16:00:02 <ildikov> #startmeeting cinder-nova-api-changes
16:00:02 <openstack> Meeting started Thu Oct 26 16:00:02 2017 UTC and is due to finish in 60 minutes.  The chair is ildikov. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:03 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:06 <openstack> The meeting name has been set to 'cinder_nova_api_changes'
16:00:08 <ildikov> johnthetubaguy jaypipes e0ne jgriffith hemna mriedem patrickeast smcginnis diablo_rojo xyang1 raj_singh lyarwood jungleboyj stvnoyes
16:00:21 <mriedem> o/
16:00:23 <stvnoyes> o/
16:00:38 <jungleboyj> @!
16:00:39 <_pewp_> jungleboyj (✧∇✧)╯
16:01:02 * smcginnis head explodes
16:01:11 <ildikov> hi All :)
16:01:44 <ildikov> smcginnis: after one minute? :)
16:01:45 <jgriffith> o/
16:01:48 <jungleboyj> No smcginnis  Nooo!!!!!!
16:02:45 <ildikov> ok, let's start
16:03:13 <ildikov> so we had a milestone last week by merging an updated version of the multi-attach spec
16:03:36 <jungleboyj> Yay!
16:03:45 <ildikov> special thanks to mriedem for the last round of cleanups on it!
16:03:55 <ildikov> and all of you for contributing to figuring it out again
16:05:18 <ildikov> we talked about capturing some of the policy bits and pieces and other relevant parts on the Cinder side
16:05:38 <ildikov> I will consult with jgriffith on this
16:05:51 <mriedem> note there were 2 todos in the spec
16:05:59 <mriedem> i've got a test up for one of those
16:06:06 <mriedem> https://review.openstack.org/#/c/515426/
16:06:19 <mriedem> to see what happens if you try to attach an attached volume to the same instance with the new v3 attach flow
16:06:28 <mriedem> regardless of the multiattach flag on the volume
16:06:36 <mriedem> i have a feeling cinder will be cool with that
16:06:40 <mriedem> but not sure
16:07:08 <mriedem> maybe not though - nova will still call os-reserve i think
16:07:19 <mriedem> and looking at the db conditional update code in cinder, that might block it
16:07:21 <mriedem> it should
16:07:36 <ildikov> it won't call that with the new flow
16:07:48 <mriedem> ok that would be a problem then
16:08:08 <mriedem> but, let's see what happens
16:08:46 <mriedem> the other todo in the spec was determining if nova would require a microversion for the change to support multiattach, and i think we would,
16:08:57 <mriedem> and we have an example to follow (2.49 for tagged attached)
16:09:35 <ildikov> I was emotionally prepared for the latter
16:09:45 <ildikov> I guess we should have a spec update too
16:09:51 <mriedem> we can do that later
16:09:54 <mriedem> just fyi
16:10:02 <ildikov> unless someone has objections to the idea
16:10:22 <ildikov> sounds good!
16:10:27 <mriedem> otherwise it looks like next up is https://review.openstack.org/#/c/514853/ and the backend_id and shared_targets stuff on the cinder side
16:10:34 <mriedem> i need to re-review ^ the latest
16:10:51 <ildikov> yep, jgriffith has just uploaded a fixed version
16:11:06 <ildikov> and the new attach patch is in a good shape too
16:11:15 <jgriffith> I'll be rewriting the shared_targets changes now that the online migration stuff is there
16:11:29 <jgriffith> mriedem: dansmith thanks for the help on that BTW
16:12:03 <mriedem> np
16:12:07 <mriedem> one little comment inline
16:12:09 <mriedem> otherwise lgtm
16:12:15 <dansmith> yessir
16:12:55 <jgriffith> mriedem: yeah, not following; but lemme run it and see
16:13:07 <mriedem> jgriffith: you run the migration once with a limit of 2 and you have 3 total
16:13:09 <mriedem> just saying,
16:13:09 <jgriffith> mriedem: assert passed
16:13:19 <jgriffith> right
16:13:20 <mriedem> run it again with count=2 and assert total 1 and updated 1
16:13:23 <mriedem> so run the migratoin twice
16:13:27 <jgriffith> mriedem: total is the total number of entries in the table
16:13:33 <jgriffith> Oh
16:13:34 <jgriffith> got ya
16:13:38 <mriedem> total w/o a uuid set
16:13:44 <jgriffith> yes
16:13:47 <mriedem> cool
16:14:13 <mriedem> and then find an intern to write some fixtures for you guys :)
16:14:39 <jgriffith> :)
16:14:43 <mriedem> if you're doing a contributor thing at the summit, might be something to put in a list of stuff people can work on
16:14:43 <ildikov> lol :)
16:14:44 <mriedem> with examples
16:14:58 <mriedem> cleaning up the low level db api mocks i mean
16:15:35 <ildikov> there are occasions to point people to tasks like this
16:16:04 <ildikov> both the training and the on boarding room, so I will let jungleboyj and smcginnis to give it some extra thoughts :)
16:16:44 <ildikov> ok, so the online migration is on track and I would expect that the shared_targets patch will be an easier bit to fix up
16:16:48 <jungleboyj> ildikov:  Good thought.
