14:00:13 <mriedem> #startmeeting nova
14:00:16 <openstack> Meeting started Thu Jul 27 14:00:13 2017 UTC and is due to finish in 60 minutes.  The chair is mriedem. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:17 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:20 <openstack> The meeting name has been set to 'nova'
14:00:20 <dansmith> o/
14:00:34 <bauzas> \o
14:00:42 <edleafe> \o
14:00:44 <takashin> o/
14:00:48 <sdague> o/
14:01:27 <mriedem> ok let's get started
14:01:28 <efried> o/
14:01:29 <mriedem> #link agenda https://wiki.openstack.org/wiki/Meetings/Nova
14:01:36 <mriedem> #topic release news
14:01:41 <cdent> o-
14:01:54 <mriedem> #link Pike release schedule: https://wiki.openstack.org/wiki/Nova/Pike_Release_Schedule
14:02:03 <mriedem> #info July 27 (today!!!) is feature freeze, final python-novaclient release, pike-3 milestone
14:02:18 <mriedem> i'll be posting the final release request for python-novaclient after the meeting
14:02:49 <mriedem> the pike-3 tag is going to be held up on getting the claims in the scheduler patch merged
14:03:22 <mriedem> #info Blueprints: 65 targeted, 64 approved, 33 completed (+4 from last week)
14:03:28 <mriedem> oh that's actually old
14:03:40 <mriedem> #undo
14:03:41 <openstack> Removing item from minutes: #info Blueprints: 65 targeted, 64 approved, 33 completed (+4 from last week)
14:03:55 <jaypipes> o/
14:04:12 <mriedem> #info Blueprints: 62 targeted, 61 approved, 44 completed (+11 from last week)
14:04:27 <mriedem> so things are winding down
14:04:37 <mriedem> #link Pike feature freeze blueprint tracker etherpad: https://etherpad.openstack.org/p/nova-pike-feature-freeze-status
14:04:59 <mriedem> any questions about the release?
14:05:14 <mriedem> alright moving on
14:05:18 * gibi lurks
14:05:19 <mriedem> #topic bugs
14:05:30 <mriedem> no critical bugs
14:05:39 <mriedem> #info 59 new untriaged bugs (down 64 from last week - thanks sdague and others for helping with bug triage)
14:05:55 <mriedem> #info Starting to tag bugs for pike-rc-potential: https://bugs.launchpad.net/nova/+bugs?field.tag=pike-rc-potential
14:06:16 <mriedem> gate status is pretty ok for nova changes
14:06:27 <mriedem> the functional test craziness seems to have dropped off a bit
14:06:49 <mriedem> which is maybe due to the 401 auth middleware hack, and the wsgi intercept stuff
14:07:03 <mriedem> thanks cdent for helping with that
14:07:14 <mriedem> #link Trying to fix the Ironic CI jobs: https://review.openstack.org/#/c/487458/
14:07:15 <patchbot> patch 487458 - ironic - Don't use multicell setup for any ironic jobs
14:07:22 <mriedem> ^ looks like it's finally passing
14:07:23 <cdent> there’s another wsgi intercept change up, for osapi, but I assume we’ll want to wait on that one?
14:07:38 <mriedem> cdent: seems we can wait for now
14:07:51 <bauzas> we have the placementfixture since a while, so I'm a bit surprised it could be hitting races now
14:08:00 <bauzas> but anyway
14:08:13 <mriedem> bauzas: the placement fixture wasn't used in the functional tests globally until <2 weeks ago
14:08:23 <mriedem> when it was added to the IntegratedHelpers mixin
14:08:30 <mriedem> which is the base mixin for like 90% of functional tests
14:08:54 <mriedem> anywho
14:08:57 <bauzas> that's correct, only a few functional tests wereusing it
14:09:00 <mriedem> #link 3rd party CI status http://ci-watch.tintri.com/project?project=nova&time=7+days
14:09:04 <bauzas> anywho
14:09:04 <mriedem> i don't have any news on 3rd party ci
14:09:15 <mriedem> questions/comments about bugs?
