21:00:00 <dansmith> #startmeeting nova
21:00:01 <openstack> Meeting started Thu Feb  4 21:00:00 2016 UTC and is due to finish in 60 minutes.  The chair is dansmith. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:02 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:04 <openstack> The meeting name has been set to 'nova'
21:00:10 <tonyb> o/
21:00:11 <dansmith> I'm not doing the ping thing because I think it's silly
21:00:17 <bauzas> heh
21:00:19 <bauzas> \o
21:00:20 <ctrath> o/
21:00:22 <dansmith> Welcome to "Dansmith reads the meeting agenda"
21:00:25 <dansmith> I'm your host, dansmith
21:00:30 <dansmith> #topic Release Status
21:00:44 <dansmith> Final date for non-priority stuff is .. tomororw
21:00:47 <dansmith> also, tomorrow
21:01:01 <dansmith> johnthetubaguy sent out a thing, which you should read:
21:01:10 <dansmith> #link http://lists.openstack.org/pipermail/openstack-dev/2016-February/085455.html
21:01:22 <dansmith> any questions or concerns?
21:01:42 <dansmith> everyone is aware that mitaka-3 is Mar 1-3 which will be here before you know it,
21:01:49 <dansmith> thanks to our tricky short month of February
21:02:16 <dansmith> okie doke
21:02:19 <Vek> heh
21:02:28 <dansmith> #topic Regular Reminders
21:02:40 <dansmith> brush your teeth twice a day
21:02:43 <dansmith> also, this:
21:02:48 <tonyb> Be nice to your monther
21:02:51 <dansmith> #link https://etherpad.openstack.org/p/mitaka-nova-priorities-tracking
21:02:57 <Vek> heh
21:02:57 <dansmith> tonyb: also a  good reminder, yes
21:03:22 <bauzas> monther ?
21:03:24 <dansmith> in addition to the nova priority reviews, you should always review my patches
21:03:30 <Vek> haha :)
21:03:42 <dansmith> bauzas: I believe he meant "mother" but .. who knows about those upside down monther-lovers
21:04:17 <Vek> I figured it was probably a response to dan's comment about February :)
21:04:27 <tonyb> I talk/type funny remember?
21:04:38 <dansmith> tonyb: you're excused, yes
21:04:44 <bauzas> :)
21:04:47 <tonyb> \o/
21:04:50 <dansmith> okay, anything else on this electrifying topic?
21:05:07 * Vek flips the switch
21:05:13 <dansmith> #topic Bugs / gate status
21:05:24 <dansmith> so, I put this first thing on here, before I realized I have to run this thing
21:05:43 <dansmith> and I also forgot this week was a late meeting, because I mostly wanted to confront johnthetubaguy here
21:06:00 <dansmith> but basically, nova consumes a lot of gate resources, which are rapidly becoming more and more rare
21:06:08 <dansmith> 21 nodes on every patch in check for nova
21:06:11 <dansmith> that's like, a lot and stuff
21:06:27 <tonyb> dansmith: Yeah I fell like we need a different crowd to make thew call but reducing things seems like an ok idea
21:06:29 <dansmith> we have a bunch of jobs that I think we should consider dropping because they cost (someone) money and I think are less useful
21:06:45 <dansmith> like v2.0 jobs (two of them) and tests for some weirdo database called postgres
21:07:00 <dansmith> I also hear that the ceph job is not very stable,
21:07:07 <bauzas> probably a good ask for some 3rd party CI instead ?
21:07:15 <Vek> *nod*
21:07:15 <dansmith> but we're spending resources on it, AND like 90% of real deployments use it
21:07:26 <dansmith> so I kinda think that we should maybe focus more on making that job suck less
21:07:41 <dansmith> which I think could come in the form of folding that into our default job, to make it everyone's problem
21:07:46 <dansmith> but I know that's not super popular
21:08:02 <dansmith> bauzas: yeah, we could try to farm out some of the weird configurations
21:08:33 <dansmith> so anyway, I propose we leave this on the agenda and bring it back up next week in the presence of folks like johnthetubaguy, mriedem, sdague, etc
21:08:43 <bauzas> ++
21:08:45 * Vek will +1
21:09:17 <dansmith> any other comments on this before we move on?
