15:00:15 #startmeeting nova_scheduler 15:00:16 Meeting started Tue Jun 16 15:00:15 2015 UTC and is due to finish in 60 minutes. The chair is bauzas. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:17 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:19 The meeting name has been set to 'nova_scheduler' 15:00:26 hola folks 15:00:48 o/ 15:00:54 bye 15:00:55 o/ 15:01:03 n0ano is unable to make the call, so I'm chairing for the best or the worst 15:01:17 feel free to buy tomatoes before we start 15:01:19 * edleafe is in another meeting on the phone 15:01:33 so sorry if I lag response time 15:01:46 edleafe: eh, wasn't my turn last week ? :) 15:02:15 okay, waiting a few more time for guests coming in, and then we can start 15:02:27 bauzas: :) 15:03:25 okay, guessing that 3 people is definitely good for having consensus, we can start 15:03:40 (at least, it is easier than 4) 15:03:49 (and better than 2) 15:04:02 #topic Spec tracking 15:04:17 so n0ano was wondering how to track our progress 15:04:33 from my perspective, we have 2 kind of things 15:04:35 didn't he create an etherpad or wiki page? 15:04:42 1/ things that are priorities 15:04:48 2/ things that aren't 15:05:11 edleafe: yup, but he recently raised the question about that during the last nova meeting 15:05:22 edleafe: and I was -1 about that 15:05:51 so, as I said, there are 2 possibilities 15:06:01 either you work on a prioritized spec 15:06:07 (or you claim for it) 15:06:08 suggestion for tracking these things? 15:06:12 or, you don't 15:06:29 so, given that, we already the Master Doom of the Etherpads for Nova 15:06:38 An etherpad to track our Liberty commitments and process sounds useful; in addition to the liberty-nova-priorities list where we may put 3 items for review 15:06:47 #link https://etherpad.openstack.org/p/liberty-nova-priorities-tracking 15:07:14 MHO is that we can add our specs there, either it's a prio or not 15:07:42 L29 is for Scheduler priority specs 15:07:58 just added my teeny spec 15:08:19 so, I'm thinking we should just add non-prio specs in another Scheduler bullet at the bottom 15:08:23 johnthetubaguy: are you around ? 15:08:51 johnthetubaguy: we have some specs that are priorities or claiming for so we can add them to https://etherpad.openstack.org/p/liberty-nova-priorities-tracking 15:08:58 We can add some extra things which aren't ready to be reviewed quite yet 15:09:14 I wouldn't like to clog it up with everything we're thinking of though 15:09:17 johnthetubaguy: but some specs couldn't be priority, so my gut is that we should add a Scheduler subteam item at the bottom 15:09:29 lxsli: as long as they are in a separate section 15:09:32 (and explicitely say those are not prio) 15:09:44 lxsli: we can move them around when they are ready for review 15:09:56 I think having a single scheduler subteam area is easiest if possible 15:10:05 lxsli: edleafe: +1, I want it very clear that the first section (L17) is only for those we consider mergeable 15:10:10 yeah, just have a separate ready vs needs review by sub team lists 15:10:14 johnthetubaguy: without a separate etherpad? 15:10:17 you can have non priority stuff separate if you want 15:10:29 johnthetubaguy: okay, I just want to make it clear that not all the specs are priority 15:10:29 lxsli: L17 is for the API 15:10:42 lxsli: you mean L35 15:10:50 lxsli: yeah, please not more etherpads, people don't seem to look at the current one at the moment, so I think adding more will be counter productive 15:10:55 edleafe: yup 15:11:21 bauzas: the blueprint and spec deadline is super soon, specs that are not up now are very unlikely to make the deadline at this point 15:11:45 johnthetubaguy: my point 15:12:05 johnthetubaguy: I mean, during Liberty, we will have 2 kind of changes 15:12:10 so yeah, I would focus on making sure you have the specs merged that you need, and focusing your view on those right now 15:12:12 johnthetubaguy: (speaking of changes, here) 15:12:31 johnthetubaguy: edleafe: lxsli: I'm just amending the etherpad now 15:13:35 cool, thanks for pushing on this folks 15:13:47 the idea is to separate non-prio changes from the prio ones, to help both nova cores and sched subteam to get the proper attraction for prio changes, while we still want to keep focus on non-prio 15:14:06 okay, I'm taking an action to refine a little bit the sections 15:14:36 #action bauzas to amend the etherpad to make clear what is a priority change vs. what is not a priority change 15:14:47 let's move on then 15:14:56 johnthetubaguy: thanks for your insights on that 15:15:12 #topic Liberty tracking 15:15:54 so, before going further, do we agree on using https://etherpad.openstack.org/p/liberty-nova-priorities-tracking for tracking ? 