17:00:17 <alaski> #startmeeting nova_cells
17:00:18 <openstack> Meeting started Wed Feb 17 17:00:17 2016 UTC and is due to finish in 60 minutes.  The chair is alaski. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:20 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:00:22 <openstack> The meeting name has been set to 'nova_cells'
17:00:26 <mriedem> o/
17:00:27 <alaski> o/
17:00:56 <belmoreira> o/
17:01:17 <alaski> this could be a short meeting
17:01:21 <alaski> but let's get going
17:01:28 <bauzas> \o
17:01:29 <alaski> #topic Cells testing/bugs
17:01:31 <mriedem> i assume doffm is here
17:02:02 <alaski> hmm, topic didn't work
17:02:08 <bauzas> openstack you there ?
17:02:21 <alaski> I have not heard of any issues regarding testing/bugs
17:02:39 <alaski> belmoreira: you had wanted to raise some high priority bugs
17:02:43 <alaski> do you have a list?
17:02:55 <alaski> or at least raise some bugs to get prioritized
17:03:12 <belmoreira> let me find my list
17:04:30 <alaski> okay, we'll swing back around in open discussion
17:04:38 <alaski> #topic open reviews
17:04:51 <bauzas> there are a lot :)
17:04:59 <alaski> https://etherpad.openstack.org/p/mitaka-nova-priorities-tracking is your one stop shop for all things review related
17:05:23 <alaski> bauzas: yep, it's awesome
17:05:34 <alaski> everyone should drop everything and go review some cells code
17:05:49 <bauzas> could we maybe try to reorganize like we did with the sched prio ?
17:06:00 <bauzas> it's a bit hairy and a bit hard to follow
17:06:06 <bauzas> like sections per topic
17:06:10 <alaski> sure
17:06:14 <mriedem> i should be +2 on the flavor tables today
17:06:19 <mriedem> just need to get back to it, doffm updated that one
17:06:28 <alaski> mriedem: excellent
17:06:38 <alaski> I'll take a look as well
17:07:06 <bauzas> mriedem: yeah was my next review
17:07:07 <belmoreira> mriedem great
17:07:23 <bauzas> that one was stale recently AFAIR
17:07:42 <bauzas> oh yeah
17:07:53 <alaski> it was updated yesterday
17:08:00 <alaski> #topic Open Discussion
17:08:02 <bauzas> yeah just saw
17:08:11 <belmoreira> just saw it
17:08:44 <alaski> belmoreira: are you ready with your list?
17:09:12 <belmoreira> bug #1532562 was critical for was some weeks ago
17:09:12 <openstack> bug 1532562 in OpenStack Compute (nova) "Cell capacities updates include available resources of compute nodes "down"" [Undecided,In progress] https://launchpad.net/bugs/1532562 - Assigned to Belmiro Moreira (moreira-belmiro-email-lists)
17:09:53 <belmoreira> I did a patch that I deployed in our infrastructure and it would be great to have some feedback
17:10:09 <alaski> okay, I'll take a look
17:11:03 <mriedem> seems pretty straight-forward
17:11:09 <alaski> anything else that is a high priority for you?
17:11:18 <bauzas> but it calls the SG API
17:11:34 <bauzas> which is a bit concerning me, but let's discuss that offline
17:11:40 <belmoreira> we started using NUMA topology and hit #1517006
17:12:11 <alaski> bauzas: as long as it's calling it within a cell I think it's fine from a cells pov, but if you have other concerns we should certainly discuss
17:12:24 <bauzas> alaski: that's the concern I wanted to be super-sure :)
17:13:16 <mriedem> isn't there an attr on the service object that can give us the same info w/o the sg API?
17:13:23 <mriedem> like last_seen_up?
17:13:31 <bauzas> mriedem: well, it's a long convo
17:13:38 <bauzas> mriedem: and we have some BPs for that
17:13:50 <belmoreira> bauzas: that was my simple way to try to fix it. happy to change it
17:13:51 <bauzas> but AFAIK, the Service object is not yet having that kind of filed
17:13:54 <bauzas> field
17:14:03 <mriedem> bauzas: it's in the object class
17:14:11 <bauzas> belmoreira: nah, just need to think about the implications of your patch
17:14:12 <mriedem> you mean it's not getting set?
17:14:13 <bauzas> mriedem: I know
17:14:30 <bauzas> mriedem: well, remember, the SG API is a pure fun
17:14:39 <bauzas> mriedem: and last_seen_up is a driver thing
17:14:56 <belmoreira> i just raised that bug because we were suffering from it for a while without noticing... I imagine other are having the same problem
17:15:03 <bauzas> that's why we have 2 specs from someone trying to fix that
17:15:03 <alaski> belmoreira: that one could be tricky (#1517006). if it's adding new functionality it may have to wait, but I wonder if you can work around it by manually ensuring your flavors in the cells match
17:16:11 <belmoreira> I have a work around... but very ugly to try to submit something
17:16:56 <alaski> right. honestly I'm not sure it's something we should spend time on, but I may not fully understand it
17:17:06 <belmoreira> fine
17:17:35 <alaski> any others?
17:17:40 <ccarmack> I have a question about cells and quotas …. if the quota usages get out of sync vs. the actual usage, there are 2 config options (until_refresh and max age) to refresh the usages  …. would those work for cells?
17:17:42 <belmoreira> my point last meeting was that most bugs related with cells were not even triaged. I see that changed
17:17:53 <belmoreira> thanks for having the time
17:18:05 <alaski> belmoreira: yep. thanks for reminding me :)
17:18:16 <alaski> ccarmack: yes
17:18:31 <ccarmack> excellent .. thanks
17:18:47 <alaski> those are used way down in the db api so they're not affected by cells
17:18:53 <ccarmack> yup
17:19:35 <alaski> anything else for today?
17:20:13 <alaski> I'm still waiting on an update on the resource pools/classes/providers work from jaypipes to get a clear picture on how that will work with cells
17:20:31 <mriedem> how much of that is actually expected to happen in mitaka?
17:20:35 <alaski> and doffm has been involved in those discussions as well
17:20:36 <mriedem> i haven't been following along
17:20:41 <alaski> mriedem: they're planning to get it all in
17:20:45 <mriedem> jeez
17:20:52 <alaski> yeah, we'll see
17:21:02 <bauzas> the resource-pools, yes
17:21:13 <alaski> it is moving
17:21:15 <bauzas> really not sure for resource-providers
17:21:17 <jaypipes> not the allocations or scheduelr bits.
17:21:37 <bauzas> we're 2 weeks from FF :)
17:21:40 <jaypipes> only the resource-providers schema, the generic-resource-pools, and a bit of the compute node inventory bp
17:21:59 <alaski> okay, that's not a ton
17:22:10 <alaski> but generic-resource-pools is still an open question
17:22:38 <jaypipes> I will have the g-r-p bp revised and sent up by end of today
17:23:01 <alaski> cool, I'm ready to go through it again
17:23:35 <alaski> other than that I think most cells related work has patches up
17:24:17 <alaski> I would still like to try to move instance creation to the conductor, but if something big comes up in that work I may not be able to work around it in time
17:24:26 <alaski> anything else?
17:25:21 <alaski> cool. everyone has an extra 30 mins to go review code :)
17:25:25 <alaski> thanks all!
17:25:31 <alaski> #endmeeting