14:09:48 <mriedem> #topic reminders
14:09:56 <mriedem> #link Consider signing up for the Queens PTG which is the week of September 11 in Denver, CO, USA: https://www.openstack.org/ptg
14:10:24 <mriedem> #link Queens PTG ideas etherpad https://etherpad.openstack.org/p/nova-ptg-queens
14:10:46 <mriedem> ^ is totally unordered
14:10:51 <mriedem> i just needed to get some ideas down somewhere
14:11:09 <mriedem> #info PTL nominations for Queens start next week (July 31 - Aug 9): https://governance.openstack.org/election/
14:11:23 <jaypipes> mriedem: you might ask cdent to fix that unordered thoughts problem.
14:11:38 <jaypipes> mriedem: you running?
14:11:47 <cdent> que?
14:11:55 <mriedem> jaypipes: don't put baby in the corner
14:12:07 <jaypipes> cdent: I was alluding to your unit test fixes on that move operation patch ;)
14:12:23 <jaypipes> mriedem: sorry honey.
14:12:29 * cdent doesn’t do allusions when coughing
14:12:30 <mriedem> i'll do PTL again for queens if no one else wants it
14:13:01 <mriedem> if anyone has questions about PTL stuff for queens and want to ask privately, feel free
14:13:15 <mriedem> moving on
14:13:19 <mriedem> #topic stable branch status
14:13:26 <mriedem> #link stable/ocata: https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:stable/ocata,n,z
14:13:31 <mriedem> #link stable/newton: https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:stable/newton,n,z
14:13:43 <mriedem> there is no real news here, i'll push for stable reviews prior to rc1
14:13:47 <mriedem> since we do a stable release around then too
14:14:02 <bauzas> mriedem: no news about a possible EOL for newton ?
14:14:06 <mriedem> no
14:14:10 <bauzas> cool then
14:14:16 <mriedem> i think that's a ways off
14:14:27 <mriedem> 2017-10-11
14:14:31 <mriedem> https://releases.openstack.org/
14:14:36 <mriedem> yeah we have a few months for newton
14:14:39 <bauzas> oh right, gtk
14:14:47 <bauzas> mriedem: thanks
14:14:52 <mriedem> #topic subteam highlights
14:15:04 <mriedem> dansmith: anything you'd like to highlight for cells v2 besides no meeting?
14:15:18 <dansmith> just that we merged the devstack patch,
14:15:32 <dansmith> and less drama than last time but ironic was still broken and they didn't realize,
14:15:40 <dansmith> but I think we've got them on a path to recovery at this point
14:15:51 <dansmith> which is pretty awesome
14:15:55 <dansmith> that's about it I think
14:16:03 <mriedem> a 12 step program to recovering from cells?
14:16:08 <dansmith> heh
14:16:10 <dansmith> from ironic
14:16:20 <mriedem> #link cells v2 todos https://etherpad.openstack.org/p/nova-pike-cells-v2-todos
14:16:26 <dansmith> oh,
14:16:34 <dansmith> I could highlight my doc patch for critical eyeballs
14:16:38 <mriedem> yup
14:16:42 <mriedem> was just going to do that
14:16:47 <dansmith> #link https://review.openstack.org/#/c/487183/
14:16:47 <patchbot> patch 487183 - nova - Add some more cellsv2 doc goodness
14:17:03 <mriedem> also, cfriesen has a fix up for unshelving with cellsv2 https://review.openstack.org/#/c/486208/
14:17:03 <patchbot> patch 486208 - nova - Ensure we unshelve in the cell the instance is mapped
14:17:04 <mriedem> i'm +2
14:17:26 * jaypipes looks
14:17:29 <mriedem> ok moving on
14:17:32 <mriedem> edleafe: scheduler?
14:17:36 <edleafe> Scheduler subteam meeting was just a recap of outstanding patches.
