20:59:57 #startmeeting nova 20:59:58 Meeting started Thu May 26 20:59:57 2016 UTC and is due to finish in 60 minutes. The chair is dansmith. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:59:59 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:00:01 The meeting name has been set to 'nova' 21:00:07 o/ 21:00:10 o/ 21:00:11 I am religiously opposed to the pre-ping, so I'm not doing it 21:00:19 o/ 21:00:26 \o 21:00:31 \o 21:00:35 /o/ 21:00:38 \o 21:00:59 * edleafe admires dansmith's convictions 21:01:01 seven people isn't even quorum, right? guess we have to cancel? 21:01:09 o/ 21:01:10 heh 21:01:10 ./ 21:01:18 #topic Release News 21:01:23 o/ 21:01:25 o/ 21:01:27 hi 21:01:27 #link Newton release schedule: https://wiki.openstack.org/wiki/Nova/Newton_Release_Schedule 21:01:31 #info June 2: newton-1, non-priority spec approval freeze 21:01:46 anyone have anything here? 21:02:03 obviously the upcoming deadline should be on everyone's minds 21:02:12 I've been trying to do mostly reviews this week including specs 21:02:22 which has me in a great mood 21:02:29 ++ 21:02:34 (sarcasm) 21:02:44 moving on in 10 seconds 21:02:49 * auggy just always assume dansmith == sarcasm 21:02:55 #topic Bugs 21:03:02 bugs, we have 'em 21:03:14 the gate has been pretty good recently I think 21:03:27 the EU/APC bugs meeting moved .... 21:03:34 #link check queue gate status http://status.openstack.org/elastic-recheck/index.html 21:03:48 tonyb: oh? to when or where? 21:03:54 2 hours earlier AFAIR 21:04:11 like 0700UTC 21:04:15 tonyb: correct? 21:04:18 2 hours earlier 21:04:21 #link http://eavesdrop.openstack.org/#Nova_Bugs_Team_Meeting 21:04:21 #info EU/APAC bugs meeting moved -2 hours 21:04:47 cool, thanks 21:04:50 well, 0700UTC is not really EU-friendly, but whatever... :) 21:05:03 #link http://ci-watch.tintri.com/project?project=nova&time=7+days 21:05:10 I think xenserver CI is back on track, right? 21:05:26 I saw some thread about the Intel CI 21:05:27 also of note for this section is that DB2 CI went away sometime in the past, intentionally, and isn't coming back 21:05:32 bauzas: 0800, but yeah 1000 was better 21:05:38 we celebrated with a removal of all the super annoying DB2isms from the migrations 21:06:02 *celebrated* is that code? 21:06:04 \o/ 21:06:10 anything else here? any critical bugs to highlight? 21:06:51 i just had another item, markus_z mentioned he'd be expiring old open bug reports, details in this ML post - http://lists.openstack.org/pipermail/openstack-dev/2016-May/095654.html 21:06:59 ah yes, thanks 21:07:06 #link http://lists.openstack.org/pipermail/openstack-dev/2016-May/095654.html 21:07:11 *very* old bugs :) 21:07:21 Automatic closing of 185 bug reports which are older than 18 months in the week R-13 21:07:32 18 months is just prehistory 21:07:42 people on twitter informed me that we're bad for doing this, but the rest of the people that do actual work all agreed this was a good plan 21:08:03 ranting on twitter is just not a fair game 21:08:04 anything else on bugs? 21:08:17 I think we're bad but it's the least terrible option 21:08:22 heh 21:08:32 just so dan can include IRC in the list of bad places ;P 21:09:01 #topic Reminders 21:09:06 #link Newton review focus list: https://etherpad.openstack.org/p/newton-nova-priorities-tracking 21:09:15 if you're not looking at that to know what to review, you should be 21:09:23 if you're not updating it for your subteam, then you can't complain 21:09:53 #info We have 59 approved blueprints: https://blueprints.launchpad.net/nova/newton - 6 are completed, 5 have not started, 3 are blocked 21:10:16 * dansmith giggles at 59 approved 21:10:34 and, final reminder: 21:10:35 #help https://wiki.openstack.org/wiki/Nova/BugTriage#Weekly_bug_skimming_duty Volunteers for 1 week of bug skimming duty? 21:11:09 looks like there were a couple people for this week 21:11:27 next week needs some non-US not-holiday-having people to sign up 21:11:54 if you're interested in helping, put your name on that list.. no license or proof of insurance required 21:12:10 also if you have questions on what it entails, feel free to ask me or markus_z 21:12:18 any other reminders? 