes early
17:49:12 <dkranz> #endmeeting