21:00:21 <mriedem> #startmeeting nova
21:00:22 <openstack> Meeting started Thu Jan  5 21:00:21 2017 UTC and is due to finish in 60 minutes.  The chair is mriedem. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:23 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:25 <openstack> The meeting name has been set to 'nova'
21:00:33 * Vek waves
21:00:35 <takashin> o/
21:00:36 <macsz> \o
21:00:36 <dansmith> moo/
21:00:42 <tovin07> o/
21:00:43 <dansmith> um, I did not mean to moo
21:00:44 <edleafe> \o
21:00:45 <macsz> cowsay? ;)
21:00:50 <thorst_> o/
21:00:57 <dtp> o/
21:00:57 <diana_clarke> o/
21:00:58 <bauzas> \o
21:01:01 <cdent> o/
21:01:04 <Vek> haha :)
21:01:55 <mriedem> since i just broke the nova gate let's get this party started
21:02:01 <mriedem> #link meeting agenda https://wiki.openstack.org/wiki/Meetings/Nova#Agenda_for_next_meeting
21:02:10 <Vek> heh
21:02:10 <mriedem> #topic release news
21:02:12 <efried> o/
21:02:17 <mriedem> #link Ocata release schedule: https://wiki.openstack.org/wiki/Nova/Ocata_Release_Schedule
21:02:23 <mriedem> #info Jan 19: final non-client library release (oslo)
21:02:31 <mriedem> ^ is the next upcoming major deadline
21:02:53 <mriedem> if you have changes that depend on a library patch to get released, that library release has to happen by the 19th
21:03:03 <mriedem> i know we at least have one for os-vif
21:03:15 <mriedem> #info Jan 26: o-3 milestone, final python-novaclient release, FeatureFreeze, Soft StringFreeze
21:03:29 <mriedem> so one week after lib freeze is feature freeze and the o-3 milestone
21:03:40 <mriedem> on a side note i'm on vacation that week so bauzas is going to handle the o-3 release tag
21:03:50 <mriedem> and jaypipes is also on vacation that week
21:03:56 * melwitt o/
21:04:07 <mriedem> questions about dates?
21:04:19 <mriedem> we've got 3 weeks until feature freeze and a lot to do yet
21:04:21 <mriedem> #link Ocata blueprints https://blueprints.launchpad.net/nova/ocata
21:04:35 <mriedem> #info 70 total blueprints, 16 implemented, 5 not started, 38 need code review
21:04:41 <jaypipes> hey, sorry, was out last few hours. back now.
21:04:48 <mriedem> so we've got about 23% of our blueprints completed
21:05:04 <mriedem> there are a few out there which need a final +2
21:05:23 <mriedem> #link Details on remaining blueprints: https://etherpad.openstack.org/p/nova-ocata-feature-freeze
21:05:39 <bauzas> I guess we could have a review day?
21:05:41 <mriedem> if you didn't see the ML thread, ^ is a thing i did last week
21:05:46 <mriedem> bauzas: getting there :)
21:06:03 <bauzas> oh yeah, just saw the agenda
21:06:13 <mriedem> i categorized the remaining blueprints and added notes on where i thought each was progress-wise in that etherpad
21:06:27 <mriedem> i plan on scrubbing through it again tomorrow
21:06:32 <mriedem> more on that later
21:06:35 <mriedem> any release questions?
21:06:54 <mriedem> #topic bugs
21:07:09 <mriedem> well
21:07:10 <mriedem> https://bugs.launchpad.net/nova/+bug/1654393
21:07:10 <openstack> Launchpad bug 1654393 in OpenStack Compute (nova) ""keystoneauth1.exceptions.catalog.EmptyCatalog: The service catalog is empty." in test_cinderclient_unsupported_v1" [High,Confirmed]
21:07:18 <mriedem> ^ is new as of today, bouncing changes
21:07:20 <mriedem> and is my fault
21:07:29 <mriedem> i suspect a racy global mock in the test added today
21:07:51 <mriedem> if no one else figures it out anytime soon i'll be working on that tonight
21:08:10 <mriedem> otherwise we're pretty good on gate status
21:08:27 <mriedem> anything else on bugs?
