14:00:46 <tdurakov> #startmeeting Nova Live Migration
14:00:47 <openstack> Meeting started Tue Dec 20 14:00:46 2016 UTC and is due to finish in 60 minutes.  The chair is tdurakov. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:48 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:51 <openstack> The meeting name has been set to 'nova_live_migration'
14:00:53 <tdurakov> hi folks
14:01:09 <tdurakov> pkoniszewski: ? :)
14:01:22 <pkoniszewski> o/
14:01:33 <tdurakov> so let's start
14:01:44 <tdurakov> #topic CI
14:01:52 <tdurakov> pkoniszewski: great research on grenade
14:02:10 <tdurakov> haven't checked potential fix, you've submitter
14:02:15 <tdurakov> s/submitted
14:02:22 <tdurakov> do you need a help with that?
14:02:24 <pkoniszewski> it does not work, no idea how to fix it yet
14:02:38 <pkoniszewski> any help will be appreciated
14:02:42 <pkoniszewski> I asked Sean for help as well
14:02:59 * tdurakov reviewing claim patch right now
14:03:07 <tdurakov> will take a look after that
14:03:22 <tdurakov> it looks like there are only 2 people
14:03:56 <pkoniszewski> everyone is prepering for christmas :D
14:04:01 <tdurakov> right
14:04:13 <tdurakov> pkoniszewski: got another question on that patch
14:04:14 <tdurakov> https://review.openstack.org/#/c/244489/50/nova/compute/rpcapi.py@453
14:04:22 <tdurakov> why do we need that?
14:05:17 <tdurakov> my understanding is that migration.dest_compute will be set there in any case
14:05:48 <tdurakov> also, I think it's safe to https://review.openstack.org/#/c/244489/50/nova/compute/rpcapi.py@440 not default these to None
14:06:23 <tdurakov> we upgrade controllers node first, by the moment someone decided to start l-m, all controllers will be up to date
14:06:30 <tdurakov> pkoniszewski: thoughts^
14:06:38 <pkoniszewski> that's right, but compute might not be
14:07:08 <tdurakov> do we care about computes?
14:07:35 <pkoniszewski> we have to, as compute might be old one, therefore we need to drop migration object, but we don't want to lose it so we save it into the database
14:07:56 <tdurakov> yes
14:08:00 <tdurakov> but
14:08:26 <tdurakov> migration object will contain dest_host by that moment, and that info will be saved already, no?
14:09:13 <pkoniszewski> it isn't in the DB yet
14:10:06 <pkoniszewski> it's saved for the first time in live_migration method in compute manager, I think
14:11:38 <tdurakov> pkoniszewski: I wonder, could you move that save to task/live_migrate.py/find_destination
14:12:00 <pkoniszewski> tdurakov: my mistake, it is firstly created in conductor manager, but the destination node property might bo None there
14:12:05 <pkoniszewski> might be *
14:12:13 * tdurakov worrying about logic in rpcapi.py
14:12:31 <tdurakov> it's kind of nit, will leave a comment
14:12:55 <tdurakov> https://review.openstack.org/#/c/244489/50/nova/compute/rpcapi.py@440 - what about this one?
14:13:07 <tdurakov> do you really need None for them?
14:13:08 <pkoniszewski> but in find_destination we don't know whether we can pass it through RPC or not
14:13:23 <tdurakov> pkoniszewski: do we need to care about that?
14:13:43 <tdurakov> even if we can, what will happen if we save dest_node info?
14:14:13 <pkoniszewski> we will do some redundant DB calls
14:14:35 <tdurakov> :) possible
14:15:12 <pkoniszewski> not sure whether we need to guard against None, we would need to ask Dan
14:15:27 <pkoniszewski> i'd say that it's better for compatibility to have default to None
14:15:36 <pkoniszewski> but can't see any normal reason
14:15:49 <tdurakov> afair None is required when we remove smth during releases
14:16:12 <tdurakov> but in case of adding a param, all the code is executed within a single node
14:16:19 <pkoniszewski> that's right
14:16:36 <tdurakov> I mean rpcapi.py and conductors code is running inplace
14:17:10 <tdurakov> and there are no calls from compute node to check_destination it's only done from conductor
14:17:27 <tdurakov> anyway, just add dansmith
14:17:53 <tdurakov> do you have anything else to discuss?
14:18:04 <pkoniszewski> not at the moment
14:18:12 <pkoniszewski> this claim patch is the most important to me
14:18:40 <tdurakov> ok, so I'll finish reviewing your claim patch first, and then will try to help with grenade
14:19:05 <tdurakov> there will be no l-m meeting next week
14:19:12 <tdurakov> so, have a great weekend
14:19:17 <tdurakov> #endmeeting