14:00:57 <PaulMurray> #startmeeting Nova Live Migration 14:00:58 <openstack> Meeting started Tue Jan 5 14:00:57 2016 UTC and is due to finish in 60 minutes. The chair is PaulMurray. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:59 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:01:01 <openstack> The meeting name has been set to 'nova_live_migration' 14:01:28 <PaulMurray> anyone here for live migration ? 14:01:32 <claudiub> o/ 14:01:34 <jlanoux> o/ 14:02:06 <markmc> hey 14:02:17 <paul-carlton1> o/ 14:02:25 <PaulMurray> I'll give it a minute 14:02:31 <andrearosa> happy new year 14:02:39 <mdbooth> o/ 14:02:50 * alex_xu lurks 14:02:53 <claudiub> happy new year :) 14:03:07 <PaulMurray> Thanks to andrearosa for reminding me of the time :) 14:03:31 <PaulMurray> Agenda is here as usual: https://wiki.openstack.org/wiki/Meetings/NovaLiveMigration 14:04:05 <PaulMurray> I would expect some people to be away so it could be short 14:04:18 <PaulMurray> but I thought it was worth trying to recap where we are 14:04:42 <PaulMurray> #topic Overall priority status 14:05:07 <PaulMurray> We are coming up to mitaka-2 milestone 14:05:36 <PaulMurray> and not many changes have been merged although there are quite a lot of patches up for review 14:06:03 <PaulMurray> So I'll go through the list of items 14:06:18 <PaulMurray> Anyone know where we are with CI? 14:06:33 <pkoniszewski> o/ 14:06:54 <PaulMurray> Hi pkoniszewski 14:07:06 <PaulMurray> is that a hand up for CI or just telling us your here? 14:07:25 <pkoniszewski> just saying hello, don't know about CI 14:07:36 <PaulMurray> ok 14:08:00 <PaulMurray> last I knew the job on the experimental queue seems to be working ok 14:08:09 <PaulMurray> remember to try it out 14:08:30 <PaulMurray> Next is API docs 14:08:48 <PaulMurray> I added some description to the concepts guide 14:09:11 <PaulMurray> As we go along it would be good to make sure the API is correct and up to date for any changes we make 14:09:23 <PaulMurray> and feel free to correct anything there 14:09:28 <alex_xu> PaulMurray: yea, thanks for the doc :) 14:09:57 <PaulMurray> How about "Instances with mix of local and shared disks" 14:10:15 <PaulMurray> We have three patches waiting for review 14:10:41 <pkoniszewski> there is one -1 i need to solve 14:10:50 <pkoniszewski> however, mikal gave +2 on series 14:11:10 <pkoniszewski> so I believe it is good to go 14:11:50 <PaulMurray> pkoniszewski, it will be good to get this one done 14:11:53 <PaulMurray> so thanks 14:12:00 <pkoniszewski> i will resubmit it today 14:12:19 <PaulMurray> next is "Force live migration" 14:12:59 <pkoniszewski> i saw your review PaulMurray, thanks for that, I will resubmit this one today also 14:13:34 <PaulMurray> Thanks - iirc it was all very small comments 14:14:00 <PaulMurray> ping me when your done and I will go over it again 14:14:09 <pkoniszewski> sure, thanks! 14:14:50 <PaulMurray> next is "Resize/Cold Migrate" - paul-carlton1 ? 14:14:56 <PaulMurray> or mdbooth 14:15:25 <claudiub> umm, is there a link for the force live migration? 14:15:30 <claudiub> i am curious 14:15:52 <claudiub> I suppose that behaviour will also have to be implemented at the driver level, right? 14:15:53 <pkoniszewski> is in the priority tracking etherpad 14:16:08 <claudiub> ok, gotcha, ty. :) 14:16:10 <PaulMurray> claudiub, if you go to the bottom of https://etherpad.openstack.org/p/mitaka-live-migration you can find links to all the specs 14:16:11 <pkoniszewski> might be missing on wiki - https://review.openstack.org/#/c/245921/ 14:16:28 * mdbooth has been working on libvirt storage pools. I'm only just back at work today. I posted some WIP patches before I left for the hols in December. It's early work, but I know where I'm going with it now. 14:16:58 <PaulMurray> claudiub, force live migration is also known as "pause vm during live migration" 14:17:18 <paul-carlton1> mdbooth, can you add me a reviewer or ping me the urls pls 14:17:29 <mdbooth> paul-carlton1: Sure thing 14:17:36 <claudiub> PaulMurray: I see, cool. Will take a look. 14:18:25 <PaulMurray> mdbooth, paul-carlton1 did we find