21:08:39 <mriedem> #topic reminders
21:08:43 <mriedem> the usual
21:08:43 <mriedem> #link Ocata review priorities https://etherpad.openstack.org/p/ocata-nova-priorities-tracking
21:09:00 <mriedem> #topic Stable branch status: https://etherpad.openstack.org/p/stable-tracker
21:09:17 <mriedem> there are some things getting into stable/newton that i'm looking to release soon, maybe next week
21:09:24 <mriedem> one is for fixing an upgrade issue from mitaka
21:09:28 <mriedem> so would like to get that released soon
21:09:55 <mriedem> #topic subteam highlights
21:10:00 <mriedem> dansmith: cells v2?
21:10:07 <dansmith> we had a meeting,
21:10:15 <dansmith> several things up for review, all on the priorities tracker page
21:10:32 <dansmith> quota stuff, multicell stuff, and some other minor things we need to change
21:10:41 <dansmith> that's pretty much it
21:11:02 <mriedem> main thing looking at right now is cells v1 job with cells v2
21:11:08 <mriedem> at least for dansmith
21:11:18 <dansmith> yeah, getting there.. have a debug patch up and slogging through
21:11:28 <mriedem> ok thanks
21:11:32 <mriedem> edleafe: scheduler?
21:11:39 <edleafe> I was out Monday, so jaypipes or cdent can fill you in
21:11:52 <cdent> jaypipes: ran the meeting and went through all the relevant reviews
21:11:59 <cdent> and that's about it
21:12:18 <bauzas> was on PTO
21:12:31 <cdent> We introduced some new contributors to the lay of the land
21:12:37 <cdent> nice way to start year
21:12:44 <jaypipes> mriedem: briefly talked about the issue with NUMA objects being recreated every time update_available_resource() is called and the trouble that causes trying to add a uuid field to the NUMA cell.
21:12:46 <mriedem> ok, so for placement, the main focus right now is going to be on getting https://review.openstack.org/#/c/392569/ in which is needed for the scheduler to use the placement API
21:12:59 <jaypipes> mriedem: yep ++
21:13:02 <mriedem> and then bauzas is going to have a change up for the scheduler
21:13:07 <mriedem> any minute now... :)
21:13:27 <mriedem> and that's the main goal for placement in ocata
21:13:36 <mriedem> there are of course lots of other patches up
21:13:41 <bauzas> yeah, the scheduler change should come up soon
21:14:09 <mriedem> ok moving on
21:14:17 <mriedem> is tdurakov or pkoniszewski around?
21:14:19 <mriedem> for live migration
21:14:40 <mriedem> guess not
21:14:46 <mriedem> ok, api
21:14:53 <mriedem> sdague: were you at the api meeting this week?
21:15:14 <mriedem> well,
21:15:17 <mriedem> two things,
21:15:19 <cdent> lots of discussion about capabiities spec under review in api-wg
21:15:31 <mriedem> the query parameters validation changes are in and merged
21:15:52 <cdent> some discussion on how to manage the fixes to filtering (avoiding joins and whatnot)
21:15:54 <mriedem> cdent: yeah i expect that'll be a cross-project discussion at the PTG
21:16:16 <mriedem> i'm not sure if i can call it cross project or not, the PTG rules are new and scary
21:16:41 <cdent> quite
21:16:45 <bauzas> there is now a wiki AFAIK
21:16:48 <mriedem> the other thing for ocata api stuff is the server sort/filter patches that Kevin Zheng is working on
21:17:12 <bauzas> https://wiki.openstack.org/wiki/Forum
21:17:17 <mriedem> those changes are here https://review.openstack.org/#/q/topic:bp/add-whitelist-for-server-list-filter-sort-parameters+status:open
21:17:42 <mriedem> looks like i need to talk to alex_xu and Kevin about what's going on there
21:17:56 <mriedem> moving on
21:18:15 <mriedem> i don't see anyone around for the sriov/pci meeting, if there was one this week
21:18:21 <mriedem> gibi: notifications updates?