16:17:02 <ildikov> but jgriffith can correct me if I'm overly ambitious :)
16:17:04 <jgriffith> ildikov: should be
16:17:18 <ildikov> jgriffith: coolio
16:17:31 <jgriffith> but then again I thought fixing up the UUID one was going to be *simple* so... who knows :)
16:18:03 <ildikov> jgriffith: it's almost weekend here, so give it a bit more positivity :)
16:18:31 <ildikov> johnthetubaguy: mriedem: any chance you can look at the new attach patch?
16:18:33 <jgriffith> That trick never works :)
16:18:44 <mriedem> not right now
16:19:04 <ildikov> jgriffith: you're so cruel to me today :)
16:19:11 <mriedem> i'm waiting on those tempest test results too
16:19:49 <ildikov> ok, let's see those and then if any Nova cores could take a look at that patch that would be great
16:19:59 <ildikov> so we can have some progress in parallel
16:20:43 <ildikov> to the margin, I also put the libvirt patch for multi-attach on top of the new attach patch
16:21:37 <ildikov> I know we're not there yet, however once we are that one should be an easy one to land
16:22:12 <ildikov> it's also small, so a quick view is also appreciated to see whether the concept used there is acceptable or not
16:23:12 <ildikov> that's what I mainly have for today
16:23:16 <stvnoyes> I noticed that tempest doesn't allow for multiple servers being validatable. A floating ip issue. So I was going to update tempest to allow multiple servers to be validatable so we can check multiattachments from inside the vms. It will be useful when we start writing MA tempest tests. Seem reasonable?
16:23:24 <ildikov> I would love to see the new attach patch landed as soon as possible
16:24:04 <mriedem> stvnoyes: probably ask the qa team
16:24:18 <mriedem> bring it up in a weekly meeting maybe, or just ask around in the -qa channel
16:24:21 <mriedem> andreaf or mtreinish
16:24:28 <ildikov> stvnoyes: I like your proactivity :)
16:24:32 <stvnoyes> ok will do
16:24:50 <ildikov> stvnoyes: tnx
16:26:07 <ildikov> ok, anything else from anyone for today?
16:26:27 <mriedem> nope
16:27:04 <mtreinish> stvnoyes: I'm pretty sure you can create 2 servers that you can ssh into in tempest, there are definitely tests for neutron stuff doing that
16:27:15 <mtreinish> we can talk about it later in -qa
16:27:44 <stvnoyes> mtreinish: ok, thanks
16:27:47 <ildikov> mtreinish: sounds good, thanks!
16:28:22 <ildikov> then this is it for today
16:28:50 <ildikov> let's keep in touch on the tempest patch and get the Cinder bits and the new attach patch landed as soon as we can
16:29:09 <ildikov> oh, one more thing
16:29:33 <ildikov> does next week work for the most of us or is it "travel day" already?
16:30:02 * jungleboyj will be over the ocean somewhere.
16:30:19 <mriedem> won't work for me
16:30:28 <mriedem> cancel next week
16:30:36 <ildikov> ok, cancel then
16:30:52 <ildikov> so don't forget about our forum session: https://www.openstack.org/summit/sydney-2017/summit-schedule/events/20457/cindernova-cross-project-session-on-multi-attach?BackURL=https%3A%2F%2Fwww.openstack.org%2Fsummit%2Fsydney-2017%2Fsummit-schedule%2Fglobal-search%3Ft%3Dmulti-attach%23eventid%3D20457
16:31:52 <ildikov> the point is to collect feedback on how people intend to use the thing with regards to the follow up plans with touched on earlier
16:32:44 <ildikov> please make it there if you can so we can answer questions and ensure we cover all the aspects we were discussing earlier
16:33:27 <ildikov> we can sync up on the Forum session on the project channels or the ML in case needed
16:33:28 <mriedem> about that,
16:33:34 <mriedem> who is seeding the agenda?
