20:59:52 #startmeeting nova 20:59:52 Meeting started Thu Oct 13 20:59:52 2016 UTC and is due to finish in 60 minutes. The chair is mriedem. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:59:53 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:59:56 The meeting name has been set to 'nova' 21:00:11 hi who's here? 21:00:13 o/ 21:00:16 o/ 21:00:17 Not me 21:00:26 mikal: you're here, just not your pants 21:00:31 \o 21:00:33 o/ 21:00:37 o/ 21:00:41 o/ 21:00:42 mriedem: I am glad we've minuted the state of my pants 21:00:46 :) 21:01:12 o/ 21:01:50 o/ 21:02:00 o/ 21:02:04 \o 21:02:08 alright let's do this 21:02:14 #link agenda https://wiki.openstack.org/wiki/Meetings/Nova#Agenda_for_next_meeting 21:02:22 #topic release news 21:02:27 o/ 21:02:31 #link Draft Ocata release schedule: https://wiki.openstack.org/wiki/Nova/Ocata_Release_Schedule 21:02:39 that's basically it for release news 21:02:45 lots of spec reviewing going on 21:02:55 #topic Summit planning 21:03:06 #link Nova Ocata design summit schedule: https://www.openstack.org/summit/barcelona-2016/summit-schedule/global-search?t=Nova%3A 21:03:15 #link Questions for ops folks welcome for the Nova & Ops: https://etherpad.openstack.org/p/BCN-ops-nova 21:03:23 johnthetubaguy is running ^ 21:03:36 #info Topic change: https://www.openstack.org/summit/barcelona-2016/summit-schedule/events/16977/ 21:03:49 ^ i posted to the dev list on that as well, swapped the docs session for security 21:03:51 Real Live johnthetubaguy ! 21:04:05 Your schedule line includes a lot of not-designs-summit stuff 21:04:06 i see the john hopkins guys aren't around 21:04:09 Like vBrownBags 21:04:32 looks fine to me 21:04:40 maybe it's your aussy browser 21:05:03 Huh, refreshing fixed it 21:05:04 Ignore me 21:05:07 ok 21:05:23 i was going to say, the results with and without the colon are very different 21:05:36 #help If you are chairing a session please work on getting your agenda written in the session etherpad: https://wiki.openstack.org/wiki/Design_Summit/Ocata/Etherpads#Nova 21:05:42 Oh, that's probably it 21:05:58 that means alaski dansmith jaypipes johnthetubaguy et al ^ 21:06:05 yessir 21:06:14 questions about the summit? 21:06:26 #topic bugs 21:06:30 there are no critical bugs 21:06:49 quite a few new untriaged bugs though 21:07:22 gate status is mostly ok 21:07:40 the newest thing is http://status.openstack.org/elastic-recheck/index.html#1630664 21:07:44 which is after TLS was turned on 21:07:50 we have random connection failures talking to neutron 21:08:07 there is a patch in ksa to be released which should give more details on what the actual failures are 21:08:20 credit to dansmith and sdague for fixing that 21:08:35 3rd party ci status 21:08:38 vmware has had issues 21:08:46 i turned off voting once this week 21:08:54 some tests were randomly failing again today 21:09:07 and there are no vmware people in here to give any details on what the issues are 21:09:49 so i'll be pushing the change to drop the vmware driver after the meeting 21:10:04 ... 21:10:14 * dansmith looks around the room 21:10:17 ha 21:10:20 i'm joking about that one 21:10:28 but really expected some response 21:10:34 mriedem: do it 21:10:38 I chuckled 21:10:39 We were all waiting... 21:10:42 ...but it never happened 21:10:47 mtreinish: oh of course you show up 21:10:53 Wait for the media to run articles about it I suppose? 21:11:05 i'm not on twitter so i wouldn't care 21:11:11 #topic stable branch status 21:11:20 #info stable/newton 14.0.1 was released this week 21:11:29 #info stable/mitaka goes to phase 2 and stable/liberty goes to phase 3 support tomorrow: http://docs.openstack.org/project-team-guide/stable-branches.html#support-phases 21:12:04 lyarwood has been active on propsing and reviewing backports, so there are things out there 21:12:08 *proposing 21:12:42 #topic subteam highlights 21:12:46 alaski: cells v2 meeting? 21:13:03 https://review.openstack.org/#/c/381275/ came up 21:13:05 has a +2 21:13:10 i just forgot to get back on it 21:13:10 not much to report, just that spec series 21:13:20 ok 21:13:36 edleafe: scheduler meeting? 