21:18:30 <gibi> short one
21:18:35 <gibi> nothing has been finished but everything is moving
21:18:40 <gibi> the first patch of the instance action notification extensions for searchlight has been merged
21:18:45 <gibi> not the whole bp scope will be finished for Ocata but we still have some patch that has chance
21:18:59 <gibi> json-schema-for-versioned-notifications still has some dependency on ovo lib due to https://review.openstack.org/#/c/411822/
21:19:04 <gibi> so this has less and less chance to be finished in Ocata
21:19:29 <gibi> the flavor notification bp has code up and it is getting in a good shape
21:19:36 <mriedem> yeah i need to get back on https://review.openstack.org/#/c/398171/
21:19:38 <mriedem> but it's close
21:19:46 <gibi> the priority etherpad is up to date
21:19:52 <gibi> that is all
21:19:57 <mriedem> ok thanks gibi
21:20:10 <mriedem> #topics PTG planning
21:20:19 <mriedem> #topic PTG planning
21:20:25 <mriedem> #link Pike PTG planning: http://lists.openstack.org/pipermail/openstack-dev/2017-January/109594.html
21:20:48 <mriedem> i've started an etherpad to jot down topics
21:21:01 <mriedem> i'm still just trying to figure out myself how this thing is going to work
21:21:30 <mriedem> but with wed-friday being like a normal midcycle i guess i'm going to start out by treating it that way, mostly open-ended
21:21:44 <mriedem> but we'll need to be discussing things for Pike deliverables for sure since it's at the start of Pike
21:22:03 <mriedem> any questions on the PTG?
21:22:15 <mriedem> #topic stuck reviews
21:22:23 <mriedem> there was nothing in the agenda - anyone have something for this?
21:22:39 <mriedem> #topic open discussion
21:22:42 <mriedem> a few items
21:22:54 <mriedem> I plan on scrubbing https://etherpad.openstack.org/p/nova-ocata-feature-freeze  before next week to maybe drop things that haven't started but to also  highlight things that need review and are close if given a good push.
21:23:04 <mriedem> Given that, I'm proposing a review sprint for next week on Wednesday  1/11 which leaves about a week before library release freeze and two  weeks before feature freeze.
21:23:30 <mriedem> are people ok with that date for a blueprint code review sprint?
21:23:35 <mriedem> i can't do monday or tuesday
21:23:52 <cdent> wednesday++
21:24:06 <bauzas> cool by me
21:24:12 <mriedem> ok - it's going to be mostly harassment in the nova channel
21:24:42 <mriedem> #info if you've got a blueprint up for review, be in the channel on wednesday 1/11 in case there are questions
21:24:49 <mriedem> #undo
21:24:51 <openstack> Removing item from minutes: #info if you've got a blueprint up for review, be in the channel on wednesday 1/11 in case there are questions
21:25:02 <mriedem> #info if you've got a blueprint up for review, be in the openstack-nova channel on wednesday 1/11 in case there are questions
21:25:13 <mriedem> (efried / thorst) PowerVM Update
21:25:19 <thorst_> I’ll give a powervm CI update.  efried will do a powervm driver review update.
21:25:34 <thorst_> We hit an issue where we need to change the name space of our out of tree driver so that in tree can just be ‘powervm’.  We’re working through that joy...  Affected our CI.
21:25:45 <thorst_> We’re also working through a keystone issue now, so while we’re running nova jobs, we shut off publishing so we don’t flood red to each patch (even if non-voting).  Plan to turn back on soon.
21:26:29 <thorst_> that's about it for where we are w/ CI.
21:26:31 <mriedem> ok
21:26:48 <efried> Four outstanding change sets for the PowerVM driver:
21:26:49 <efried> #link https://review.openstack.org/#/q/powervm-nova-compute-driver+status:open
21:27:02 <efried> First one is a bit large (+1732 at current patch set).
21:27:02 <efried> #link https://review.openstack.org/#/c/391288/
21:27:14 <efried> Would like to get some feedback from those present as to whether we should attempt to split this up further.  This won't be trivial, or we would have done it already, but I'm sure we can figure it out if it's going to make the difference getting core reviews.