21:09:52 <tonyb> just on bugs
21:09:55 <dansmith> #topic Bugs / 3rd party CI
21:10:08 <dansmith> so looking at the report:
21:10:12 <dansmith> #link http://ci-watch.tintri.com/project?project=nova&time=7+days
21:10:13 <tonyb> I guess we shoudl start lookign at things that aren't going to land in M but would be worth a backport
21:10:28 <dansmith> looks like hyper-v isn't doing so hot
21:10:47 <dansmith> but someone from Intel NFV CI popped up today sounding eager to expand and stabil-ify things there, which is good
21:10:52 <tonyb> dansmith: Yeah I've been talking to cbeliu aboiut it
21:10:56 <dansmith> okay
21:11:00 <tonyb> they kno the problem but not how to fic
21:11:04 <tonyb> fix even
21:11:06 <Vek> oh, cool..
21:11:07 <dansmith> okay
21:11:22 <dansmith> the intel one seems more stable than I recall, according to the report, which is nice
21:11:23 <tonyb> the problem is there are a few hyperv related patches backing up ....
21:11:37 <dansmith> I'd like to go look at some of those failures to see whether they're legit or not
21:12:11 <melwitt> tonyb: what does that mean?
21:12:12 <dansmith> tonyb: problem meaning they're stacking up because we don't have CI reads on them and thus aren't approving?
21:12:36 <tonyb> dansmith: Yeah, kinda
21:12:46 <dansmith> I'm pretty okay with that
21:13:00 <tonyb> dansmith: I don't know that they have core attention but there are a few that I wont +1 without a vote
21:13:01 <dansmith> we wouldn't be landing any patches if regular infra CI was hosed
21:13:08 <dansmith> tonyb: yeah
21:13:41 <tonyb> at least one is security related which would be good to get before M3
21:13:56 <tonyb> anyway it's a problem it sucks but we can't fix it :(
21:14:14 <dansmith> yeah, well... :/
21:14:31 <Vek> tonyb: is there enough unit testing on that security-related one to test the security hole being closed?
21:14:53 <Vek> or is it something that can only be properly tested by a CI job?
21:15:02 <tonyb> Vek: that's a good question.
21:15:09 <dansmith> Vek: well, the testing most people are waiting for is to make sure it doesn't break anything, not necessarily just that it plugs the hole
21:15:13 <tonyb> Vek: the fix adds unit tests which I'm happy enough with
21:15:17 <melwitt> dansmith: +1
21:15:28 <Vek> good point.
21:15:58 <dansmith> okay, anything else on 3rd party CI?
21:16:31 <dansmith> #topic Bugs / reminders
21:16:44 <dansmith> So, markus has a couple things here:
21:16:55 <dansmith> the first is looking for people to do bug skimming duty:
21:17:02 <dansmith> #link https://wiki.openstack.org/wiki/Nova/BugTriage#Weekly_bug_skimming_duty
21:17:25 <dansmith> requested 1 week compulsory bug skimming duty - no salary or stipend offered
21:17:41 <dansmith> and also, that meetings are starting up Feb 9th, which is Monday I think:
21:17:47 * alaski sneaks in late
21:17:48 <dansmith> #link http://lists.openstack.org/pipermail/openstack-dev/2016-January/084543.html
21:17:58 <dansmith> alaski: good of you to join us
21:18:11 <dansmith> any other bug-related reminders?
21:18:12 <alaski> uh oh, the teacher noticed
21:18:30 <tonyb> alaski: you're not very sneaky
21:18:31 * Vek slips alaski an apple to give to the teacher
21:18:54 <dansmith> Vek: unless it's coated in fat or candy, it ain't gonna help
21:19:01 <Vek> hehe :)
21:19:07 <auggy> i don't mind signing up again to help out with the bug skimming
21:19:20 <dansmith> auggy: cool
21:19:36 <bauzas> no criticals \o/
21:19:49 <dansmith> oh, heh, I totes missed the critical section
21:20:02 <auggy> i'm still pretty new to nova so i'm slow at it but everyone has been really helpful :)
21:20:05 <dansmith> #topic Bugs / {critical,stable status}
21:20:20 <dansmith> I don't know that we have anything for either of these bits.. anyone?
21:20:23 <dansmith> bueller?
21:20:35 <bauzas> (22:19:36) bauzas: no criticals \o/
21:20:36 <Vek> *crickets*
21:20:41 <dansmith> schweet
21:20:55 <dansmith> #topic Stuck reviews
21:20:57 <bauzas> tonyb: stable is stable ?
