14:00:27 <tellesnobrega> #startmeeting sahara
14:00:28 <openstack> Meeting started Thu Jan 10 14:00:27 2019 UTC and is due to finish in 60 minutes.  The chair is tellesnobrega. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:29 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:32 <openstack> The meeting name has been set to 'sahara'
14:00:42 <jeremyfreudberg> o/
14:01:16 <tellesnobrega> thanks jeremyfreudberg for joining at the exact right time, I almost lost track of time
14:01:26 <tosky> hi
14:01:33 <jeremyfreudberg> hi!
14:01:45 <tosky> me too, 10 minutes ago I knew about this meeting and I was fixing a commit message and I forgot
14:02:00 <tellesnobrega> hi all
14:02:04 <tellesnobrega> #topic News/Updates
14:03:24 <tellesnobrega> I'm really focused on getting the split plugin patch to merge
14:03:29 <tosky> a) helping with the bureacratic part of the plugin splitting (done, we have the new repositories); writing the APIv2 support for both tempest tests (almost done, thanks jeremyfreudberg and tellesnobrega  for the fixes) and scenario tests (basically working, waiting for multihost results)
14:03:56 <tellesnobrega> also worked on some enhancements on apiv2
14:04:38 <jeremyfreudberg> i was deep into apiv2 with a patch finally coming out last night; i spent the rest of the time marvelling at tosky's productivity
14:05:03 <tellesnobrega> jeremyfreudberg, he is amazing
14:05:07 <tosky> come on, I couldn't have done anything without your (both) fixes
14:06:24 <tellesnobrega> moving on
14:06:32 <tellesnobrega> #topic APIv2
14:06:51 <tellesnobrega> we got some really good progress yesterday
14:07:01 <tellesnobrega> we have some merged patches, some under review
14:07:18 <tellesnobrega> what is missing at this point?
14:07:52 <tosky> Jeremy's patch with the fixes, and your patches with OSC support
14:07:57 <tellesnobrega> policies inconsistencies, osc, the polish of apiv2 and microversioning
14:07:59 <tosky> I think I may found another smaller issues
14:08:09 <tellesnobrega> on osc?
14:08:10 <tosky> with job binaries (see my last comment)
14:08:30 <tosky> the same thing that I touched; waiting for the next results from sahara-v2 tempest tests
14:08:44 <tosky> microversioning is probably more relevant (pending the answer from the TC)
14:09:07 <tosky> is the policy fixes something that affects the API, or is it something that we can better recheck next week?
14:09:31 <tellesnobrega> it doesn't affect the API itself
14:10:23 <tellesnobrega> it is just policy checking, I guess we can wait a little more
14:10:25 <jeremyfreudberg> doesn't affect the api itself, but it is part of making the api ready and great... since it is server side i would give it more priority than anything client-side, for the purposes of m2
14:11:09 <tellesnobrega> that makes sense too
14:11:09 <tosky> client-side == the OSC patch?
14:12:16 <jeremyfreudberg> client side is the OSC patch, plus some currently not-yet-written dashboard patches (I think telles's patch covers everything for osc, but i didn't look)
14:13:04 <tellesnobrega> jeremyfreudberg, I think it covers, but as usual, when you get too close you may not see clearly
14:13:14 <tellesnobrega> so please take a look whenever you can
14:14:58 <jeremyfreudberg> so, just to put it clearly, the server side stuff is my new patch, policy things, and microversion / declaration of stable
14:15:37 <jeremyfreudberg> i didn't quite understand what the concern yesterday with microversion was, can someone sum it up again?
14:17:01 <tosky> we have two pending patches, declaration of stable and microversion
14:17:51 <tosky> the question is: in order to reduce the amount of work, and especially if some final touches are needed for the microversioning patch,
14:18:34 <tosky> if we land the 'declaration of stable' now before M2, and discuss and merge the microversion patch next week, will this break the API?
14:18:44 <tosky> or is it allowed?
