15:00:04 #startmeeting gantt 15:00:05 Meeting started Tue Feb 24 15:00:04 2015 UTC and is due to finish in 60 minutes. The chair is n0ano. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:06 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:09 The meeting name has been set to 'gantt' 15:00:17 anyone here to talk about the scheduler? 15:00:42 \o 15:00:57 o/ 15:01:15 edleafe, are you a lefty too or are you just being ironic? 15:01:16 \o/ 15:01:34 I'm ambidextrous, actually 15:01:42 even better 15:01:58 write with right hand, mouse with left 15:02:23 my mother claims I started life as a lefty and then bowed to peer pressure and switched, before I remember things 15:02:51 lefty is smarter then righty :) 15:03:03 s/then/than :) 15:03:06 eliqiao, arguable :-)\ 15:03:11 anyway, let's get started 15:03:18 #topic Remove direct nova DB/API access by Scheduler Filters 15:03:18 Only left-handed people are in their right minds! 15:03:41 edleafe, I see you're finally getting all +'s on the spec, anything to talk about it? 15:03:42 ok, so bauzas finally added his +1 to the spec 15:03:55 \o 15:03:55 so I ran it by ndipanov and johnthetubaguy 15:03:58 * bauzas waves late 15:04:02 got a +2 from John 15:04:10 edleafe, saw that, good news 15:04:10 nick is still reviewing 15:05:02 I think it should be approved (finally) shortly so, unless you have anything, we can move on 15:05:22 Once that spec is approved, the WIP code can be changed to actual reviewable code 15:05:25 edleafe, fwiw I don't have +2 on that repo 15:05:39 patch series starting at https://review.openstack.org/#/c/149430 15:05:55 edleafe, but am unlikely to protest the spec now... as for the code... :D 15:06:02 ndipanov, you're not core for Nova? 15:06:03 ndipanov: I assume you're just being thorough :) 15:06:10 I am on nova 15:06:12 not on nova specs 15:06:22 that is the job of nova-drivers 15:06:26 nova-drivers are only +2'ing specs 15:06:33 ndipanov: ah, I thought you were core for specs, too 15:06:35 first among equals 15:06:52 guess I'm going to have to bug dansmith 15:07:03 edleafe, yeah 15:07:05 * n0ano thinks the beaurocracy is getting a little large 15:07:17 n0ano: getting? :) 15:07:19 n0ano: ssssht 15:07:20 thou I think if he sees +1s from me and jay you'd have to bug him a bit less 15:07:46 if need be we can bring it up at the nova meeting this week, let's hope we don't have to wait until then 15:07:52 ndipanov: but it's fun to bug Dan! 15:07:56 n0ano, yathink? 15:08:05 edleafe, that it is 15:08:11 :) 15:08:18 OK, let's move on 15:08:25 #topic Status on cleanup work 15:08:45 bauzas, I believe your patches are under control, need any help:? 15:09:42 n0ano: bp/detach-services is taking the most of my time 15:09:54 n0ano: but I'm rushing to provide bp/isolate-sched-db 15:10:08 n0ano: bp/detach-service deserves reviews 15:10:22 n0ano: and bp/isolate-sched-db is WIP 15:10:36 (once I'll fix all the demands on bp/detach-service) 15:11:03 n0ano: bp/requestspec-object is planned to be done on Liberty 15:11:10 #action review patches for https://blueprints.launchpad.net/nova/+spec/detach-service-from-computenode 15:11:11 because FF is Mar 3rd 15:11:30 March 5, no? 15:11:38 bauzas: 5? 15:11:44 edleafe: not exactly 15:12:04 branches are done on Thursdays but code has to be merged on Tuesdays in general 15:12:05 johnthetubaguy, said 5th to me 15:12:37 that's a matter of trusting Zuul for merging all patches close to a FF 15:12:40 https://wiki.openstack.org/wiki/Kilo_Release_Schedule 15:12:50 says 5th March for Feature Proposal Freeze 15:13:07 I'm usually targeting Tuesdays for deliveries, even if branching is done by Thursdays 15:13:37 anyway, you got the idea 15:13:37 bauzas, nothing has to be merged - just uyp for review 15:13:45 s/uyp/up 15:13:59 bottom line, early next week 15:14:10 yuy 15:14:33 eh, hold on 15:14:40 bauzas: there is no branching or tagging for FeatureProposalFreeze, but thats true for FeatureFreeze 15:14:57 bauzas, FF is 19th 15:14:59 johnthetubaguy: right, I confounded FPF and FF 15:15:20 so yeah, let's plan to rush on the patches by mid-next week 15:16:04 Time for an update from PaulMurray ? 