21:21:14 <tonyb> bauzas: for nova yeah
21:21:31 <tonyb> actually in general it's about as good as it gets
21:21:35 <dansmith> #undo
21:21:36 <openstack> Removing item from minutes: <ircmeeting.items.Topic object at 0x95ac3d0>
21:21:50 <dansmith> sorry, I was thinking that without mriedem here, we'd have nothing
21:22:03 <tonyb> dansmith: we still have nothing ;P
21:22:08 <dansmith> excellent
21:22:11 <dansmith> #topic Stuck reviews
21:22:13 <bauzas> \o/
21:22:22 <dansmith> I believe there to be no stuck reviews because the list on the wiki is empty
21:22:29 <dansmith> any disagreements?
21:22:36 <bauzas> go go go
21:23:03 <dansmith> alrighty then
21:23:15 <dansmith> #topic Open Discussion
21:23:25 <dansmith> who put this docimpact thing on the agenda?
21:23:27 <dansmith> I don't have context
21:23:34 <bauzas> sdague ?
21:23:36 <tonyb> probably sdague
21:23:40 <bauzas> I remember some convo we had
21:23:46 <bauzas> + markus
21:23:55 <tonyb> I think the impact for us is DocImapct will now land in the nova queue to triage
21:24:08 <bauzas> so, basically, just a reminder at least that adding DocImpact is bugging the doc team
21:24:14 <tonyb> which will suck for markus and auggy ;P
21:24:26 <alaski> http://lists.openstack.org/pipermail/openstack-dev/2016-February/085733.html I think
21:24:31 * auggy writes a perl script to make them all wontfix ;D
21:24:32 <dansmith> hrm
21:24:39 <Vek> heh
21:24:41 <bauzas> tonyb: it's set for doc, not nova AFAIK
21:24:49 <tonyb> auggy: *cough* python *cough*
21:24:51 <dansmith> auggy: you have got to stop answering things with "write.*perl"
21:24:56 <bauzas> but the problem is that it mostly overloads the doc team
21:25:10 <bauzas> while it should really be just a reno note IMHO
21:25:13 <tonyb> bauzas: okay I'll re-read but that's not what I recall
21:25:25 <anteaya> my read of the docimpact thing is that it stops overloading the docs team
21:25:35 <bauzas> #link https://wiki.openstack.org/wiki/Documentation/DocImpact
21:25:36 <anteaya> and that puts bugs in project queues
21:25:43 <anteaya> but I may be out of date
21:25:49 <tonyb> bauzas: I'm not sure I agree with you and sdague that everything that needs documenting implies a reno but ....
21:25:56 <melwitt> I thought we were supposed to use reno note instead of DocImpact. are we supposed to do both?
21:26:02 <dansmith> I'm thinking that will cause us to just stop using DocImpact, for better or worse
21:26:41 <bauzas> https://git.openstack.org/cgit/openstack-infra/project-config/tree/gerrit/projects.yaml#n2188
21:26:55 <tonyb> If nothgin else DocImpact will now be a review flag to ask what need documenting and then eitehr update devref or reno
21:26:59 <bauzas> => openstack-manuals
21:27:08 <bauzas> so
21:27:12 <bauzas> my point is
21:27:20 <tonyb> bauzas: right that's *now* but there is a spec out to alter the behaviour
21:27:39 <bauzas> tonyb: because the docs team can't handle all our load
21:27:46 <dhellmann> bauzas : http://lists.openstack.org/pipermail/openstack-dev/2016-February/085733.html
21:27:50 <tonyb> https://review.openstack.org/#/c/276065/1/gerrit/projects.yaml
21:27:50 <bauzas> hence the point we're discussing now
21:28:06 <bauzas> I see
21:28:08 <tonyb> So yeah it'll be down to us to
21:28:10 <auggy> i can talk to markus about if there's any stats we might want to keep track of regarding the nova bug impact
21:28:19 <tonyb> a) decide if it's a reno or a devref
21:28:22 <tonyb> an ;
21:28:34 <tonyb> b) add openstack-manuals if it impacts the stuff they write
21:28:40 <bauzas> our code-review policy says that what needs DocImpact needs a note
21:29:16 <bauzas> honestly, I'm by far in favor of having only one way to get the release comments - and that's reno IMHO
21:29:40 <tonyb> bauzas: sure but reno doesn't cover all the cases
21:29:50 <bauzas> examples ?