16:33:47 <mriedem> like, what questions are going to be asked? assuming policy and r/w r/o stuff
16:33:55 <mriedem> maybe some background on the sticky parts in the spec
16:33:57 <mriedem> like boot from volume
16:33:58 <jungleboyj> mriedem:  Good question.
16:34:14 <mriedem> the schedule has jay's face on it so i'm deferring to him
16:34:27 <mriedem> who will likely defer to ildikov
16:34:44 <ildikov> mriedem: those two were the main questions/concerns
16:35:00 <jungleboyj> mriedem:  Ouch ... but you are kind-of right.
16:35:45 <jungleboyj> So, what we talked about last time:
16:35:49 <ildikov> mriedem: beyond that we can share current limitations on libvirt
16:36:00 <jungleboyj> How do people want to use multi-attach?
16:36:11 <jungleboyj> What are the expectations for the functionality?
16:36:17 <mriedem> i'm not sure i'd ask that,
16:36:22 <mriedem> with 40 minutes,
16:36:23 <ildikov> and depending on the audience raise the flag for people on the Cinder back end side
16:36:27 <mriedem> i'd be as specific as possible
16:36:31 <jungleboyj> mriedem:  Ok ...
16:36:35 <mriedem> like, this is what we're going to do in queens
16:36:37 <mriedem> these are the limitations
16:36:51 <mriedem> is that cool with people (yes/no)
16:37:01 <ildikov> I would rather tell what is the bare minimum people can expect and recruit as many people as possible to test it out
16:37:03 <mriedem> then if there is time, get into whatever future stuff you want to find out
16:37:14 <jungleboyj> ildikov:  Ok.  That makes sense.
16:37:30 <mriedem> asking "what do you want?" will be a mes
16:37:32 <mriedem> *mess
16:37:34 <jungleboyj> So, we tell them ...
16:37:35 <ildikov> mriedem: +1, I think we're on that same page
16:37:42 <mriedem> "i want to pass volume type to nova!"
16:38:03 <jungleboyj> We support one r/w volume and other volues are r/o.
16:38:08 <jungleboyj> Sorry, attachments.
16:38:22 <jungleboyj> We are not supporting boot from volume with multi-attach.
16:38:46 <ildikov> we support all of them being r/w and make people aware that they can screw things up if they are not careful enough very easily
16:38:56 <ildikov> and then wish good luck and smile :)
16:39:06 <jungleboyj> ildikov:  Did we agree to that?
16:39:17 <jungleboyj> ildikov:  Or are you messing with me?
16:39:48 <ildikov> jungleboyj: we didn't figure out the R/O part when I last checked, but I might be missing something here
16:40:13 <jungleboyj> jgriffith:  mriedem  ?
16:40:27 <mriedem> jungleboyj: you need to re-read the nova spec
16:40:31 <jgriffith> :)
16:40:38 <jungleboyj> mriedem:  Saw that coming.
16:40:43 * jungleboyj head explodes
16:41:16 <jungleboyj> Ok, so, I am going to go read the Cinder and Nova Specs and put together an etherpad for you guys to approve.
16:42:08 <ildikov> jungleboyj: I can take the burden of MC-ing that session if you want
16:42:29 <jungleboyj> ildikov:  That would be good.  :-)  People will be nicer to you.
16:42:37 <ildikov> jungleboyj: and thanks for the etherpad in advance :)
16:43:11 <ildikov> jungleboyj: haha, not 100% sure about that, but we will see :)
16:43:12 <jungleboyj> ildikov:  Yep, I think that is a fair split of work. Will get me caught up.
16:43:24 <ildikov> jungleboyj: +1, thanks
16:43:55 <ildikov> mriedem: any remaining concerns to the session?
16:44:05 <ildikov> or anyone else?
16:45:10 <mriedem> no
16:45:32 <ildikov> cool
16:45:41 <ildikov> then I think we're now good for today
16:46:09 <ildikov> please review the two Cinder patches and the new attach patch in Nova before the Summit
16:46:30 <ildikov> that would at least make me very happy, I know it doesn't matter that much, but still
16:46:46 <ildikov> and safe travels for next week and see you soon!!!
16:46:58 <ildikov> thanks everyone!
16:47:09 <ildikov> #endmeeting