13:59:57 #startmeeting nova_scheduler 13:59:58 Meeting started Mon May 2 13:59:57 2016 UTC and is due to finish in 60 minutes. The chair is edleafe. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:59:59 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:01 The meeting name has been set to 'nova_scheduler' 14:00:14 is anyone here today? Or still recovering? 14:00:29 o/ 14:00:44 jet lag again 14:01:05 o/ 14:01:12 Yingxin: I'm in the same time zone and I feel like I have jet lag 14:01:48 OK, I'm just finally sitting down at a keyboard for the first time since the summit, so I'm totally unprepared for this meeting. 14:01:54 I think it will be quick 14:02:10 #topic Specs / Reviews 14:02:27 Anything new to bring attention to? 14:02:28 edleafe: I'm cleaning up all resource-providers specs for a final review. 14:02:37 jaypipes: cool 14:02:40 ^ is what i wanted to know 14:02:48 edleafe: I am also completing a new one for dynamic resource classes based on the discussion with Ironic and FPGA folks. 14:03:05 edleafe: should have them all pushed by this afternoon. 14:03:14 jaypipes: yeah, that sounds like a more difficult task 14:03:23 hopefully we can get generic-resource-pools and resource-provider-allocations specs merged this week 14:03:29 mriedem: ++ 14:03:32 ++ 14:03:36 ++ 14:04:00 jaypipes: I'll be sure to review the qualitative / capabilities stuff 14:04:07 edleafe: danke 14:04:38 Sooooo... anything else on specs? 14:04:53 edleafe: yes 14:05:03 jaypipes: the floor is yours 14:06:05 edleafe: never mind. lemme finish this work first before adding any more to my plate ;) 14:06:25 jaypipes: awwwww 14:06:36 #topic Opens 14:06:55 Anything else? Or should we get back to digging ourselves out? 14:07:01 i've got a question, sec 14:07:04 sure 14:07:07 oh, hey, I'm like here and stuff 14:07:20 claudiu's re-proposed hypervisor capabilities spec https://review.openstack.org/#/c/286520/ 14:07:34 i'm holding off on a fast re-approval of that given the discussion about capabilities at the summit 14:08:12 jaypipes: are there competing specs for the capabilities stuff now? or is https://review.openstack.org/#/c/286520/ still valid? or 'don't ask me right now'? 14:08:18 mriedem: added to my review list 14:09:32 mriedem: I'm trying to add standardize-capabilities blueprint before that one. 14:09:54 ok, that makes sense 14:11:29 i don't have anything else. top priorities seem to be getting those two specs cleaned up, since the code, or part of it, is up for them already 14:11:48 and we have quite a way to go on the quantitative changes before we really get into the qualitative changes, besides just specs 14:11:55 on that code: I could do with some advice on how best to stack that set so that ready stuff can merge: 14:12:09 starts here: https://review.openstack.org/#/c/282442/ 14:12:36 the allocation stuff is on a different spec from generic-resource-pools 14:12:43 but is used in the tests for resource pools 14:14:14 seems fine to stack them like that, it's going to get messy b/c obviously there are more changes needed in there for the allocations bp 14:14:45 roger that 14:14:58 sort of related, but we closed the resource-providers-inventories spec but never implemented the online data migration for pci and numa 14:15:06 so i'm wondering where/when that happens? 14:15:33 sorry, compute-node-inventory-newton bp 14:15:57 mriedem: qual and quan sides can definitely be worked in parallel, but you are right that the specs and roadmap/direction need to be focused on the qual side. 14:16:35 mriedem: for the PCI and NUMA stuff, I'm working on a proposal to align those. Will have it done by end of week. 14:16:45 ok 14:17:10 mriedem: work can continue as-is in the meantime. have lots to do before that will become a blocker and me and Moshe are working on cleaning up the PCI tracking in prep for such work. 14:17:11 jaypipes: was that getting held up on the pci refactor changes in the resource tracker? 14:17:17 zactly 14:17:21 yay, ok 14:17:26 at least we're on the same page :) 14:17:29 yup 14:18:12 OK, anything else? 14:18:27 jaypipes: Is pci and numa qual or quan? 14:18:34 Yingxin: both :) 14:18:42 Yingxin: but mostly quan 14:18:48 and therein lies the problem 14:18:57 ok 14:19:10 Yingxin: some folks wish to add "NIC and PCI device capabilities" which is qual, but we have to get the quan side lined up first :) 14:19:51 so rp makes pci and numa quan. 14:20:26 Yingxin: resource providers is focused on the quan representation of those things, yes. 14:20:45 jaypipes: get it, thanks 14:21:52 Anything else? 14:22:08 the notes from the etherpad 14:22:10 mriedem, edleafe: what is the next step for https://review.openstack.org/#/c/263898/? 14:22:25 suggested that the "where does the data go" question opened up again? 14:22:29 are we in a position to approve it now? 14:23:05 mlavalle: we have to approve the generic-resource-pools spec first IMO 14:23:10 since it's a dependency 14:23:17 cdent: yes, but we never had the discussion on Friday, as some people had left 14:23:52 mriedem: ok, will keep an eye on that one. thanks 14:25:03 cdent: so for now it looks like we're still going to segment the data 14:25:31 * cdent stays up on the balls of his feet 14:26:35 mriedem: what is next step for https://blueprints.launchpad.net/nova/+spec/sriov-live-migration 14:27:23 svenkat: not really related to the scheduler meeting... 14:27:30 oh sorry. 14:28:09 anything else? 14:28:24 3... 14:28:28 2... 14:28:34 1... 14:28:37 #endmeeting