15:16:14 sure, go for it 15:16:28 btw. I'm taking the opportunity here to ask people to amend https://etherpad.openstack.org/p/kilo-nova-priorities-tracking and set the correct statuses on their ownerships for Launchpad BPs 15:16:34 hey n0ano - I have one patch I hacked to bring up (sorry not following along fully) is it appropriate now 15:16:41 after paul that is 15:17:01 ndipanov, sure, no fast rules here, but PaulMurray has the floor for now 15:17:09 there are a few patches for review here: https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/make-resource-tracker-use-objects,n,z 15:17:44 I have one ready for review that starts a (small) series on changing compute node to object 15:18:07 lxsli, has some migration object patches that could do with rebase and work 15:18:32 those are in a chain, starting with Jay's numa patch going through sahid's 15:18:51 thanks - just havin trouble typing that 15:18:55 PaulMurray, lxsli do you need help with that rebase/work or are reviews sufficient? 15:18:59 lxsli: but jaypipes seems to have problems with his own patch 15:19:13 yes getting hay's patch merged is prio #1 15:19:21 lxsli: so I guess it's impacting your chain, right ? 15:19:45 he mentioned some issue in nova room, sahid said he'd look at that 15:19:54 lxsli: yeah, I understand it but Jenkins seems to be a badass by sending continously red tomatoes 15:20:07 I'm working on rebasing sahid's patch to master atm so he has an easier time rebasing it onto jay's patch when it's ready 15:20:11 lxsli: cool 15:20:26 I have quite some IRC lag atm so apologies for typos 15:20:53 lxsli, NP, the typos are minor so far 15:21:02 as https://review.openstack.org/#/c/148904/ is sitting ready it would be good to get it through - it moves a couple of methods around so it tends to get hit by rebases 15:21:41 there are two more follow-ons fro that then compute node will be done 15:22:25 My aim is to get rid of references to conductor api with everything going through objects so online updates can work 15:22:53 Not worry so much about making all the code perfect as it will get changed soon anyway with changes comming to RT 15:22:58 and scheduler etc. 15:23:23 PaulMurray, I agree not to worry, we'll just change in future as needed 15:24:10 But of course, my code will be perfrect.... :) 15:24:14 PaulMurray: seen some patches from Hans Lindgren ? 15:24:24 PaulMurray: lemme find those 15:24:30 PaulMurray, so, in general, sounds like you're making good progress, anything else? 15:25:05 n0ano, actually - been on vacation - so progress will be better than was 15:25:10 PaulMurray: https://review.openstack.org/#/c/149224/ 15:25:38 vacation - I've heard of that, I should try one sometime :-) 15:25:48 n0ano: take a French resident passport 15:25:50 bauzas: that is abandoned 15:25:59 lxsli: orly ? 15:26:06 bauzas, that's out of date by the look of it - should be abandoned. 15:26:07 are there others which are active? 15:26:07 lxsli: indeed, I missed that point 15:26:27 lxsli, oh -your right - is abandoned 15:26:48 lxsli: https://review.openstack.org/#/c/133663/ also 15:26:51 bauzas: because it was covered by one of your patches no less :) 15:27:02 lxsli: right 15:27:25 lxsli: I probably made 15 patches in my series, so I'm sometimes loosing memory 15:27:47 bauzas, thanks - I had better keep an eye out for things comming from left field 15:28:02 OK, let's move on a bit... 15:28:03 PaulMurray: I have severe delays in reviewing 15:28:07 ndipanov, you wanted to say something? 