a way around ploop with the parallels guys 14:18:40 * PaulMurray says "we" meaning "you" 14:18:43 <mdbooth> PaulMurray: No, but it's not that important imho 14:18:50 <paul-carlton1> PaulMurray, sorry I've made no real progress on this, working on internal issue right now, expect to be able to focus on upstream next week 14:18:54 <mdbooth> That's medium term work 14:19:26 <mdbooth> paul-carlton1: https://review.openstack.org/#/c/259148/ 14:19:43 <paul-carlton1> parallels are going to fix libvirt so it supports ploop for storage pools 14:19:48 * mdbooth boggles at +1s from 3 CIs on that patch :) 14:20:36 <paul-carlton1> until they do users of ploop can't move to release of nova that requires storage pools 14:21:07 <mdbooth> paul-carlton1: Nah, it just means that some storage types won't use libvirt storage pools 14:21:37 <mdbooth> We can treat different storage types differently in this regard quite easily 14:22:14 <paul-carlton1> That is another way of doing it, I was hoping we could still make storage pools the default but if we can support different models easily then no issue 14:22:39 <mdbooth> It will be the default for storage types which support it 14:23:44 <PaulMurray> mdbooth, do you expect to make progress in the coming weeks? 14:24:04 <PaulMurray> (I mean something significant before mid cycle) 14:24:15 <mdbooth> PaulMurray: Yes. I've also been distracted by a high priority internal issue, but apart from that I'm working on this. 14:25:08 <PaulMurray> mdbooth, incidentally I will have a couple of HP guys working on libvirt/qemu coming along - so let us know if any help is needed there 14:25:40 <PaulMurray> next item 14:25:52 <PaulMurray> is "split network" - anyone working on that one here? 14:26:20 <mdbooth> PaulMurray: I'm not attending midcycle, btw. However, if there's going to be a day which contains discussion relevant to this I'll make a daytrip down. 14:27:20 <PaulMurray> oh, my mistake - if you are interested I might propose doing that - there will be at least 4-5 libvirt/qemu specialists at the mid cycle 14:27:40 <mdbooth> Definitely interested if it's useful. 14:28:26 <PaulMurray> It looks like no-one has update for split network 14:29:00 <PaulMurray> There is this patch for split network: https://review.openstack.org/#/c/245005/ 14:29:28 <PaulMurray> it only has a current review from pkoniszewski - so more attention would be good 14:29:55 <PaulMurray> next is "Cancel migration" - paul-carlton1 / andrearosa ? 14:29:58 <andrearosa> I am working on a non-priority spec which is high-prioirty as we are close to the non-priority feature freeze :( I hope I can start working on cancel live-migration soon 14:30:36 <andrearosa> maybe by the end of this week, but I am not sure yet. 14:30:39 <PaulMurray> andrearosa, paul-carlton1 is it dependant on the pause migration? 14:31:33 <andrearosa> we have to define a common new GET method and pkoniszewski already put something 14:31:49 <andrearosa> we agree that I'll follow what is there once I start working on it. 14:31:54 <andrearosa> pkoniszewski: am I right> 14:31:56 <andrearosa> ? 14:32:19 <pkoniszewski> andrearosa: yes 14:32:51 <pkoniszewski> andrearosa: there is one GET already under review, part of migration details exposed through API 14:33:11 <andrearosa> pkoniszewski: ack 14:33:20 <pkoniszewski> link -> https://review.openstack.org/#/c/258771/ 14:34:01 <PaulMurray> pkoniszewski, andrearosa we should group the migration details, force migration and cancel together I guess - they make one thread of work 14:34:09 <PaulMurray> I think? 14:34:31 <PaulMurray> does that make sense? 14:34:34 <andrearosa> PaulMurray: what do you mean by group? 