21:12:42 alrighty then 21:12:50 #topic Stable branch status 21:12:57 #link Stable branch status: https://etherpad.openstack.org/p/stable-tracker 21:13:11 just before the meeting I looked at the two open stale branch bug lists: 21:13:16 #link stable/mitaka: https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:stable/mitaka,n,z 21:13:24 #link stable/liberty: https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:stable/liberty,n,z 21:13:32 nothing looked too badly on fire there 21:13:34 heeh 'stale' ;P 21:13:43 anyone need to highlight something? 21:13:49 tonyb: heh, oops, but not really :D 21:14:12 trumpian slip 21:14:19 tonyb: certainly you have something important and profound to say about stable things? 21:14:31 dansmith: Umm yeah ;P 21:14:39 my, how profound! 21:14:49 * dansmith contemplates deeply 21:14:55 just the normal ... the queue is a little long but I'll start on that today 21:15:00 more reviews is good 21:15:03 the gate is fine 21:15:07 woot 21:15:08 kilo is dying but not dead 21:15:21 * dansmith hands tonyb some heavier artillery to finish it off 21:15:31 :) 21:15:41 #topic Subteam Highlights 21:15:44 people always love to loose kilo(s) 21:15:48 alaski: say things 21:15:58 bauzas: boom tish 21:16:05 melwitt is helping to push on devstack changes around transport_url 21:16:06 :D 21:16:13 which will help with our migration commands to cellsv2 21:16:21 tonyb: too hard for people using imperial metrics, I guess 21:16:30 we discussed a potential race condition, and then found a solution 21:16:36 tons of stuff to review 21:16:39 fin 21:16:44 nice 21:16:54 edleafe: say fewer things 21:17:04 dansmith: ... 21:17:08 perfect 21:17:21 ROFL 21:17:57 edleafe: anything to actually report on scheduler? 21:18:25 still some conflict over fixed vs dynamic resource classes, etc 21:18:34 this took longer than laski, gw dansmith 21:18:46 jroll: heh 21:19:26 maybe some more eyes on https://review.openstack.org/312696 21:19:35 ^^ dynamic resource class spec 21:19:44 yeah, I need to look at that and rain some hate 21:20:04 ++ for d-r-c 21:20:07 * jroll kinda needs that 21:20:47 edleafe: anything else? you're holding up the schedule 21:20:50 * dansmith cracks himself up 21:21:02 lol 21:21:12 I thought this was async 21:21:21 nah, that's it 21:21:24 okay cool 21:21:35 I'm guessing PaulMurray isn't around 21:21:43 anyone want to say things about live migration? 21:21:57 I know that a bunch of storage pool stuff is getting worked out right now, 21:22:22 we've landed a good number of cleanups, the spec is under review, and folks are stacking up refactor patches like nobody's business 21:22:49 okay then 21:23:02 I think sdague is EOD.. alex_xu are you around to say things about API? 21:23:16 or oomichi ? 21:23:31 auggy: I'm a meeting agenda literalist 21:23:53 didn't mean to rain on your parade ;) 21:24:01 I guess they were still targeting at the api-ref changes 21:24:06 moshele isn't around to talk about sriov things, so we'll skip that 21:24:28 bauzas: yep, and I know oomichi has the patch up to remove 2.0, discussion about timing to follow 21:24:50 gibi: notifications? anything? 