15:16:12 #startvote do we agree on using https://etherpad.openstack.org/p/liberty-nova-priorities-tracking for tracking? yes or no 15:16:14 Begin voting on: do we agree on using https://etherpad.openstack.org/p/liberty-nova-priorities-tracking for tracking? Valid vote options are yes, or, no. 15:16:15 Vote using '#vote OPTION'. Only your last vote counts. 15:16:17 sounds like that's what john prefers 15:16:19 #vote yes 15:16:29 edleafe: ? 15:16:48 #vote yes 15:17:25 I would love if edleafe could give his view on that, even if I think we got his idea :) 15:17:30 ed's on the phone 15:18:01 right, that's why I make edleafe's IRC client blinking like Xmas 15:18:04 :) 15:18:17 heh 15:18:23 ok, let's close the vote 15:18:26 #vote yes 15:18:27 #endvote 15:18:27 Voted on "do we agree on using https://etherpad.openstack.org/p/liberty-nova-priorities-tracking for tracking?" Results are 15:18:28 yes (3): bauzas, edleafe, lxsli 15:18:32 woooh 15:18:35 whew, just made it! 15:18:42 edleafe: heh :) 15:18:54 ok, so, given that 15:19:19 #action all to update https://etherpad.openstack.org/p/liberty-nova-priorities-tracking to mention their changes if necessary for tracking 15:19:50 #info changes that are not given in the etherpad list are not considered for tracking yet 15:19:59 do you agree, guys ? 15:20:20 yep 15:20:33 I mean, I have the RequestSpec object patch series that is up for review, but I don't consider it yet good for being tracked 15:20:47 What's the significance of being tracked? 15:21:07 lxsli: being discussed periodically by the team to make sure we make our deadlines ? :) 15:21:23 so there's some kind of promise of delivery? 15:21:30 yup 15:21:36 OK, thanks 15:21:59 that doesn't mean that we can't discuss on some spec implementations now, but I really want to use the etherpad for things that are needing reviews 15:22:58 but let's loop back with n0ano next week 15:23:50 One thing I wanted to mention 15:24:00 given that, does anyone want to discuss about a specific spec or implementation they want to share ? 15:24:03 edleafe: fire it 15:24:06 If you review one of these specs and give a -1, please be sure to follow up 15:24:18 edleafe: wise words 15:24:21 n0ano gave me a -1 with some questions 15:24:29 and then left it all last week 15:24:38 as a result, no one reviewed the spec on Friday 15:24:46 bauzas, I guess it doesn't apply to specs that necessarily don't fall under the priority list? Is there a way out for specs that are deemed essential for someone beyond the ones listed? 15:24:56 edleafe: don't hesitate to bug people if you don't get feedback rapidely 15:25:02 bauzas: +1 15:25:11 edleafe: on my own, I seriously lag from reviews atm 15:25:38 bauzas: I lag reviews, but I always star those I give a -1 to so I can follow up 15:25:41 sudipto: we just agreed on tracking those in the etherpad too, so you could get attention from the sched team 15:25:51 bauzas, ok sure... 15:26:01 sudipto: https://etherpad.openstack.org/p/liberty-nova-priorities-tracking L43 15:26:11 bauzas, thx! 15:26:20 edleafe: fair point, but you can't assume everybody does that :) 15:26:44 we are a small team, it's reasonable to bug people if so 15:26:54 bauzas: hence my reminder :) 15:27:04 edleafe: cool 15:27:22 any spec/bp/change that we want to discuss now ? 15:27:46 I'm happy with Jay's resource objects spec, it'd be great to get some more reviews on that now 15:28:12 I could just tell that I'm suffering severe PITA with the ReqSpec object BP, but I guess you see my complaints all the day on IRC 15:28:22 lxsli: in my queue 15:28:32 lxsli: I saw you squashed the 2 specs, great and ty 15:28:47 bauzas: yep always the idea, just needed to work out the kinks with jay first 15:28:54 cool 15:29:04 moving on then ? 15:29:23 edleafe: I saw you were proposing a log information for NoValidHost ? 15:29:30 that's on my queue too 15:29:43 yep 15:29:51 edleafe: you could put it as non-prio stuff (or prio, don't remember your claim) 15:30:13 I had originally tied it to the ReqSpec, but since that isn't going to be used for filtering in Liberty, I simplified it 15:30:29 It was a priority from the summit, IIRC 15:30:34 edleafe: I missed that point, I'll take some time to decently -1 it 15:30:35 :p 15:30:50 (just kidding) 15:30:50 bauzas: I expect nothing less! 15:31:08 ok, moving on 15:31:21 #topic New meeting time 15:31:53 so I was horribly lagging last week, was there any consensus reached ? 