14:00:14 #startmeeting nova_scheduler 14:00:15 Meeting started Mon Jul 6 14:00:14 2015 UTC and is due to finish in 60 minutes. The chair is n0ano. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:16 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:19 The meeting name has been set to 'nova_scheduler' 14:00:24 anyone here to talk about the scheduler? 14:00:34 o/ 14:00:59 * alex_xu found the meeting is changed to better time~ 14:01:02 lxsli, now's your chance :-) 14:01:05 Scheduler 2: scheduler rides again 14:01:49 alex_xu, glad at least one person likes the new time (a little early for me but I can handle it). 14:02:29 n0ano: heh, how the vote out without person glad 14:02:38 mornin. 14:02:55 \o 14:03:33 * bauzas is also happy to see the meeting earlier :) 14:03:55 * n0ano is a little bleary eyed but NP 14:03:57 it just means that US people will be just awake, so we can try to bypass them \o/ 14:04:18 ha 14:04:29 OK, let's start, hopefully edleafe & Paul will catch up 14:04:39 #topic liberty specs 14:04:53 n0ano: PaulMurray didn't make a promise about that one 14:05:05 PaulMurray won't make it 14:05:18 He asked me to mention his compute node patch has finally landed, woot 14:05:29 lxsli, that's excellent news 14:06:14 looking at the priority tracking page looks like 179224 (migration) is the only one not approved, is that on track? 14:06:55 my bad, that one and the request spec object are still not striked out 14:07:12 n0ano: the spec is a priority one 14:07:28 n0ano: as the freeze is only for non-prio specs, we don't have to be afraid of it 14:08:00 n0ano: jaypipes liked it, I hope to see a bit more sponsors by the next days once the non-prio specs are freezed 14:08:15 looks like there's recent comments on this and no -1's so hopefully it's get +2 soon 14:08:24 n0ano: but for the moment, I leave all the reviewers looking at non-prio ones 14:08:55 about the reqspec implementation, well, it deserves reviews 14:09:12 bauzas: I had a look, few small comments but generally +1 14:09:23 lxsli: coolness 14:09:40 so, re the patch series the left things are: 14:09:49 1/ fix the unittests for the WIP patch on filters 14:10:09 2/ change the sched RPC API and provide a method using a ReqSpec object 14:10:34 3/ change the conductor to hydrate the Spec there instead of using build_req_spec() 14:10:51 so, good progress, I hope to put it all done by EOW 14:10:58 and then, review time 14:11:08 and iteration and upload dance 14:11:32 so sounds good, you don't need any help on this do you (I have people that are available) 14:12:17 n0ano: nope, no neede 14:12:20 need 14:12:40 n0ano: tbh, it's a huge change 14:12:56 bauzas, NP (I'm one of the few people in my company that don't think 9 women can make a baby in 1 month) 14:13:20 he 14:13:51 * bauzas takes the opportunity to thank SublimeText and Ctrl-D 14:14:02 sounds to me like the specs are well in hand so, unles someone has an issue, let's move on 14:14:22 #topic Nova mid-cycle meetup 14:14:43 the sprint is in 2 weeks, have we thought about scheduler issues we want to bring up there? 14:15:56 the only one I can think of is splitting the scheduler out, we should have the APIs well defined by Liberty so we should seriously think about what needs to be done for the split in M 14:15:56 I should hassle johnthetubaguy with an expected agenda so I could see where I could lurk 14:16:18 bauzas, I checked the etherpad for it today and it's currently completely empty 14:16:45 looks like no one has thought about the meetup yet. 14:16:53 n0ano: I'm not that worried, but I will appreciate any attempt to clear that out for remote attendees :p 14:16:56 n0ano: I'd like to discuss how we model flavor and image metadata during the launch request. not sure that's a real scheduler-related thing. kinda. 