14:18:57 <tosky> in any case, if we can merge the microversion support now, so be it
14:20:30 <jeremyfreudberg> i see, i don't know the official advice (But I know you have gone looking for it), but i wouldn't want to declare stable without the microvresions (i consider it a breaking change to add the microversions later, since some headers will go from failing silently to failing loudly)
14:21:07 <tellesnobrega> jeremyfreudberg, do you think we can make microversion get in today?
14:21:53 <jeremyfreudberg> i think the microversions patch is mergable toay-- we only didn't merge it in the summer because i wanted more time to fixup apiv2 before naming 2.00-- with last night's patch i guess the fixups are there now
14:22:34 <tellesnobrega> awesome, if that is the case, all concern is pointless since we can make it all happen today
14:22:52 <tellesnobrega> which is the best case scenario
14:23:10 <jeremyfreudberg> tosky, does it make sense?
14:23:41 <jeremyfreudberg> that all being said, it does still require some blind confidence to declare it stable and declare 2.00 -- because then there is no going back
14:24:09 <tosky> jeremyfreudberg: sure
14:24:22 <tosky> and that's the reason why I'd like to have the tempest tests working :)
14:24:33 <tosky> scenario tests seems to be fine, but I'm waiting for the multinode results
14:25:01 <tosky> even if we don't merge the sahara-tests patch
14:25:22 <jeremyfreudberg> right
14:25:27 <tosky> but at least so far the API itself seems fine (the endpoints and the results, at least)
14:26:17 <jeremyfreudberg> my hunch is that we are in good shape, but doubting can be useful at times
14:26:43 <jeremyfreudberg> for what it's worth, i have much more confidence in the plugin split than apiv2 being 100.0% perfect (i'd guess 99%)
14:27:56 <tellesnobrega> jeremyfreudberg, I would say that is pretty high, but we are close :)
14:28:47 <jeremyfreudberg> ok, any more admisitrative comments on apiv2? if not, we should take a second to discuss the policy names
14:29:07 <tosky> jeremyfreudberg: can you please rebase the "Give the illusion of microversion support" on top of the "Some polish for APIv2" patch, so that I can test both?
14:29:48 <jeremyfreudberg> yes, i can do that
14:30:44 <tosky> thanks, I don't have other questions for now
14:31:02 <tellesnobrega> jeremyfreudberg, anything else on apiv2?
14:31:18 <jeremyfreudberg> tellesnobrega: just the policy names
14:31:36 <tellesnobrega> go ahead
14:34:06 <jeremyfreudberg> the idea is that with apiv2 we have made big efforts to entire rename jobs->job_templates and job_executions->jobs, but the policy names are still in the old way: a job-executions policy controls the v2 jobs endpoints
14:34:19 <jeremyfreudberg> i think we all agree that accurate naming is best
14:34:49 <jeremyfreudberg> i guess the problem (in my head, anyway) is that we can't create a "jobs" policy for v2 jobs because that policy name is already taken for the v1 thing
14:35:08 <tellesnobrega> yes
14:35:41 <tellesnobrega> I only see one way out of this, is to create a jobs_v2 policy
14:35:55 <tellesnobrega> and once we drop v1 we can change that back
14:36:35 <tosky> are we sure we can change it back?
14:36:48 <tosky> well, maybe following the deprecation policies
14:37:01 <jeremyfreudberg> tosky: yea, policies can be deprecated
14:37:04 <tosky> do policies support aliases?
14:38:57 <jeremyfreudberg> it looks like not, but i could be wrong
14:39:13 <jeremyfreudberg> i'm reading the "naming policies" section here, by the way: https://docs.openstack.org/oslo.policy/latest/user/usage.html#naming-policies
14:39:24 <jeremyfreudberg> and i guess i found the solution
14:39:32 <jeremyfreudberg> because the resource name should be _singular_
14:39:36 <jeremyfreudberg> as in, not plural
14:40:27 <tellesnobrega> so job?