21:27:42 <mriedem> so,
21:27:52 <mriedem> this is actually paired down from when i first looked at it,
21:27:55 <mriedem> before it had tests :)
21:28:11 <mriedem> so now the first one is just spawn/destroy + infra
21:28:30 <mriedem> when i looked last week, the powervm ci was running wayyyy more tests than i thought it would for just spawn/destroy
21:28:35 <mriedem> which was probably why a ton of it was failing
21:28:40 <efried> Correct.  If we were gonna split it further, it would be infra with no functionality, then spawn/destroy in a subsequent.
21:28:49 <thorst_> mriedem: so that's the issue we're working on.  The OOT was what ran on that
21:28:50 <efried> Well, we have a question about the CI in a sec.
21:28:54 <mriedem> i would prefer to have spawn/destroy minimal in the first change
21:29:09 <thorst_> we will have a second CI for the in tree...but its a bit of a chicken / egg there on the CI and driver there.
21:29:16 <mriedem> if there is something in there that looks like it could be split out, people can comment as such
21:29:41 <efried> Sounds good to me.
21:29:46 <mriedem> thorst_: i'd expect that ci to just run on these patches
21:29:51 <mriedem> not other nova patches
21:30:08 <thorst_> the in tree or out of tree?
21:30:16 <mriedem> and not be doing anything with volumes, or attaching interfaces, pause/suspend, any of that stuff
21:30:21 <mriedem> the in-tree
21:30:31 <mriedem> i expect the in-tree ci to only be running on the patches for the in-tree driver right now
21:30:36 <mriedem> it doesn't really make sense otherwise
21:30:42 <thorst_> OK - yeah, we are setting up the skip lists and config for that one in parallel
21:30:47 <mriedem> ok
21:30:55 <thorst_> its...fun...
21:30:59 <mriedem> i had figured the tests being run were for the out of tree driver, which didn't make sense for the in-tree patches
21:31:11 <mriedem> thorst_: honestly i'd probably start with a whitelist
21:31:13 <mriedem> and build that up
21:31:16 <efried> Right, so the question is whether we need to have (a drastically pared-down version of) our CI actually passing before we can expect to merge each change?
21:31:36 <mriedem> efried: well i expect to see the powervm ci reporting on those changes to show green
21:31:54 <mriedem> if that means you're only running 20 tempest tests for spawn/destroy, so be it
21:31:54 <efried> Okay, that's what I figured, and that's totally reasonable.
21:32:08 <thorst_> sounds good
21:32:10 <mriedem> start small and build out the test set
21:32:14 <mriedem> otherwise you're going to die
21:32:22 <efried> For the other change sets, we expect to be able to wipe the WIP off in the next week or two.
21:32:42 <efried> Realizing this means some of this function will probably slip to pike.
21:32:49 <mriedem> just to level set expectations, i think getting the first patch for spawn/destroy in this release is going to be an achievement
21:33:02 <thorst_> understood.
21:33:14 <mriedem> ok let's move on
21:33:17 <mriedem> thanks for the updates guys
21:33:24 <mriedem> (tovin07): OSprofiler support in Nova is almost done and marked as good progress.
21:33:26 <thorst_> thx
21:33:32 <tovin07> yes
21:33:48 <tovin07> i just want to get some attention for this patch
21:33:54 <tovin07> #link osprifiler in nova review: https://review.openstack.org/#/c/254703/
21:34:02 <tovin07> patch in novaclient is OK now, waiting for patch in nova...
21:34:19 <mriedem> yeah it'd be nice to get that in finally, but i haven't gone through it since newton i don't think
21:34:40 <mriedem> but i know we relied on it heavily for our cloud performance testing
21:34:44 <tovin07> yup, i see it too
21:34:58 <mriedem> so, be around in the nova channel on wed 1/11 in case questions come up
21:34:59 <mriedem> or issues
21:35:15 <mriedem> ok, anyone else have anything to discuss?
21:35:17 <tovin07> okay, sound good :D
21:35:28 <tovin07> thanks
21:35:50 <mriedem> i'm not hearing anything, and we're already about 20 minutes over where dansmith wanted this to be, so thanks everyone
21:35:55 <dansmith> yup.
21:35:56 <mriedem> #endmeeting