00:01:18 <cyeoh> #startmeeting nova api 00:01:19 <openstack> Meeting started Fri Feb 27 00:01:18 2015 UTC and is due to finish in 60 minutes. The chair is cyeoh. Information about MeetBot at http://wiki.debian.org/MeetBot. 00:01:20 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 00:01:22 <openstack> The meeting name has been set to 'nova_api' 00:01:31 <cyeoh> hi anyone around for the nova api meeting? 00:04:37 <gmann> hi 00:05:14 <cyeoh> gmann: hi 00:05:37 <oomichi> hi 00:05:42 <cyeoh> hey 00:05:59 <cyeoh> should we start with v2.1/microversions? 00:06:19 <oomichi> cyeoh: yeah, ready to go :) 00:06:27 <cyeoh> #topic v2.1 00:07:21 <cyeoh> so I think the only thing blocked now is CUMENTATION: At least some basic documentation on how to write a V2.1 plugin 00:07:21 <cyeoh> https://review.openstack.org/156882 00:07:54 <cyeoh> jaypipes has some problems with it, but I think we can still release without a doco patch 00:08:12 <oomichi> cyeoh: yeah, I agree with that 00:08:28 <oomichi> the patch is just for v2.1 plugins, not microversions 00:08:37 <cyeoh> other than some other docs patches is anyone aware of something we need to merge for v2.1 urgently? (I know there are test cleanup etc) 00:08:49 <oomichi> so we can push microversion patches without the patch. 00:09:04 <gmann> cyeoh: oomichi : who about new attribute in V2.1 response 00:09:43 <cyeoh> gmann: do we have any in 2.1? 00:09:55 <cyeoh> for the keypairs v2.2 I know we do 00:10:10 <oomichi> gmann: ah, yeah. nice point. 00:10:34 <oomichi> cyeoh: v2.1 API has new attributes which doesn't exist v2 API. 00:10:47 <oomichi> cyeoh: so gmann is trying to remove them from v2.1 API. 00:11:06 <cyeoh> oomichi, ah, how many are there - do we have a list? 00:11:11 <oomichi> cyeoh: that means case2 on previous discussion. 00:11:14 <gmann> cyeoh: yes - https://review.openstack.org/#/c/156130/ 00:12:04 <oomichi> cyeoh, gmann: yeah, we need to pass v21-full job on the patch. 00:12:37 <gmann> cyeoh: oomichi : 2 of them i released -https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/v2-on-v3-api,n,z 00:12:40 <cyeoh> hrm. How did they get in? 00:13:13 <oomichi> cyeoh: https://wiki.openstack.org/wiki/NovaAPIv2tov3 00:13:28 <oomichi> cyeoh: please see "new attribute" on the wiki. 00:14:07 <oomichi> cyeoh: maybe we changed v3 attributes to v2.1 attributes even if they didn't exist on v2. 00:14:29 <cyeoh> are they all new or just renamed from v2? 00:14:55 <oomichi> cyeoh: all new 00:15:03 <oomichi> I guess 00:15:18 <gmann> cyeoh: oomichi : i also think all new. renamed one we already fixed 00:15:54 <gmann> cyeoh: oomichi while V2.1 conversion we considered that new att is no prob. 00:15:56 <oomichi> gmann: yeah, If missing renaming, tempest can detect it because of "required" on the schema. 00:16:11 <gmann> oomichi: yeah right 00:16:57 <cyeoh> it is backwards compatible so technically I guess ok 00:17:02 <cyeoh> would have liked to avoid it though 00:17:44 <cyeoh> wel also have the whole pci plugin which is only in v3 which i've bbeen trying to work out what to do with 00:18:28 <gmann> cyeoh: yes, that also adds new att in hypervisor and server 00:19:45 <oomichi> so I think v2.1 API should return the same response as v2 one without extra/new attributes, right? 00:20:25 <cyeoh> yes 00:20:32 <oomichi> even if these attributes are backwards compatible, we need to remove them from v2.1 API. 00:20:41 <cyeoh> that list is a very long list! 00:20:48 <gmann> oomichi: yeah 00:20:57 <oomichi> OK, got it :) 00:21:18 <gmann> cyeoh: list is long but i think its due to some common new att like acessip in all create server 00:21:31 <gmann> cyeoh: oomichi may be i will finish those by today 00:21:49 <oomichi> gmann: great! will review them soon after that 00:22:28 <gmann> cyeoh: oomichi : only PCI thing i want to discuss as cyeoh mentioned 00:22:54 <cyeoh> yea we need to fix these asap.... 00:23:14 <gmann> remove completely or keep only PCI controller which does not add new attributes? 00:23:15 <oomichi> PCI extension will be as new microversion ? 00:23:34 <cyeoh> I can help if we split some of them up 00:23:43 <cyeoh> gmann: so I think we should disable it 00:24:10 <gmann> cyeoh: thanks. 