14:00:11 <edleafe> #startmeeting nova_scheduler
14:00:12 <openstack> Meeting started Mon Jun 12 14:00:11 2017 UTC and is due to finish in 60 minutes.  The chair is edleafe. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:13 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:15 <openstack> The meeting name has been set to 'nova_scheduler'
14:00:20 <edleafe> WHo's here today?
14:00:20 <jaypipes> o/
14:00:34 <cdent> o/
14:00:35 <dtantsur> o/
14:00:47 <edleafe> bauzas mentioned that he's doing daddy duty and probably won't be here
14:01:34 <mriedem> GET /edleafe
14:02:07 <jaypipes> PUT /mriedem
14:02:10 <edleafe> 409
14:02:31 <edleafe> #topic Specs & Reviews
14:02:31 <cdent> if we get an earlier microversion of edleafe, will that result in a smooth shit-talking new yorker and not some dude from texas?
14:02:37 <edleafe> #link Nested Resources: series starting with https://review.openstack.org/#/c/415920/
14:02:56 <edleafe> jaypipes: anything of note there?
14:02:58 * dtantsur watches how every meeting with edleafe and cdent turns into an API WG meeting
14:03:27 <jaypipes> edleafe: yeah, that series was just rebased to keep up with master. it's got a few issues cdent identified, and I'll try to address those tomorrow (today not good)
14:03:47 <edleafe> cool
14:03:55 <jaypipes> edleafe: I believe traits integration with scheduler and the claims stuff is prioritized before nested
14:04:03 <edleafe> dtantsur: did you come for scheduler issues, or just to be amused?
14:04:18 <dtantsur> edleafe: the latter of course!</kidding>
14:04:27 <edleafe> :)
14:04:33 <dtantsur> edleafe: I'm watching for things related to changes in scheduling bare metal stuff
14:04:42 <edleafe> ok, next up:
14:04:43 <edleafe> #link Alternate Allocations https://review.openstack.org/#/c/473377/
14:04:44 <dtantsur> since Jim is no longer our driving force in this field...
14:04:47 <edleafe> #link Spec: https://review.openstack.org/#/c/471927/
14:05:05 <jaypipes> so, a question for you all on the above...
14:05:15 <edleafe> New revision of the spec, and new code
14:05:26 * alex_xu waves late
14:05:34 * edleafe waves back to alex_xu
14:06:05 <jaypipes> would it be more useful for me to push the next few patches that add a REST API response for the PlacementAlternatives thing and integration into the scheduler? or should I wait/block on the first patch that I've pushed in order to get folks' ideas.
14:06:28 <mriedem> what is https://blueprints.launchpad.net/openstack/?searchtext=placement-allocation-requests ?
14:06:31 <mriedem> oh nvm
14:06:33 <mriedem> the new spec
14:06:58 <jaypipes> mriedem: right, and https://review.openstack.org/#/c/473377/ is the implementation start.
14:07:20 <jaypipes> mriedem: still rough, of course. want people's thoughts on it.
14:07:34 <edleafe> jaypipes: haven't had a chance to dig into it yet
14:07:49 <jaypipes> edleafe: no worries, man. just pushed it this morning. :)
14:08:06 <edleafe> but in general, I'd prefer to see the series in order to get a better idea where things are heading
14:08:18 <jaypipes> edleafe: k, I can do that :)
14:09:14 <edleafe> Anything else to add to this?
14:09:54 <edleafe> Next on the agenda:
14:09:56 <edleafe> #link project_id and user_id in PUT /allocations: https://review.openstack.org/#/c/469634/
14:10:28 <edleafe> Any issues to discuss with that?
14:10:53 <mriedem> so,
14:10:58 <mriedem> melwitt started it,
14:11:01 <mriedem> jaypipes updated it,
14:11:05 <mriedem> dansmith +2ed it
14:11:14 <mriedem> are there issues with it beyond the rebase that's needed?
14:11:52 <edleafe> dunno - it's monday morning and I still need to reload my brain with where things are
14:12:35 <cdent> I think it's okay, given that we've decided to worry about consumer_uuid "safety" later
14:12:47 <mriedem> can we start -1ing if there is no api reference?
14:13:16 <cdent> mriedem: that's a good question
14:13:30 <cdent> ....probably, or at least soon
14:14:05 <mriedem> ok, on the compute side i'd -1 if there weren't an api reference update for a microversion change
14:14:31 <edleafe> It's probably a good thing to start doing for placement
14:14:35 <cdent> I reckon you get to set the tone mriedem as part of your onerous leadership responsibilities
14:14:57 <mriedem> i don't want it to be like that,
14:15:03 <mriedem> i just think it's good to have the docs in the same change for context
14:15:10 <mriedem> and so we don't miss them
14:15:19 <jaypipes> mriedem: we haven't merged the allocations api-ref patch yet. https://review.openstack.org/#/c/470933/
14:15:46 <jaypipes> mriedem: so it would be rough to -1 something for an api-ref section that doesn't yet exist.
14:16:02 <mriedem> ok i'll remove the -1 and just hound
14:16:06 <jaypipes> thx
14:16:57 <edleafe> Moving on...
14:17:00 <edleafe> #link Delete all inventory: https://review.openstack.org/#/c/460147/
14:17:13 <edleafe> I see that cdent has been helping rfolco with that
14:18:08 <rfolco> edleafe, in good shape I guess, waiting for CI and reviews, close to merge
14:18:32 <edleafe> rfolco: cool; it's in my review queue
14:18:51 <jaypipes> mine too
14:19:16 <edleafe> Finally, and would have made a better segueway one spot earlier:
14:19:18 <edleafe> #link Placement API ref docs: https://review.openstack.org/#/q/topic:cd/placement-api-ref+status:open
14:19:56 <mriedem> so there are duplicates
14:21:15 <mriedem> would be nice if avolkov can be co-author on the existing allocations api-ref
14:21:19 <mriedem> rather than creating duplicates
14:21:32 <jaypipes> hmm?