21:30:04 <tonyb> bauzas: for waht's designed for it's great but we need a policy that covers the rest
21:30:06 <anteaya> reno is for release notes, docimpact is for the openstack-manuals you need both
21:30:20 <anteaya> or I hope you think you need both
21:30:23 <bauzas> sorry I was unclear
21:30:27 <tonyb> bauzas: say the api v2 -> v2.1 switch
21:30:30 <bauzas> what I see is
21:30:37 <tonyb> it's needs a reno *and* install/ops updates
21:30:58 <tonyb> so we handle the reno in nova and *also* add it to the docs queue
21:31:17 <tonyb> somethings might be appropriate only for devref
21:31:28 <bauzas> how can us know what can be modified in docs land ?
21:31:35 <bauzas> devref is differnet
21:32:01 <tonyb> bauzas: You make a "best guess" to triage it if we're wrong then docs will reject it but it will still lighten thier load
21:32:08 <bauzas> we have the install guides, the ops guides, the security guide, the CTO guide (kidding)
21:32:15 <dansmith> heh
21:32:18 <melwitt> the example I had recently was a new nova-manage command. reno covers the announcement of the new command, DocImpact would be needed to get the usage info in the manuals
21:32:33 <bauzas> tonyb: hence me thinking that what changes for our operators are in release notes
21:32:47 <tonyb> bauzas: sure devref is differnt but today we tagg things as DocImact to mean all 3 things (reno, devref and manuals) we're just refining the process
21:32:51 <bauzas> melwitt: sure, but the tag won't help
21:33:28 <alaski> bauzas: the tag is a reminder to go and do it
21:33:52 <tonyb> Or to not +W tha review without the work being in the patch or a dependant patch
21:34:09 <bauzas> okay, fair enough, but how can we define when asking to add that tag ?
21:34:16 <bauzas> what could be the policy ?
21:34:31 <bauzas> like we have for http://docs.openstack.org/developer/nova/code-review.html#when-a-release-note-is-needed
21:35:03 <tonyb> bauzas: I'll take a TODO to write up something to have a robust discussion over :)
21:35:19 <tonyb> bauzas: in short I don't really think it's that hard
21:35:50 <tonyb> moving on?
21:35:57 * dansmith fell asleep
21:35:58 <bauzas> mmm, thanks for helping, I'd love to see that list
21:36:10 <bauzas> dansmith: on my TZ ?
21:36:14 * tonyb has to take the kids to school in 5 ....
21:36:22 <anteaya> tonyb: safe travels
21:36:26 <dansmith> does anyone know anything about this novaclient cinder keystone thing?
21:36:39 <anteaya> I do not
21:36:52 <Vek> other than that it caused functional tests to fail, not really.
21:36:55 <tonyb> dansmith: I'm guessing it's fallout from the turn on v3 in devstack from last week
21:37:14 <dansmith> tonyb: right, but .. do you know what we need to discuss about it?
21:37:24 <Vek> https://review.openstack.org/264764 probably needs work
21:37:49 <dansmith> melwitt was a reviewer.. any comments?
21:37:59 <Vek> looks like it hasn't been touched since beginning of January, too...
21:38:30 <melwitt> dansmith: just that the work needs to be done to add v3 compat in novaclient
21:38:51 <melwitt> but most of the time the answer is, just use openstackclient
21:39:02 <auggy> i remember notmorgan was working on a patch that sdague was reviewing, but i don't know if it's related
21:39:07 <auggy> i'll see if i can find it
21:39:20 <dansmith> that's catalog related I think
21:39:27 <dansmith> which may be somewhat related, I dunno
21:39:33 <auggy> yeah i got nothin'
21:40:03 <tonyb> auggy: that was different
21:40:10 <tonyb> auggy: that was glance stuff
21:40:30 <auggy> tonyb: ah ok, i was thinking it was keystone auth related, sorry for the distraction
21:40:41 <Vek> Monty was working on https://review.openstack.org/#/c/245200/, but I don't think that's actually related; it just comes up when I search my mailbox for "keystone"
21:41:00 <tonyb> auggy: just trying to save you looking for a thing that doesn't exist ;P
21:41:32 * tonyb ducks out early.
21:41:41 <alaski> o/
21:42:18 <dansmith> okay, I think the midcycle thing is old now,
21:42:26 <dansmith> so .. anything else?
21:42:38 <dansmith> frankly I'm sick of this meeting after 42 whole minutes of it
21:42:45 <Vek> heh :)
21:42:49 <anteaya> lovely reading of the agenda dansmith
21:42:49 <Vek> stick a fork in it, then :)
21:42:54 * dansmith bows
21:43:15 <dansmith> going thrice...
21:43:44 <dansmith> #endmeeting