21:13:48 sorry, on a call 21:13:58 i know the scheduler guys have been very quiet in channel the last 2 days 21:14:10 *rimshot* 21:14:15 thanks 21:14:29 Heh 21:14:31 ok tdurakov gave me the live migration meeting highlights before the meeting 21:14:35 CI (grenade job for live migration testing is in progress) 21:14:46 ^ the job actually exists for grenade + live migration, 21:15:00 the changes are for running live migration old <> new nodes back and forth in that job 21:15:07 with a tempest change from raj_singh 21:15:09 i'm +1 on the series 21:15:20 just need to grease some QA palms 21:15:33 Bugs (markus_z brought up the change for https://review.openstack.org/#/c/275801/ - johnthetubaguy to help with https://bugs.launchpad.net/nova/+bug/1566622 ) 21:15:33 Launchpad bug 1566622 in OpenStack Compute (nova) "live migration fails with xenapi virt driver and SRs with old-style naming convention" [High,Confirmed] 21:15:55 Open discussions (backward compatability during upgrades on hypervisors layer) 21:16:02 i'm not really sure what ^ means 21:16:35 but, that's part of what the live migratoin + grenade back and forth stuff is for testing 21:16:47 sdague: alex_xu_: api subteam meeting highlights? 21:16:57 mriedem: sorry about that. The gist of the scheduler meeting was getting jaypipes to focus on code and not so much on specs 21:17:24 the rest was on dividing the work on the placement API changes 21:17:30 ok 21:17:40 the api meeting was mostly about topics for the API session at the summit 21:17:41 which led the aforementioned quiet in the channel 21:17:56 alex_xu has an etherpad with some of that info 21:18:15 #link https://etherpad.openstack.org/p/nova_api_ocata_ideas 21:18:36 there was no sriov/pci meeting this week 21:18:41 gibi: notifications meeting? 21:19:06 there are changes https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/versioned-notification-transformation-ocata 21:19:10 #link https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/versioned-notification-transformation-ocata 21:19:26 there are also a few specs related to new notifications 21:19:32 several around compute node / host metrics 21:19:41 which i've nacked 21:19:50 #topic stuck reviews 21:19:59 nothing in the agenda - anyone need to pounce on this? 21:20:18 #topic open discussion 21:20:32 (jroll): Ironic portgroups BP, can we approve or does it need a spec? https://blueprints.launchpad.net/nova/+spec/ironic-portgroups-support 21:20:37 ohai 21:20:42 jroll: so at one point there was a spec for that 21:20:49 munged with the tenant networking spec 21:20:52 yeah, so this was split out from the general networking stuff 21:21:03 * jaypipes is here... mostly. 21:21:30 we dropped the bonding things from that spec 21:22:08 that said, we'll need to update instance metadata to get the bond configs into the configdrive / metadata api 21:22:08 #link ironic port groups support POC https://review.openstack.org/#/c/206163/ 21:22:13 so I could see some bikeshedding there 21:22:30 yeah so that poc doesn't have any of the metadata stuff right? 21:22:35 right 21:22:38 I think a poc for that exists 21:22:43 * jroll looks 21:22:47 There are a few requests for new metadata things 21:22:58 I was assuming we'd discuss those in the vendordata summit session 21:23:00 jroll: and duplicate things here https://review.openstack.org/#/c/382438/ 21:23:03 But that session seems more... keystoney 21:23:31 so here's a poc for doing the metadata in our driver, which doesn't solve metadata api https://review.openstack.org/#/c/289412/1/nova/virt/ironic/driver.py 21:23:52 mriedem: that's some dynamic portgroups thing that mirantis keeps pushing on, which we are totally not ready for yet 21:23:58 so ignore that pls 21:24:14 heh, tell them to slow down, you're not that kind of project 21:24:24 lol 21:24:28 so this doesn't sound trivial 21:24:33 or feature parity-ish 21:24:41 so i'm leaning toward spec city 21:24:43 or fun 21:24:46 or that 21:24:47 yeah, we don't get a hypervisor to HA networks for us :( 21:24:49 any disagreements? 