15:32:21 no, since there was no single time where everyone was available 15:32:22 or are we just doomed because of a stupid French wanting to take some time with his daughters ? 15:32:44 edleafe: so, what was the outcome ? 15:32:46 we are doomed because stupid jaypipes wants to go to his meetings :) 15:32:50 They should move America east a bit 15:33:13 n0ano just punted 15:33:16 things would be far easier if anyone could just use the Paris TZ 15:33:20 * bauzas whispers 15:33:25 * edleafe wonders if 'punted' translates well... 15:33:56 ok, so we're stuck ? 15:34:03 should we think about alternating ? 15:34:27 even weeks using that timeslot, and odd weeks using a new timeslot ? 15:34:37 bauzas: there is another option 15:34:55 edleafe: I'm hanging at your lips 15:35:03 (looking it up...) 15:35:08 (don't know if that translates well too...) 15:35:16 1600 UTC was ok for everyone except johnthetubaguy 15:35:27 he's not that important 15:35:35 +1 15:35:36 thinking about a rebellion ? 15:35:44 lxsli: heh, I'm sure he'll be glad to hear that :) 15:35:52 "coup, coup" tweeted the pigeons 15:36:03 I have a regular meeting at that time, but its moving soon anyways 15:36:04 don't forget we cut our King's head 15:36:06 but thats fine 15:36:16 but those were the only two times that jaypipes was available 15:36:16 its not on IRC so I can be vaguely around for both 15:36:24 I would go for that one 15:36:40 * johnthetubaguy wonders where in the agenda to raise his crazy spec and docs patches 15:36:50 the other option is to move it to monday or thursday 15:37:05 johnthetubaguy: it was during the previous topic, but let's discuss that during open time 15:37:22 bauzas: yeah, I figured I missed that, cool 15:37:40 bauzas: how about I ping jaypipes for his availability, and then create a new doodle based on that? 15:37:41 we only have that topic and then we open the discussion 15:37:47 edleafe: huge +1 15:37:58 seems like we can just go for 1600UTC though? 15:38:16 #action edleafe to ping jaypipes for availability and start new doodle for meeting time 15:38:22 ignore my no vote on that, I can lurk like I normally do just fine at that time, if it helps 15:38:28 johnthetubaguy: yes, but jay was a maybe for that 15:38:35 edleafe: ah, gotcha 15:38:38 edleafe: not sure you logged the action 15:38:42 johnthetubaguy: so you're not the only problem :) 15:38:45 #chair edleafe 15:38:46 Current chairs: bauzas edleafe 15:39:00 edleafe: action should be to find out the issue with 1600 first then 15:39:01 bauzas: yeah, actions and links can be posted by anyone 15:39:13 lxsli: agreed 15:39:17 #undo 15:39:18 Removing item from minutes: 15:39:32 ooh undo, fancy 15:39:46 #action edleafe to ping jaypipes for his issue with Tuesday 1600 UTC 15:39:53 edleafe: before we agree on that, could we just consider using the 1600UTC slot and check with jaypiprsd 15:39:58 ? 15:40:01 yeah, that one 15:40:16 I missed the new doodle thing when I shouted "hell yea" 15:40:23 bauzas: is there a meeting room available? 15:40:46 edleafe: we can sort that out quickly, or complain about that to ttx if no 15:40:54 edleafe: I'm not that worried 15:41:03 I remember the 1700 was open 15:41:08 not sure about 1600 15:41:11 edleafe: the #openstack-meeting room was sexy, but I can leave with that 15:41:12 shall we move on? 15:41:21 lxsli: +1, that can be figured out later 15:41:25 #topic open discussion 15:41:34 johnthetubaguy: had a status to share ? 15:41:44 #link https://review.openstack.org/192260 15:41:57 so I had an action to write up a summary of the scheduler evolution plans 15:42:03 thats a first draft of that effort 15:42:17 oh cracking, I was thinking of doing that but not looking forward to it 15:42:19 help extremely welcome, like please take over that patch and make it better 15:42:28 any one other thing tooo 15:42:28 johnthetubaguy: okay, are you fine with me providing some updates ? 15:42:36 bauzas: please do, that would be awesome 15:42:45 johnthetubaguy: sure, that's on my todo list 15:42:53 starring it even :) 15:43:16 so I wrote up some stuff around the parallel scheduling effort too 15:43:22 johnthetubaguy: ooooooh 15:43:28 based on the summit session, and stuff I was mumbling about during that session 15:43:31 #link https://review.openstack.org/#/c/191914/ 15:43:35 its a backlog spec 15:43:35 johnthetubaguy: saw ndipanov's proposal btw ? 