14:17:07 jaypipes: I like this 14:17:28 jaypipes, not a bad idea, certainly related to scheduling but probably more a Nova API issue 14:17:30 jaypipes: we have many "need an owner/plan" including the AZs stuff 14:17:45 jaypipes: I would appreciate any think tank by the midcycle time about those stuff 14:18:21 bauzas: I was hoping tdurakov could own the AZ stuff. but I agree with you we need to discuss it. 14:18:32 jaypipes: yeah, I know your spec 14:18:45 jaypipes: what I would like to see is some agreement by the midcycle 14:18:50 - even if I'm not here :( 14:19:00 I mean, there 14:19:06 bauzas: sure, agreed. it's on my list. 14:19:14 n0ano: +1 to flavor being / becoming a Nova API issue 14:19:37 jaypipes: the hints definition is also not yet agreed 14:19:57 jaypipes: I mean, the hints taxonomy 14:20:09 bauzas: what do you refer to? 14:20:13 I remember alaski having concerns about that 14:20:22 jaypipes: sec, giving you the mail thread 14:20:33 bauzas, do we need to address the persistance of hints, I seem to remember someone working on that in the past 14:20:41 alaski: prefer not use json-schema validae the hints 14:21:08 jaypipes: that's pretty related to how someone is describing something like a flavor or image, but to the difference it's for end-users 14:21:21 n0ano: that's not exactly the problem 14:21:26 I'm fine with validation of existing hints, as long as additionalProperties is True 14:21:28 n0ano: we don't disagree persisting the hints 14:21:40 bauzas, no but I think persisting the hints needs to be done 14:21:51 n0ano: that's planned :) 14:22:01 n0ano: there is no problem with that 14:22:04 well, I'm not totally fine with it. but am growing less adamant over time 14:22:25 sec, a mail thread is better than me trying to explain 14:22:48 #link http://lists.openstack.org/pipermail/openstack-dev/2015-June/067996.html 14:22:58 alaski: could you be more specific on what you're not totally fine with? I'm having a had time understanding precisely what we are talking about here. 14:23:11 ahhhhhhhhhhhhhhhh 14:23:12 alaski: like I said, some grey area is still there, that's not only about API validation 14:23:14 it's in the mail bauzas linked 14:23:22 alaski: I owe you a respond on that thread. sorry man :( 14:23:28 np 14:23:34 jaypipes: I owe that too :( 14:24:23 so, yeah, something like AZs, hints, flavor/image definitions and reviewing sprint for the ReqSpec would be awesome 14:24:32 #action all to cogitate on and reply to alaski mail thread on hints 14:25:31 bauzas, good, I'll update the etherpad with those to begin with (maybe I can get a first post :-) 14:25:34 * edleafe just woke up after 3 hours of sleep 14:26:11 edleafe, that's what you get for agreeing to this time :-) 14:26:22 jaypipes: like I said previously, even if not attending the midcycle, I'll be set on Rochester TZ 14:26:30 kk 14:26:49 jaypipes: so it means it will be possible to do some sprint for the ReqSpec if necessary 14:26:54 n0ano: no, that's what I get for flying the red-eye home from holiday :) 14:27:09 edleafe, ouch 14:27:57 OK, let's think about the mid-cycle and talk again about it next week, the sprint is only 2 weeks away 14:28:13 #topic opens 14:28:18 anything new from anyone? 14:28:59 * johnthetubaguy notes the need for some kind of remote co-ordination with bauzas 14:29:16 johnthetubaguy: don't worry, I'll be online 14:29:35 johnthetubaguy: so, provided some help can be done, I'll be there 14:29:43 johnthetubaguy, will there be a possibility of a google hangout with video at the meetup? 14:29:48 as a heads up, I have been thinking about a few totally remote follow ups post the mid-cycle if we need those 14:30:11 n0ano: we hope so, although the room logistics don't sound perfect for that kind of thing 14:30:35 johnthetubaguy, believe me, I understand, we'll just have to do the best we can 14:31:13 n0ano: +1 14:31:27 anything else? 14:31:31 n0ano: I'm already pretty silent during the summits, I actually just need to hear :) 14:31:59 n0ano: if I hear something that I dislike, I can hassle you guys directly in IRC 14:32:00 :p 14:32:07 johnthetubaguy: ^ 14:32:07 silent? are we talking about the same bauzas I know :-) 14:32:30 hah 14:32:45 don't make me wrong, I can talk about cheese for like one hour 14:33:00 better cheese than wine 14:33:09 * bauzas thinks he's seriously diverting 14:33:17 indeed 14:33:18 gotta run o/ 14:33:19 bauzas: I have a better IRC client now, so that should be possible 14:33:31 OK, I think we're done for today, talk again next week 14:33:34 tnx everyone 14:33:35 ++ 14:33:40 #endmeeting