15:28:27 ah yes 15:28:54 so I was working on the bug 15:29:20 where we don't consider the right flavor for migrating with NUMA 15:29:31 https://bugs.launchpad.net/nova/+bug/1370390 15:29:32 Launchpad bug 1370390 in OpenStack Compute (nova) "Resize instace will not change the NUMA topology of a running instance to the one from the new flavor" [Medium,In progress] - Assigned to Nikola Đipanov (ndipanov) 15:29:42 and came up with this https://review.openstack.org/#/c/158245/ 15:29:57 which kinda steps into request_spec teritory 15:30:13 but is only relevant inside the filter code 15:30:13 ndipanov: right, I quickly reviewed your patch 15:30:35 I mean I CAN fix the bug without it 15:30:51 ndipanov: anyway, I think that we need to rework on the RequestSpec BP, because there was some good comments on the implementation 15:30:53 but this would be a nice cheap win for moving all the filters to use flavor objects 15:31:14 bauzas, this really makes no assumptions about that other than it will be an object at one point 15:31:23 that will expose this one method 15:31:46 the reason I need this one method that tells me if the flavor is changing 15:32:37 is that we can tell when we don't need to recalculate stuff 15:32:51 like I said - not critical to go this route 15:32:59 ndipanov: right, the problem is that we're dehydrating the Flavor object when sending it to the wire 15:33:16 ndipanov: we should just pass to the Scheduler what we already have 15:33:25 yeah that's not a problem for me for now but agreed 15:33:28 it should all be objects 15:33:34 ndipanov: cool that you found a workaround 15:33:47 so basically 15:34:05 I'd like a review and ack that this is not insane from sched folks and dansmith if possible 15:34:20 but not sure if I can put up the whole series this week since I am a bit swamped 15:34:46 if not - ah well, but I think moving to flavor objects in the scheduler with this is an easy win unless I missed an angle 15:35:01 ndipanov, so you're asking for a review of your patch 158245, right? 15:35:12 well more like - 15:35:27 try to see the approach 15:35:35 and tell me if it;s OK to land it for a bugfix 15:35:43 and get some refactoring early and cheap 15:36:10 ndipanov: ack, will give you my voice 15:36:21 next patch in that series moves all use of flavor in filters to objects 15:36:28 at this point I am hesitant about refactoring at this late stage in Kilo, even if it's easy 15:36:29 but still failing tests so not posted 15:36:38 n0ano, that is reasonable 15:36:43 * PaulMurray has to drop off now - sorry 15:36:51 ndipanov: +1 15:36:58 PaulMurray, NP, just read the logs 15:37:04 but we;re still not that late from priorities pov 15:37:07 so think about it 15:37:26 ndipanov, OK, we'll check it out 15:37:50 moving on... 15:37:55 #topic opens 15:38:01 anything new for today? 15:38:57 hearing crickets... 15:39:02 I got one 15:39:08 bauzas, go ahead 15:39:34 basically, restating what I said in the crowd 15:39:35 (16:16:28) bauzas: btw. I'm taking the opportunity here to ask people to amend https://etherpad.openstack.org/p/kilo-nova-priorities-tracking and set the correct statuses on their ownerships for Launchpad BPs 15:40:24 ie. NeedsCodeReview, Slow Progress or In Progress 15:41:20 bauzas, +1 15:41:44 anything else? 15:42:06 nothing really much 15:42:28 OK, so, as always, lots of reviews needed, everyone sharpen your pens and we'll talk next week. 15:42:33 tnx for attending 15:42:40 #endmeeting