21:25:02 dansmith: orly? I'd so happy to use my axe 21:25:10 (for the v2.0 removal patch) 21:25:16 bauzas: yeah 21:25:28 okay, well, let's move on.. this is taking way too long 21:25:33 #topic Stuck Reviews 21:25:38 dansmith: :( 21:25:43 #undo 21:25:44 FWIW, impressive progress on api-ref http://burndown.dague.org/ 21:25:45 Removing item from minutes: 21:25:49 rlrossit: more things to say? 21:25:57 both specs in 1 nova, 1 oslo 21:26:06 syjulian working on oslo schema generation 21:26:15 gibi working on most of the transformation work 21:26:17 * rlrossit is done 21:26:24 okay cool, thanks 21:26:29 #topic Stuck Reviews 21:26:37 There are none listed on the wiki 21:26:52 also, I personally surveyed every open nova review before the meeting and have deemed none to be stuck 21:27:22 #topic Open Discussion 21:27:38 ohai, I have a thing 21:27:38 jroll: ironic serial console... did you see someone from NEC put up a spec for this? 21:27:45 I added you to it earlier today 21:27:49 dansmith: yes, there's a spec, I don't think we need a spec 21:27:59 the spec is pretty detail-free so having no spec is mostly equivalent to me 21:28:03 cool 21:28:06 dansmith: https://review.openstack.org/#/c/319507/ right? 21:28:09 jroll: you're just trolling for approval? 21:28:18 well, johnthetubaguy said the next step is add to agenda 21:28:20 jroll: that's the one 21:28:24 so I did 21:28:39 approval would be cool, should be pretty trivial code 21:28:52 alaski: you're the only other approver here.. you cool with that? if so I'll stamp 21:28:59 yes, I'm good 21:29:00 ironic start-console $node_id; return (host, port) 21:29:18 <3 21:29:20 thanks y'all 21:29:24 jroll, that's right, thank you 21:29:24 cool 21:29:27 * jroll goes off to grill things 21:29:36 alaski: you had a thing 21:29:42 yes, https://review.openstack.org/#/c/320416 21:29:55 this ties into the live-migrate refactor that was discussed at the summit 21:30:19 but it would change the error returns from the API 21:30:24 okay I'm thinking maybe sdague and PaulMurray as reviewers here, but neither are around 21:30:35 as in, it would no longer check preconditions on computes before returning a response 21:30:43 yes, they would be good to get input from 21:31:15 I mostly want to get consensus on supporting two paths, or just changing the error conditions 21:31:24 and I'd rather not support two paths 21:31:48 yeah 21:31:49 anyone with interest please comment on the review 21:32:00 and it should mostly be admin-facing api at this point, so.. probably not a huge deal 21:32:05 I hate to keep using that excuse, but.. 21:32:11 yes 21:32:19 and errors would surface via instance-actions 21:32:23 so they're not invisible 21:32:37 alaski: that's not what I understand when reading the spec or I'm wrong 21:32:58 bauzas: it changes the two calls into compute nodes to be casts 21:33:47 alaski: so, that does mean you're not guaranteed that a migration has been started when you return from the call 21:33:53 which might make some HA people twitchy 21:34:04 you should be able to monitor that it started through other mechanisms, but... 21:34:04 alaski: oh, I see, what he calls 'blocking' is the call from conductor to compute, not API to conductor 21:34:13 bauzas: yes 21:34:29 dansmith: I'd have to look, but we may not make that guarantee today either 21:34:37 ah, okay, based on above 21:34:38 okay cool 21:34:47 I think we're already async on the API side 21:34:49 it's the prechecks that block 21:35:01 that's just the conductor blocking for a while because of the compute limbo callds 21:35:01 okay, so anyway, comments requested on the review.. anything else we should discuss here/ 21:35:08 i have one quick thing 21:35:16 auggy: I mean about this 21:35:19 nope, just wanted awareness since it ties into something we agreed to do. 21:35:20 oh sorry 21:35:21 auggy: we have one more thing on the list before we move on 21:35:25 and it's sort of a large change 21:35:32 * alaski sits down 21:35:46 takashin: you have a re-approval spec you want 21:35:51 #link https://blueprints.launchpad.net/nova/+spec/add-swap-volume-notifications 21:35:57 takashin: that is a link to the blueprint, do you have a spec too? 21:36:04 It was approved for Mitaka. I would like to make it be approved for Newton. 