14:21:40 <cdent> i think that's unintentional
14:21:54 <mriedem> https://review.openstack.org/#/c/470933/ is a duplicate of https://review.openstack.org/#/c/450550/
14:22:02 <mriedem> looks like he made several comments and then pushed his own
14:22:24 <jaypipes> mriedem: no.
14:22:43 <jaypipes> mriedem: Kevin's is GET, andrey's is everything except GET.
14:23:07 <jaypipes> mriedem: and andrey reviewed Kevin's and specifcally de-duped his pathc.
14:23:09 <jaypipes> patch
14:23:36 <mriedem> ok, that's not obvious in commit message or how they are aligned
14:23:38 <mriedem> as in, they aren't aligned
14:23:44 <mriedem> but i didn't -1 so whatever
14:23:56 <mriedem> so i just won't review these
14:24:07 <jaypipes> mriedem: agreed
14:25:06 <jaypipes> mriedem: I will review them and ask andrey/kevin to work closer together.
14:25:23 <mriedem> i posted some comments in andrey's change
14:26:06 <mriedem> moving on?
14:26:35 <mriedem> edleafe: ?
14:26:50 <edleafe> waiting for you guys to finish
14:27:00 <edleafe> #topic Bugs
14:27:08 <edleafe> #link Placement bugs https://bugs.launchpad.net/nova/+bugs?field.tag=placement
14:27:16 <edleafe> One new one this week
14:28:13 <cdent> that bug has interesting implications
14:28:23 <edleafe> how so?
14:28:25 <cdent> the deeper we get into the comments the more strange it becomes
14:28:35 <cdent> #link https://bugs.launchpad.net/nova/+bug/1696830
14:28:36 <openstack> Launchpad bug 1696830 in OpenStack Compute (nova) "nova-placement-api default config files is too strict" [Low,Confirmed] - Assigned to Corey Bryant (corey.bryant)
14:28:50 <cdent> the last comment is about how conf is being loaded for the auth middleware
14:28:58 <cdent> (in deploy.py)
14:29:44 <cdent> the initial fix doesn't fix the whole problem and the initial code didn't really imagine the use case presented by the bug
14:31:08 <edleafe> ok, well, be sure to ping on -nova if it gets stranger
14:31:18 <edleafe> #topic Open discussion
14:31:31 <edleafe> What's on everyone's mind?
14:31:38 <mriedem> i have something
14:31:54 <mriedem> not that at least 4 major bp efforts aren't enough for placement,
14:31:58 <mriedem> but what do we want to do about https://blueprints.launchpad.net/nova/+spec/custom-resource-classes-in-flavors ?
14:32:00 <dtantsur> I wanted to ask (probably jaypipes) about the status of https://blueprints.launchpad.net/nova/+spec/custom-resource-classes-in-flavors, but this can wait
14:32:06 <dtantsur> LOL
14:32:11 <cdent> yay, jinx!
14:32:44 <mriedem> i don't expect jaypipes to own nested rps, shared resources, alternative allocations, and this at the same itme
14:32:46 <mriedem> *time
14:33:08 <mriedem> so, do we find another owner or defer?
14:33:17 <dtantsur> how complex is this code-wise?
14:33:27 <mriedem> dtantsur: have you read the spec?
14:33:39 <dtantsur> mriedem: yep, but it's hard to estimate the work
14:33:45 <dtantsur> I've never contributed anything to nova outside of nova/virt/ironic
14:34:03 <jaypipes> shouldn't be too difficult to do, but as mriedem mentioned, I'm starting to get stretched.
14:34:15 <edleafe> I might be able to pitch in
14:35:19 <cdent> I might as well, but I've got a two week gap comin up
14:35:37 <cdent> so I don't want to make promises
14:35:45 <dtantsur> I can help, but I'll probably require too much hand-holding
14:36:15 <mriedem> helping on the test side would be useful, as in actually verifying the code proposed works with an ironic environment
14:36:31 <mriedem> so black box testing
14:36:40 <edleafe> Yeah, that's hard for me to test :)
14:37:06 <dtantsur> oh, this one I can do!
14:37:33 <edleafe> OK, anything else?
14:37:45 <mriedem> so to summarize this one,
14:38:04 <mriedem> it actually sounds like 3 changes in nova: 1 in the placement api for a new filter parameter to getting RPs,
14:38:15 <mriedem> then a change in the filter scheduler to request based on these extra specs,
14:38:20 <mriedem> and a data migration
14:38:35 <cdent> 1st step is pretty much done
14:38:40 <mriedem> maybe split up the work so cdent does the api, edleafe does the scheduler, etc
14:38:58 <alex_xu> i can help also
14:39:01 <cdent> with the resources filter, and if jay's doing resource filter on the /allocations_request
14:39:44 <jaypipes> we need to get traits into the pipeline too, ASAP.
14:39:55 * cdent nods
14:40:00 <mriedem> alright well i'm not as involved so i guess you guys all sort out who wants what piece of the pie
14:40:07 <edleafe> wil do
14:40:16 <dtantsur> ping me for any testing with ironic
14:40:19 <cdent> so many pies
14:40:26 <mriedem> cdent being an englishman wants the kidney part
14:40:32 * cdent voms
14:40:44 <edleafe> I have to run, so if there's anything else, continue on in -nova
14:40:48 <edleafe> Thanks everyone!
14:40:52 <edleafe> #endmeeting