00:24:13 <cyeoh> we have a blacklist variable I think 00:25:15 <gmann> cyeoh: ok 00:26:16 <gmann> cyeoh: oomichi : after these 2 we can get more clear list - https://review.openstack.org/#/c/159352/ & https://review.openstack.org/#/c/159399/ 00:26:46 <gmann> Once they gets merged we can run https://review.openstack.org/#/c/156130/ again 00:27:10 <oomichi> gmann: ok, will review them after the meeting 00:27:15 <oomichi> gmann: thanks 00:27:22 <gmann> oomichi: thanks.. 00:28:15 <cyeoh> ok, gmann - could you put up an etherpad with things that need working on and we'll just pull them off the list as we start 00:28:28 <cyeoh> I'll start looking at the pci extension and the best way to disable 00:28:44 <gmann> cyeoh: sure. 00:28:46 <cyeoh> we'll then re-enable it as a microversion 00:29:21 <oomichi> cyeoh: +1 for enabling it on a microversion 00:29:45 <cyeoh> oomichi, yea I think thats safe since noone has been using it 00:30:12 <oomichi> cyeoh: so btw, how about https://review.openstack.org/#/c/140313/ ? 00:30:52 <oomichi> we are ready to go a first microvresion, I guess ;-) 00:31:39 <cyeoh> yea we're ready, but if these v2.1 things are easy/quick to fix I think we should do them first and aim for monday/tue to merge 140313 00:32:02 <oomichi> cyeoh: that is a clean way, +1 00:32:17 <gmann> cyeoh: yeah thats nice way. 00:33:01 <cyeoh> so other things which need review are: https://review.openstack.org/159075 00:33:12 <cyeoh> which does need to be fixed before some microversions can merge 00:33:58 <oomichi> cyeoh, gmann: can you write necessary patches on https://etherpad.openstack.org/p/v2_1_ReleaseCriteria ? 00:34:03 <cyeoh> and the novaclient one needs priority for reviewing. technically not required, but certainly very useful (can use rest api directly) and we want it in soon 00:34:14 <oomichi> cyeoh, gmann: we have a bunch of patches ;-) 00:34:53 <oomichi> and they would be necessary as criteria. 00:34:54 <gmann> oomichi: yes. 00:35:40 <cyeoh> yep! I guess we'll put the url for the review once we upload them. Hopefully we can get them merged fast 00:36:09 <cyeoh> might want to mention in the commit message that these came out of some recent tempest testing 00:36:31 <oomichi> cyeoh: yeah, agree. better to concentrate on them for fast merging 00:36:46 <gmann> cyeoh: oomichi : agree 00:37:31 <oomichi> please add me as a reviewer for this kind of patches 00:37:52 <oomichi> just after posting them 00:37:57 <gmann> oomichi: yeah 00:38:05 <cyeoh> definitely will, me too please 00:38:24 <oomichi> nice! 00:39:11 <cyeoh> ok, is there anything else we need to talk about? 00:39:28 <oomichi> one point: how to manage microversions? 00:40:14 <cyeoh> manage number allocation? 00:40:15 <oomichi> maybe we will have to manage these microversions patches in parallel 00:40:24 <oomichi> cyeoh: yeah, that is. 00:40:30 <cyeoh> sadly/happily they will merge conflict 00:40:39 <cyeoh> so will handle it like db migrations 00:40:48 <cyeoh> there will be last minute rebases 00:41:07 <oomichi> cyeoh: heh, interesting way ;-) 00:41:17 <oomichi> cyeoh: +1 for the way 00:41:26 <cyeoh> I think we should probably recommend using variables to define the min version 00:41:40 <cyeoh> that way the rebase should just be changing something in one spot 00:42:03 <oomichi> ok, I will avoid my -1 for versioning 00:42:11 <gmann> cyeoh: thats good idea. 00:42:45 <cyeoh> oomichi: yea I don't think we can auto-allocate versions 00:44:16 <oomichi> gmann: your 159352 is approved now, we can fix them soon I feel/hope ;) 00:44:28 <oomichi> cyeoh: I agree 00:46:35 <cyeoh> ok if there's nothing else might finish early and get on with the todo list? 00:46:51 <oomichi> yeah, nothing from me 00:47:39 <gmann> nothing from my side too 00:47:45 <cyeoh> ok, lets finish now then. thanks everyone! 00:48:06 <gmann> Thanks all 00:49:51 <oomichi> thanks! 14:00:55 <openstack> crinkle: Error: Can't start another meeting, one is in progress. Use #endmeeting first. 14:01:25 <mfisch> can we end someone else's meeting? 14:01:33 <crinkle> #endmeeting