14:41:33 <jeremyfreudberg> yeah, at the very least apiv2 jobs would be controlled by "job"... for ultimate consistency we could deprecate all the old stuff and use singular nouns for everytinhg
14:41:54 <tellesnobrega> sounds good
14:43:07 <jeremyfreudberg> yeah, as long as we do the deprecation properly i think that that is best
14:43:14 <jeremyfreudberg> maybe it should wait till after m2 though
14:43:21 <jeremyfreudberg> tosky, any thought?
14:43:29 <tellesnobrega> for deprecating stuff, yes
14:43:45 <tellesnobrega> I can fix the inconsistencies now by adding the job policy for v2
14:43:50 <tosky> if we can do it without breaking the API, sure
14:43:54 <tosky> talking about APIv2, did anyone have some time to check that small quirck with /plugins endpoint that I pointed out last time?
14:44:07 <tosky> or was it just a non-issue?
14:44:16 <jeremyfreudberg> i don't recall what it was
14:44:47 <tellesnobrega> and we start deprecating the rest after m2? makes sense?
14:44:56 <tellesnobrega> tosky, I don't recall either, sorry
14:46:48 <tosky> uh, maybe we discussed it outside the meeting
14:47:40 <tosky> ok, so I guess we need a) deploy with unversioned endpoint b) fix quirks by Jeremy c) microversioning support d) stable
14:47:43 <tosky> as basic requirements
14:48:56 <tellesnobrega> yes
14:49:07 <tellesnobrega> and I'm fixing the policy patches as well
14:49:30 <jeremyfreudberg> tosky, if you end up remembering the /plugins quirk, let us know
14:49:36 <tosky> scenario tests with APIv2 passed, in the meantime; both single (fake) and simple multinode (spark)
14:50:00 <jeremyfreudberg> tellesnobrega, i'm in the process of commenting on the policy patch
14:50:01 <tosky> I'm waiting for a rebase of the microversion patch to test it too
14:50:11 <tellesnobrega> jeremyfreudberg, awesome
14:50:19 <jeremyfreudberg> tosky: i'm mid-rebase now
14:50:30 <tosky> should we merge first the APIv2 patches before the plugin split?
14:50:30 <tellesnobrega> awesome tosky
14:50:56 <tellesnobrega> yes, I can rebase it after the apiv2 merges
14:51:04 <tosky> probably yes, better APIv2, then split (which is "just" one patch right now)
14:51:30 <tellesnobrega> yes
14:51:37 <tosky> the big work with split will be next week, with the cascade breakages in the deployment systems (puppet, ansible, tripleo) and packaging
14:52:30 <tellesnobrega> tosky, count me in for that
14:52:45 <tellesnobrega> not sure how much help I will be, but I can try
14:52:56 <tellesnobrega> playing with tripleo has been on my list forever
14:52:59 <tosky> let's discuss it next week :)
14:53:36 <tellesnobrega> yeah
14:53:41 <tellesnobrega> 7 minutes left
14:53:55 <tellesnobrega> split-plugins seems pretty straight forward now
14:54:03 <tellesnobrega> I'm disabling grenade job for it to pass
14:54:16 <tellesnobrega> not a huge deal, we can fix it later
14:54:24 <tosky> yes, and better recheck it after APIv2 as just said
14:54:32 <tellesnobrega> of course
14:54:32 <tosky> so we know that everything is more or less fine
14:54:58 <tellesnobrega> btw, split plugins patch passed tempest
14:55:45 <tosky> yep, noticed
14:55:54 <tosky> we need to refine the jobs, but *after*
14:56:05 <tellesnobrega> yeah
14:56:41 <tellesnobrega> I guess we have a good definition of what we need to get it done today
14:56:53 <tellesnobrega> lets keep the focus and make it happen
14:57:56 <tellesnobrega> thanks jeremyfreudberg and tosky
14:58:18 <tellesnobrega> I know it has been a crazy cycle, but it most certainly will be a great one for Sahara
14:58:46 <tosky> crossing fingers
14:59:14 <tellesnobrega> thanks again, you both have been great :D
14:59:42 <tellesnobrega> #endmeeting