21:25:01 well whose fault is that? 21:25:08 * jroll points at deva 21:25:34 i'm also interested in the QA stuff which i know you mentioned in the bp whiteboard 21:25:42 spec is fine with me, just wanted to be told so instead of assuming 21:25:42 sounds like new cloud-init required 21:25:57 no, current cloud-init works 21:26:00 and you guys have a special ironic multi-tenant networking job right? 21:26:03 cirros doesn't load the bonding kmod though 21:26:06 ah 21:26:15 right, but this could be added to any job 21:26:21 so i was wondering if the updated cirros (when that happens) could be in that job to test this 21:26:24 oh 21:26:35 i guess we just update devstack for the cirros version to use, sure 21:26:40 yeah 21:26:42 anyway, i think we agree on a spec 21:27:06 yeah, the metadata bits sell it for me 21:27:31 thanks 21:27:34 ok 21:27:42 (thorst / adreznec): Initial PowerVM Driver Integration BP proposed: https://review.openstack.org/#/c/381772/ 21:27:48 Hi all. We have another blueprint up to propose the *initial* integration of the PowerVM driver in tree to nova. 21:27:54 https://review.openstack.org/#/c/381772/ 21:28:00 Intention is to do this over multiple releases to be cognizant of reviewer bandwidth. 21:28:07 We’re asking for feedback on the blueprint, any requirements we haven’t considered, etc... 21:28:15 We’ve also tee’d up a discussion for it at the unconference in Barcelona. 21:28:27 * mikal braces 21:28:33 i haven't gone through the whole spec, but the gist is, 21:28:44 start with a couple of minimum operations in tree 21:28:50 create/delete, start/stop, console 21:29:02 and work things in over time? 21:29:12 exactly. All picked from the out of tree driver 21:29:15 Yeah, that's essentially the plan 21:29:16 we keep the two 'in sync' 21:29:29 which, is kind of how hyperv does new features btw 21:29:37 So, I don't think we should debate this in detail now, but I'd like to understand how we hedge against IBM having yet another change of corporate direction 21:29:53 So can we put that on the agenda for whenever we discuss this please? 21:30:04 yes it's in the unconference session 21:30:09 mikal: especially since there are like eight IBM hypervisors 21:30:09 Cool 21:30:18 3 z ones, two power ones 21:30:22 so only five I guess :) 21:30:23 3 power ones 21:30:27 3? 21:30:30 you forgot powervc 21:30:37 Standard exaggeration factor? 21:30:39 pkvm, powervc, powervm v2 21:30:42 yeah, so um.. great 21:31:10 But yeah, let's not get side tracked here but I do want to be part of that conversation please 21:31:35 sounds good. Hoping for some feedback on spec before conference if possible, but we'll chat more at the unconference 21:31:44 thanks! 21:31:47 sure, but before ammo'ing up to shoot them down at the summit, they've been working on this and attending things for 2 years now 21:31:57 mriedem: you take the fun out of everything 21:32:07 I think its about hedging 21:32:09 if it helps i -2ed the spec for the latest z driver yesterday 21:32:23 For example, if we can get gating into infra, then that would ease my concerns at least a little (for example) 21:32:26 right, so part of the hedging thing is getting adoption and more people involved besides 3 guys at ibm 21:32:30 and ci 21:32:34 ci is a requirement 21:32:41 _upstream_ ci 21:32:53 upstream doesn't ci vcenter? 21:32:54 ci is up and running, but we don't vote per last mid cycle guidance (I think it was midcycle) 21:32:55 or hyperv 21:33:12 mriedem: sure, those vendors don't have 8 hypervisors and indecisive VPs though 21:33:15 openstack-infra isn't going to host a power 8 system 21:33:15 but runs against all nova patches - and its public config/logs are available 21:33:26 and people use those hypervisors for cloudy things 21:33:31 Still don't want to side track this meeting 21:33:35 Let's argue about this at the unconference 21:33:54 I will take a look at the spec too because I am only partially a terrible person 21:34:03 fair enough, but i'd rather not argue about it every 3 months 21:34:21 powervm-just-the-tip is a decent starting course 21:34:25 b/c we go slow just like ironic 21:34:26 :) 21:34:36 anything else for open discussion? 21:34:56 ok, that's it then. thanks everyone. 21:34:58 #endmeeting