21:36:13 Yes. 21:36:38 Sorry.No. 21:36:41 takashin: okay I don't see one 21:36:42 okay 21:36:55 I guess it was specless? 21:37:12 Yes.It was specless. 21:37:38 alaski: I don't think I've seen this before or know anything about it, but it's just adding a few notifications, so I guess we're good? 21:38:04 yeah, seems totally non contentious 21:38:17 it seems okay to me 21:38:27 okie dokie, done 21:38:45 okay that's the end of the list of open discussion things.. auggy you had a thing? 21:38:47 Thank you. 21:38:49 hopefully a very very short thing? 21:38:50 last week i updated the Mentorship wiki page with tasks for new contributors - https://wiki.openstack.org/wiki/Nova/Mentoring 21:39:12 it's still incomplete but if you have things to add, either let me know and i'll do it or feel free to add them 21:39:14 that's all :) 21:39:24 alright 21:39:30 anything else from anyone? 21:39:52 anyone? 21:39:58 anyone? 21:40:07 beuller? 21:40:13 newbie meta question about spec approval process 21:40:42 esnyder: shoot 21:40:53 is it appropriate to be proposing specs with api changes where upstream dependencies are still up in the error, or is that always bad form? 21:41:08 (re https://review.openstack.org/#/c/310050/) 21:41:43 esnyder: the up in the air bit here is the VSOCK stuff? 21:41:53 yes 21:42:10 right, so because that has to filter down to us through the kernel, qemu, and libvirt at least, 21:42:18 and then the distros, 21:42:35 approving that before things are much closer to being consumable is not likely to happen, IMHO 21:42:46 exactly; we're just trying to keep as much parallel motion going as possible, but don't want to be out of turn here 21:42:59 ok, thanks 21:43:08 yep, and that's cool.. I also appreciate the desire the use 9p as the proving ground, 21:43:17 a backlog spec might be more appropriate for now to capture discussion, and then move to a full spec when dependencies are in place 21:43:34 it's just that 9p has a lot of bad issues for nova, etc that it's not really a viable alternative even for a PoC I think 21:43:57 alaski: yeah, although I think they mostly have things fleshed out, so just merging the spec when the time is right is probably fine too 21:44:28 alaski: it's not like they have an idea and no plan to do it, they're just waiting for the goodness to get finished in another community so it can filter down to us 21:44:29 sure. the backlog spec just avoids the -2 parade at the end of the cycle 21:44:46 alaski: or we could -2 it now .. but yeah :) 21:44:49 either way 21:44:59 yeah, just mentioning it as an alternative 21:45:04 so i should just move it from newton/approved to newton/backlog? 21:45:05 yup 21:45:12 esnyder: you can if you want 21:45:17 esnyder: it's an option, not necessary 21:45:20 ok, thanks for the feedback 21:45:37 okay, anything else/ 21:45:51 this has been the worst "shortest meeting ever" attempt, so I apologize for that 21:45:57 I blame all the people showing up and talking 21:46:12 we should look at the logs and point out names 21:46:20 they know who they are 21:46:23 * auggy hides 21:46:30 I tried to be quiet 21:46:40 shush! 21:46:43 edleafe: yeah, you were part of the solution 21:46:56 okay, I believe we're done here.. someone second my motion to adjourn 21:47:03 +2 21:47:03 +2 21:47:04 seconded 21:47:04 seconded 21:47:05 i concur 21:47:08 * claudiub prepares tumble weeds 21:47:12 excellent, consensus 21:47:15 #endmeeting