00:02:38 #startmeeting nova api 00:02:39 Meeting started Fri Nov 14 00:02:38 2014 UTC and is due to finish in 60 minutes. The chair is cyeoh. Information about MeetBot at http://wiki.debian.org/MeetBot. 00:02:40 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 00:02:42 The meeting name has been set to 'nova_api' 00:02:55 hi sorry I'm late is anyone around for the nova api meeting? 00:03:44 hi 00:04:53 gmann: do you know if oomchi is back from summit yet? 00:06:14 cyeoh: yes, he came back in office. 00:06:47 gmann: ok I was hoping he could give us an update from summit 00:07:00 cyeoh: but not sure whether he is going to join meeting. 00:07:09 cyeoh: yes. 00:07:43 ok looks like alex is having trouble staying connected today too so just us 00:07:58 gmann: is there anything that you'd like to talk about - any urgent reviews etc? 00:08:03 cyeoh: oh 00:08:23 cyeoh: nothing urgent from my side. 00:08:54 I'm here~ 00:09:04 ok. I've just come back from holidays recently so I don't really have much either. 00:09:08 #link https://etherpad.openstack.org/p/kilo-nova-microversions 00:09:22 cyeoh: we have around 30 reviews pending on V2.1. we can try to merge those. 00:09:39 the api microversions spec seems to be pretty close - the summit meeting seemed to go well with nothing scary/surprising 00:10:03 gmann: ok how about we all put effort into getting those merged over the next week. That will clear things up for getting started on microversions 00:10:12 alex_xu: hi 00:10:23 cyeoh, do we still need continue think about the implementation for microversion 00:10:38 cyeoh: +1. 00:10:43 alex_xu: so I've looked over your proposal. What I'd like to do is defer it for now as I think its something we can quite easily merge in later 00:10:57 because I want to get the microversions spec approved asap 00:11:03 cyeoh, ok, got it 00:11:39 alex_xu: there's bits of it I really like but I think it'll take a while to sort out and its more of a potential improvement in the future. 00:12:11 cyeoh, yea 00:12:16 so I've heard that at the api microversions session the goal was sort of set to have microversions merged in kilo and say just 1 or 2 microversions implemented 00:12:35 ... just to set expectations.... 00:13:11 #topic open discussion 00:13:21 did anyone have anything else they wanted to talk about? 00:13:47 just think we have some extension porting patch didn't merge for v2.1 yet. 00:14:17 alex_xu: but they are submitted to gerrit? 00:14:20 After those merged, then we can think v2.1 is most done? the json-schema and share unit is low priority... 00:14:32 yes, I guess all on the gerrit 00:14:45 yes, that sounds good. We don want to have 100% json schema input validation coverage 00:14:57 alex_xu, cyeoh: i also think all are in progress 00:15:12 at that point I think we can pass it over to ops/users and ask them to test 00:15:42 emm... before fix this https://review.openstack.org/#/c/131639/? 00:16:19 alex_xu: well yea we want to be able to pass tempest :) 00:16:28 after all merge and before asking user to test, we can have look into https://review.openstack.org/#/c/131639/ for V2.1 stability 00:16:55 yep, we'll lock it in with that, but be run as just a nightly job though (due to gate resources) 00:17:01 heh :) that looks scary now~ 00:17:03 we'll need to make sure we look at the nightly job output 00:17:29 yea we should prioritise fixing failures, but I'm sure many of them are just due to unmerged code 00:17:47 ok 00:17:53 and some are expected failure due to extra response attribute etc 00:18:34 I thought we allowed extra response attributes in our tempest response schema testing? 00:19:29 yes, in schema side it is fine but some tests like list etc tests the complete response dict as whole 00:19:58 ah ok, so if you know of specific examples of those we should fix the tests 00:20:12 do you want to list them on the v2.1 worklist page? 00:20:40 you mean we should fix tempest tests? 00:20:50 yep 00:21:29 so should not that be part of V2.1 tempest testing work 00:21:47 just wondering why we fixing those tests as it works fine for V2 :) 00:21:55 yes, just thinking we can get ahead of them a bit 00:22:09 so the v2 tests will be used to test v2.1 - we won't have a separate tree of tests 00:22:32 hummm, nice idea 00:22:37 i'm not sure what we'll do with microversions yet.... 00:22:51 ok i will list those today. 00:23:00 gmann: thanks! 00:23:25 is there anything else? 00:23:26 cyeoh: not sure about any discussion on moving functional testing from tempest to services 00:23:31 last thing, heard from nova meeting. k-1 is spec merge freeze...so policy stuff should be merged before that. 00:23:59 alex_xu: ok we need to keep on those specs then. 00:24:19 gmann: I think we should be doing it - I'm not sure if tempest-lib needs to mature a bit first? 00:24:20 yes 00:24:24 so do not know whether it is good to write V2.1 tempest tests to run those on gate in parallel with V2 00:25:26 gmann: so I don't think we need to write any v2.1 tempest tests in parallel yet 00:25:35 until we have microversions. For now we just keep adding to the v2 tempest tests and use that test suite against v2.1 00:25:47 or am I misunderstanding what you're saying? 00:25:47 yes, tempest-lib needs more mature in term of accommodating rest client etc 00:26:20 cyeoh: looks fine. 00:26:37 gmann: ok I guess if we have spare time we can help out there too 00:26:55 basically we should wait on https://review.openstack.org/#/c/96661/ till micro version? 00:28:05 gmann: so I think its ok to work on that now 00:28:23 that looks like just compatibility work for the tests to work between v2 and v2.1 00:28:39 but won't need to support say backwards incompatible stuff 00:28:52 the microversions changes will require extra headers etc 00:29:48 cyeoh: yes, sounds good. thx 00:30:36 was there anything else from anyone? 00:30:52 that's all from me 00:30:59 from me too 00:31:14 ok lets finish early then. 00:31:20 thanks everyone for turning up! 00:31:29 thanks all. have a good day 00:31:32 #endmeeting