14:17:36 <edleafe> Reviews are focused on Jay's claims in the scheduler series, starting with https://review.openstack.org/#/c/487589/
14:17:37 <patchbot> patch 487589 - nova - placement: account for move operations in claim
14:17:39 <edleafe> Still working out the claim workflow for new instance vs. moving an instance
14:17:42 <edleafe> Decided that the next steps (returning alternate hosts, returning allocation_requests) will wait until Queens.
14:17:45 <edleafe> "Th-Th-The, Th-Th-The, Th-Th... That's all, folks!"
14:18:22 <mriedem> ok i'm +2 on https://review.openstack.org/#/c/487589/ since i was last night, only changes since have been the sorting fixes for the tests
14:18:22 <patchbot> patch 487589 - nova - placement: account for move operations in claim
14:18:49 <bauzas> I need to see Jenkins first
14:18:49 <mriedem> we need to see a good run with the claims in the scheduler change and check the logs to make sure the migration/resize tests are doing ok with it
14:18:56 <bauzas> yeah that
14:19:10 <mriedem> also, gibi is working on a functional test for some of the stuff that is going on in these two changes
14:19:24 <cdent> gibi++
14:19:34 <mriedem> to do a resize confirm/revert and make sure allocations are cool both ways
14:19:36 <bauzas> oh man, s/Jenkins/jobs :)
14:19:43 <gibi> sure
14:20:06 <bauzas> so used to say J after 5 years of thinking Hudson :)
14:20:18 <mriedem> but as https://review.openstack.org/#/c/483566/ is holding up pike-3, we need re-review on that to make sure we are ok with what's there while the tests run
14:20:19 <patchbot> patch 483566 - nova - claim resources in placement API during schedule()
14:20:38 <mriedem> i think the -1s are handled in the change below
14:20:53 <mriedem> moving on
14:20:57 <mriedem> alex_xu: api?
14:21:09 <alex_xu> we talk about the things should be merged before FF
14:21:10 <bauzas> mriedem: that's why I want to see the logs before removing my -1
14:21:39 <mriedem> https://blueprints.launchpad.net/nova/+spec/api-no-more-extensions-pike is complete
14:21:43 <alex_xu> #1 novaclient 2.53, #2 the left patches of no more extension
14:21:56 <alex_xu> all of those are merged
14:22:07 <cdent> no more extensions!!!! yay!
14:22:19 <mriedem> the wsgi_log_format option deprecation is merged too
14:22:24 <sdague> nice job alex_xu
14:22:30 <alex_xu> also talk about some api doc in the devref should be updated, that will continue after FF
14:22:32 <mriedem> the only other thing to do now is update the docs for the api plugins and stable api pages
14:22:59 <alex_xu> mriedem: yea
14:23:02 <alex_xu> sdague: thanks
14:23:21 <mriedem> alex_xu: ok is that all?
14:23:30 <alex_xu> also talk about a bug from Kevin_Zheng
14:23:32 <alex_xu> #link https://review.openstack.org/486850
14:23:33 <patchbot> patch 486850 - nova - Wrong href link returned when providing non-existe...
14:24:20 <alex_xu> we should correct the URL in the version API, sdague also suggestion return 404 for invalid path. Then probably just needs to review the patch
14:24:28 <alex_xu> mriedem: that is all now :)
14:24:37 <mriedem> thanks
14:24:45 <mriedem> gibi: notifications?
14:24:58 <gibi> The bp additional-notification-fields-for-searchlight is completed for Pike. We will open a follow up bp for Queens.
14:25:01 <gibi> The rest of the notification transformation patches also needs to move under a Queens bp.
14:25:04 <gibi> Nothing critical is missing from the current master for Pike from notification perspective.