15:43:46 bauzas: yeah, this is mostly unrelated it turns out 15:43:53 so its a backlog spec 15:43:59 so its collecting a problem definition 15:44:07 and its talking about alternatives 15:44:15 johnthetubaguy: cool, I should probably say the same that for the last spec 15:44:16 I am not trying to pick a winner, or assign the work 15:44:19 s/spec/devref 15:44:22 just doument the ideas 15:44:28 sure, that's a backlog spec 15:44:42 somehow, its over 400 lines long 15:44:43 oops 15:44:46 okay, that's pretty fancy 15:44:47 johnthetubaguy: curious - backlog specs don't go in nova-specs? 15:45:02 edleafe: they do, but on a separate path 15:45:09 edleafe: http://specs.openstack.org/openstack/nova-specs/specs/backlog/index.html 15:45:46 ok, that's cool, we could even put those both changes in the etherpad for reviews 15:45:59 johnthetubaguy: just confused, since your spec is in nova, not nova-specs 15:46:07 edleafe: hell no 15:46:10 johnthetubaguy: haven't created a backlog spec myself 15:46:16 edleafe: one is a nova doc on architectural evolution 15:46:21 edleafe: https://review.openstack.org/#/c/191914/ is in nova-specs 15:46:21 edleafe: the other is in the nova-specs backlog specs 15:46:35 edleafe: https://review.openstack.org/192260 is a devref doc 15:46:42 johnthetubaguy: ah, I see 15:46:43 huh, jinxed 15:46:55 edleafe: the nova doc stuff is designed to span multiple nova-specs, giving a high level overview, then pointers once thats possible 15:47:07 johnthetubaguy: I saw the .rst and assumed it was a spec 15:47:09 edleafe: this is all very shiny and new process right now 15:47:25 edleafe: ah, easy done! 15:47:27 johnthetubaguy: too shiny - it blinded me! :) 15:47:42 :) 15:47:49 I keep -2ing specs because they don't have a spec merged, so I can't throw stones 15:48:13 johnthetubaguy: thanks a lot for your proposals, we'll follow up on those 15:48:27 so there are other actions folks have from the summit 15:48:36 it would be good to update the etherpad when they are done 15:48:44 #link https://etherpad.openstack.org/p/YVR-nova-liberty-summit-action-items 15:48:46 #link https://etherpad.openstack.org/p/YVR-nova-liberty-summit-action-items 15:48:49 #undo 15:48:50 Removing item from minutes: 15:49:24 johnthetubaguy: yeah we're just starting to collect feedback 15:49:59 johnthetubaguy: given all the things we are, I'm at least wanting to have some design discussions for things that are in the list but not yet with a clear plan 15:50:13 johnthetubaguy: so we could propose and implement those for Muppet cycle 15:50:51 bauzas: I really want backlog specs for that stuff 15:50:55 johnthetubaguy: +1 15:51:09 bauzas: so folks outside the usual groups can be more invovled 15:51:33 johnthetubaguy: that's what I say "design discussions", translate it to 'draw a draft' 15:51:47 johnthetubaguy: which is even one step further than a backlog spec 15:52:20 bauzas: perhaps 'requirement discussions' 15:52:23 johnthetubaguy: +1 on welcoming new contributors, hence the idea to open up the reviews to things that people want to share with us 15:52:30 bauzas: just agreeing the problem in a backlog spec I think is good step forward 15:52:54 I got your idea :) 15:53:06 collect feedback, iterate over that 15:53:44 so if someone has an outlandish new approach, a backlog spec would be the place to propose it? 15:53:57 edleafe: a backlog spec is for stating a problem 15:54:19 bauzas: not for proposing solutions? 15:54:30 edleafe: technically, there is no section for that IIRC 15:54:55 edleafe: but that doesn't mean we're avoiding to think about the solution 15:55:03 hmmm... I thought that if there was a proposed solution that couldn't be implemented in the current cycle, it went into the backlog 15:55:07 so you can do both 15:55:10 aiui backlog specs you don't plan to implement yourself 15:55:25 so I am about to update the doc page on this 15:55:26 so you can propose a solution but it might not get implemented that way 15:55:39 but basically, if you miss out any section of the spec 15:55:41 its on the backlog 15:55:49 lxsli: sure. Propose a solution, and let others have at it 15:55:51 that include assignee or implementation 15:55:54 maybe this discussion is wider than just for our subteam ? :) 15:56:17 bauzas: yeah - 3 minutes left 15:56:18 bauzas: sure, I thought it was already "agreed" though, as defined here: http://specs.openstack.org/openstack/nova-specs/specs/backlog/index.html 15:56:37 johnthetubaguy: let's discuss that back on #openstack-nova, I guess some people could voice there 15:56:49 sounds good 15:57:10 (and you would get better exposure for your thoughts) 15:57:23 ok, any other topic for the 2 mins left ? 15:57:24 Anything else then/ 15:57:49 crickets ? 15:58:22 fair, let's call the wrap 15:58:26 #endmeeting