14:01:09 <n0ano> #startmeeting nova-scheduler
14:01:10 <openstack> Meeting started Mon Aug  3 14:01:09 2015 UTC and is due to finish in 60 minutes.  The chair is n0ano. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:11 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:01:13 <openstack> The meeting name has been set to 'nova_scheduler'
14:01:27 <n0ano> anyone here to talk about the scheduler (other than early bird lxsli :-)
14:01:36 <edleafe> o/
14:02:16 <jaypipes> o/
14:02:52 <n0ano> #topic Liberty specs
14:03:30 <n0ano> So, I've noticed that bauzas_off patch https://review.openstack.org/199205 is failing for no good reason and I've hit it with a recheck a couple of times and it's still failing
14:03:50 <n0ano> do I just keep doing rechecks until the gate decides to work or is there something else going on?
14:04:26 <jaypipes> n0ano: no.
14:04:48 <jaypipes> n0ano: the failures are in https://review.openstack.org/#/c/188487/
14:04:59 <jaypipes> n0ano: which is WAY up in the dependent patch chain.
14:05:41 <jaypipes> n0ano: basically, we all need to do reviews of the whole patch series, starting from the beginning one: https://review.openstack.org/#/c/145528/38
14:06:35 <n0ano> jaypipes, ah, I didn't realize the whole chain was in flight
14:07:04 <n0ano> reviewing the whole chain is good but bauzas_off won't be back for a week which is unforutnate
14:07:55 <edleafe> n0ano: we can always push new patches to fix the test failures
14:07:57 <jaypipes> n0ano: sure, but we can provide feedback.
14:08:15 <edleafe> that too :)
14:08:53 <lxsli> So anyone want to discuss Liberty Specs?
14:09:23 <n0ano> lxsli, kind of what we were doing (these are the patches that implement the specs), what were you concerned about?
14:10:27 <johnthetubaguy> btw, Mitaka is open for specs, if that helps anyone
14:11:01 <n0ano> johnthetubaguy, tnx, good to know, we should be proposing a few
14:11:30 <johnthetubaguy> i need to send quite a few ML mails, thats on my TODO, once this freeze is under control
14:12:12 <n0ano> johnthetubaguy, with great power comes great responsibility :-)
14:12:15 <lxsli> n0ano: we just seemed to be off-topic and furthermore to have wound down
14:12:55 <lxsli> n0ano: I'm unsure check-migr-dest will make it given bauzas_off is off and I also depend on alaski's patches that he hasn't written yet
14:13:01 <n0ano> lxsli, the reality is I've been loose with terminology, the important specs for Liberty are approved so the real discussion is on the patches
14:13:56 <n0ano> lxsli, is that spec on the priority page?  That's what I've been using for tracking
14:15:38 <n0ano> lxsli, have you talked to alaski to see how close he is on his patches?
14:15:39 <lxsli> uh, the spec is merged to Liberty
14:15:56 <lxsli> not yet I can do that
14:16:12 <n0ano> lxsli, so, no problem on the spec, but we do need to track it's patches
14:17:26 <n0ano> lxsli, can you write preliminary code based upon what alaski should do and then update it when he actually has his implmentation?
14:17:42 <lxsli> n0ano: I'm having a go but necessarily that's vague
14:17:58 <lxsli> I guess I primarily need to sync with alaski, let's move on
14:18:19 <n0ano> lxsli, I feel your pain but that's the nature of dependencies
14:18:30 <n0ano> OK, anything else on the current specs/patches?
14:18:57 <lxsli> On resource-objects, I hope to sync with Jay today to resolve the "last" question
14:19:20 <n0ano> lxsli, jaypipes is right here, you want to sync right now?
14:19:51 <lxsli> during meeting time?
14:20:01 <n0ano> if it
14:20:13 <lxsli> I can't promise it'll be short :)
14:20:32 <n0ano> lxsli, let's do it at the end then, we should finish here soon.
14:20:41 <jaypipes> lxsli: last I checked, ndipanov was negative on the final patch in the series because he wanted to see it used in followup code.
14:21:47 <lxsli> jaypipes: I heard that but as a +0
14:22:13 <lxsli> jaypipes: my question is around whether #consume needs to call numa_fit_instance_to_host
14:22:15 <jaypipes> lxsli: I have a patch series I will be pushing shortly that refactors unit tests to use mock instead of mox (and not rely on DB in unit tests). I noticed that any time i change the resource tracker, a metric shit-ton of unit tests fail. So, I was hoping to refactor these tests to prevent a giant 5000LOC patch.
14:23:08 <jaypipes> lxsli: I thought we chatted about that last wednesday... yes, I think it should use numa_fit_instance_to_host. I need to update that patch.
14:23:42 <lxsli> jaypipes: OK, didn't catch your agreement. Thanks!
14:24:06 <n0ano> lxsli, so this was short after all?
14:24:11 <jaypipes> lxsli: no worries, man. I was in and out of eleventy billion meetings that day.
14:24:18 <lxsli> as it turns out I didn't have to bully Jay hardly at all!
14:24:50 <jaypipes> hehe
14:24:55 * n0ano bummer, watching bullying on someone else is fun
14:24:59 <n0ano> anywy
14:25:08 <n0ano> #expose host capabilities
14:25:33 <n0ano> so, eventually, I did send out an email late last night, I want to start a discussion...
14:25:51 <n0ano> my hope is to gather ideas on the problem space before we dive into a solution...
14:26:03 <n0ano> would love to hear people's thoughts on the ML
14:26:42 <jaypipes> n0ano: yes, I owe you a response on that.
14:27:09 <n0ano> jaypipes, slacker, you've literally had from midnight to 6AM to do that :-)
14:27:21 <jaypipes> :)
14:28:00 <ndipanov> jaypipes, no I wasn't negative
14:28:09 <ndipanov> I just said it would be nice to see it in action
14:28:17 <n0ano> hopefully we can talk about this subject in more detail at the next meeting when everyone has a chance to think about it
14:28:25 <n0ano> #topic opens
14:28:27 <ndipanov> and +1ed it iirc
14:28:49 <n0ano> anyone have anything new today?
14:29:43 <lxsli> *crickets*
14:29:51 <n0ano> I'm hearing the dreaded sound of crickets
14:29:55 <jaypipes> n0ano: http://www.memes.com/img/510571
14:30:23 <n0ano> run away, puppy pics
14:30:34 <lxsli> jaypipes: I felt bad about bugging you for a good half-minute after you posted that
14:30:47 <jaypipes> lxsli: :) I just love that pic.
14:31:02 <n0ano> tnx everyone, back to reviewing and we'll talk next week
14:31:17 <n0ano> #endmeeting