14:25:08 <gibi> But we are still working on some bugs like #link https://bugs.launchpad.net/nova/+bug/1684860
14:25:08 <openstack> Launchpad bug 1684860 in OpenStack Compute (nova) "Versioned server notifications don't include updated_at" [Medium,In progress] - Assigned to Balazs Gibizer (balazs-gibizer)
14:25:10 <gibi> That is all
14:25:24 <mriedem> the fix for ^ is https://review.openstack.org/#/c/475276/ and has a +2 from me
14:25:25 <patchbot> patch 475276 - nova - Add 'updated_at' field to InstancePayload in notif...
14:25:32 <mriedem> if someone else can take a look
14:25:54 <mriedem> moving on
14:26:00 <mriedem> nova/cinder stuff
14:26:04 <mriedem> there are 2 changes left
14:26:26 <mriedem> the live migration patch from stvnoyes https://review.openstack.org/463987
14:26:27 <patchbot> patch 463987 - nova - update live migration to use v3 cinder api
14:26:35 <mriedem> which has had quite a bit of testing and review already
14:26:46 <mriedem> and the api change at the end of the series https://review.openstack.org/330285
14:26:46 <patchbot> patch 330285 - nova - Implement new attach Cinder flow
14:27:10 <mriedem> i'm -1 on ^ because there are a bunch of blockdev errors in the logs on that one, so something regressed in the new flow there, and jgriffith is investigating
14:27:44 <mriedem> i think it's safe to say at least that last one isn't going to make pike at this point
14:27:57 <mriedem> but i'm hopeful for early queens
14:28:25 <mriedem> #topic stuck reviews
14:28:38 <mriedem> there was nothing on the agenda, so does someone have a patch they need to mention here?
14:28:59 <mriedem> ok then
14:29:02 <mriedem> #topic open discussion
14:29:09 <darland> Hello everyone, me and my team have written following spec https://review.openstack.org/#/c/486947/
14:29:10 <mriedem> there is one item
14:29:10 <patchbot> patch 486947 - nova-specs - SPEC: nova disk quota tracking
14:29:31 <darland> so, we are looking for review
14:29:44 <sdague> mriedem: did you get any clarity on all the doc urls being 404?
14:29:45 <mriedem> darland: one point i'll make regarding timing,
14:30:07 <darland> this spec is about adding quota for root and ephemeral disc storage allocated to VM during boot based on flavor's value
14:30:12 <mriedem> darland: today is feature freeze for pike, so most people are focused on that, and will be focused on stabilizing and documenting pike for the rc1 in 2 weeks
14:30:25 <mriedem> darland: queens specs likely won't get any attention for a few weeks
14:30:57 <mriedem> darland: if anyone from your team is going to be at the PTG you can add it as a topic here https://etherpad.openstack.org/p/nova-ptg-queens
14:31:08 <darland> that's is ok
14:31:14 <mriedem> dansmith: sdague: jaypipes: any obvious historical reasons we haven't tracked disk quota in the past?
14:31:31 <sdague> mriedem: nothing that I remember
14:31:45 <dansmith> there's no quota on disk?
14:31:49 <mriedem> no
14:31:49 <sdague> other than the horrible inaccuracies around shared storage
14:31:50 <jaypipes> mriedem: no idea. it's news to me, actually :(
14:31:57 <mriedem> sdague: same
14:31:59 <dansmith> that seems really odd to me
14:32:10 <mriedem> the problem being the PhD you need to know how much disk is used and where
14:32:22 <mriedem> based on the flavor and bfv
14:32:49 <sdague> yeh, mostly I think it's just a hard counting problem. The autosizing disks as well.
14:32:56 <darland> mriedem: none of us, our team, is going to be at PTG
14:33:02 <mriedem> darland: ok
14:33:14 <sdague> darland: who is your team?