14:35:00 <PaulMurray> I mean they are in the same area - looking at and acting on migrations 14:35:51 <andrearosa> yes they are 14:36:26 <andrearosa> we just need to stay in sync 14:36:41 <PaulMurray> Is there any items in approaved specs that I have not covered so far? 14:37:27 <pkoniszewski> there is also alex_xu effort to remove block migration flag from the API 14:37:55 <PaulMurray> pkoniszewski, ah yes, we will come back to that in a minute 14:38:27 <PaulMurray> just to skip over the next couple of things 14:38:30 <PaulMurray> #topic bugs 14:38:57 <PaulMurray> I am not sure if mikal has made any progress organising a bug day 14:39:07 <PaulMurray> I am sure we will get back to that soon 14:39:27 <PaulMurray> I'll try to catch up with him about it 14:39:58 <PaulMurray> #topic Stuck Reviews 14:40:17 <ankit> We have addded few possible solutions to fix this issue on LP bug https://bugs.launchpad.net/nova/+bug/1470420/comments/6 14:40:19 <openstack> Launchpad bug 1470420 in OpenStack Compute (nova) "Set migration status to 'error' instead of 'failed' during live-migration" [Low,In progress] - Assigned to Rajesh Tailor (rajesh-tailor) 14:40:20 <ankit> It would be really helpful if we can get some consensus on finalizing the alternet approach to fix this issue. 14:41:11 <ankit> or if there any consenses on any other approach not mentioned on LP bug note 14:41:52 <PaulMurray> ankit, have you been able to get in touch with ndipanov about it? 14:42:59 <PaulMurray> I am also behind on the ML - only just got back to work 14:43:10 <PaulMurray> was there any more discussion there since the last meeting ? 14:43:20 <ankit> PaulMurray: Nikola has added his comments on the patch itself but we are still not finalized on which approach we should go with 14:45:03 <PaulMurray> so I see a -1 from ndipanov and alaski seems to agree we should not change the user facing messages (status I guess) 14:45:25 <ankit> PaulMurray: Yes 14:46:02 <PaulMurray> I don't see any comments from cores supporting the status change 14:46:41 <PaulMurray> but i also don't see any specific suggestion of what to do except some other cleanup 14:47:12 <ankit> PaulMuarry: Yeah, thats the point we want to confirm from cores 14:47:33 <PaulMurray> Those people are not in this meeting 14:47:45 <ankit> before implementing the next approach 14:48:28 <PaulMurray> we should try and get hold of them - I think they will be around this afternoon so lets see if we can get hold of them in #openstack-nova after this meeting 14:48:33 <ankit> PaulMurray: So I think we should add this as stuck reviews in Nova meeting 14:48:56 <PaulMurray> we can do that too - but lets see if we can have a chat with them first 14:48:56 <ankit> Please suggest 14:49:13 <PaulMurray> as soon as we are done here 14:49:33 <PaulMurray> #topic open discussion 14:49:43 <PaulMurray> The mid cycle is coming up in a few weeks time 14:49:57 <PaulMurray> we should start thinking about what we need to get out of it for this sub team 14:50:14 <PaulMurray> we will be expected to contribute to the agenda as one of the priorities 14:50:30 <PaulMurray> it would also be good to show progress with the "nearly done" reviews 14:50:48 <PaulMurray> so please spend some time on reviews 14:51:06 <PaulMurray> Next is Deprecating the live_migration_flag and block_migration_flag config options 14:51:15 <markmc> hey 14:51:23 <PaulMurray> I haven't read the thread - can you give a summary 14:51:24 <markmc> just wanted to mention some patches I posted last night 14:51:34 <markmc> #link http://lists.openstack.org/pipermail/openstack-dev/2016-January/083275.html 14:51:55 <markmc> the idea is that live_migration_flag is too fine grained a config option 14:52:11 <markmc> allowing operators to do things like remove the PEER2PEER flag just makes no sense 14:52:28 <markmc> it's almost like having a config option with a chunk of arbitrary python code 14:52:45 <markmc> instead, we should have specific config options for behavior changes that have use cases 14:52:57 <markmc> and allow those config options influence the migration flags 14:53:04 <markmc> that's about it 14:53:18 <markmc> just raising here for feedback on the thread 14:53:30 <PaulMurray> markmc, thanks for bringing it up 14:54:28 <PaulMurray> Anyone want to raise anything else? 14:55:44 <PaulMurray> ok - lets end - thank you all for coming 14:55:55 <andrearosa> thanks 14:56:04 <pkoniszewski> thanks 14:56:17 <PaulMurray> ankit, lets go on #openstack-nova and see if we can find ndipanov and alaski 14:56:22 <PaulMurray> #endmeeting