14:33:26 <mriedem> darland: maybe shop the idea around the operators community too to see if anyone has a need for this as well, or already implemented a solution
14:34:03 <darland> sdague: sberbank dev group
14:34:29 <sdague> darland: cool, thanks
14:34:30 <mriedem> darland: there is an upcoming ops meeting,
14:34:38 <mriedem> might also be worth trying to get discussion ther
14:34:39 <mriedem> *there
14:34:46 <mriedem> s/meeting/meetup/
14:34:50 <bauzas> there are old traces
14:34:57 <bauzas> https://blueprints.launchpad.net/nova/+spec/nova-disk-quota-tracking
14:35:02 <mriedem> https://www.eventbrite.com/e/mexico-city-openstack-operators-meetup-tickets-34989052197
14:35:05 <sdague> honestly, is it controversial? I feel like the concept is ++
14:35:14 <sdague> it's just sorting out the details of making it accurate
14:35:31 <bauzas> the fact is, looks like disk quota is more Cinder-ish
14:35:41 <bauzas> as per the findings Google gives me
14:36:13 <sdague> bauzas: that's the issue in the spec. You can do it with cinder, if you make your entire environment bfv for everything
14:36:23 <mriedem> i don't know if it's controversial, but i also don't know if there was a good reason for not having it years ago
14:36:24 <sdague> but they want it to work with cloudy things
14:36:36 <sdague> mriedem: I would not assume there was a good reason
14:36:40 <bauzas> https://ask.openstack.org/en/question/8518/disk-quota-for-projects/ has an answer
14:36:52 <bauzas> but not for local disks
14:36:57 <bauzas> assuming those are cheap, maybe?
14:37:00 <sdague> I would just assume we never had reasonable enough storage accounting to make it viable
14:37:07 <bauzas> probably yeah
14:37:15 <jaypipes> darland: as sberbank is Mirantis' customer, we can work together to discuss it at the PTG (I'll be in Denver)
14:37:16 <dansmith> I dunno, in the early days,
14:37:21 <dansmith> storage accounting was much simpler
14:37:44 <dansmith> so I'm really surprised we don't have it from the local-only days, and it's just really wrong now
14:38:09 <sdague> pre folsom... I'm not that surprised :)
14:38:27 <dansmith> um, what?
14:38:37 <dansmith> I'm saying before bfv and ceph,
14:38:42 <dansmith> when most of the quotas stuff was done,
14:38:52 <dansmith> accounting for how much disk you were using was much more straightforward
14:38:59 <mriedem> ok, so it sounds like there are some threads to pull - the mechanics of how to actually do this based on the variables involved has to be hashed out, especially now that we're counting quotas
14:39:00 <sdague> yeh, it just looks like quotas were only ever implemented on block devices
14:39:16 <sdague> probably because rax didn't need them for ephemeral
14:39:25 <dansmith> I'd expect that disk was never a limiting factor
14:39:30 <sdague> yep
14:39:33 <dansmith> too cheap to care
14:39:36 <bauzas> yup
14:39:51 <bauzas> (16:36:57) bauzas: assuming those are cheap, maybe?
14:40:02 <mriedem> ok so if there is nothing else i say we get back to it
14:40:03 <sdague> yep, so we shouldn't assume there was any good reason. Move forward as ++ on the idea. Details need to be ironed out
14:40:09 <sdague> mriedem: docs 404?
14:40:19 <mriedem> sdague: https://review.openstack.org/487863
14:40:20 <patchbot> patch 487863 - openstack-manuals - Update sitemap
14:40:47 <sdague> mriedem: will that also put in redirects?
14:40:59 <sdague> also, my browser is crashing trying to open that file
14:41:13 <mriedem> i'm not sure, but that's what AJaeger said was one thing that needed to be done
14:41:20 <mriedem> i expect we'll have to put in redirects
14:41:33 <mriedem> i made a note about that in https://etherpad.openstack.org/p/nova-pike-cells-v2-todos L5
14:42:05 <mriedem> this is how suse wins all the LOC championships right? :)
14:42:10 <mriedem> 18K lines of xml
14:42:36 <mriedem> ok anything else?
14:42:56 <mriedem> alright thanks everyone
14:42:59 <mriedem> #endmeeting