17:00:18 <ildikov> #startmeeting cinder-nova-api-changes
17:00:19 <openstack> Meeting started Mon Aug 15 17:00:18 2016 UTC and is due to finish in 60 minutes.  The chair is ildikov. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:20 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:00:22 <openstack> The meeting name has been set to 'cinder_nova_api_changes'
17:00:23 <ildikov> scottda ildikov DuncanT ameade cFouts johnthetubaguy jaypipes takashin alaski e0ne jgriffith tbarron andrearosa hemna erlon mriedem gouthamr ebalduf patrickeast smcginnis diablo_rojo gsilvis
17:00:41 <scottda> hi
17:00:45 * scottda is on-time for once
17:00:59 <ildikov> scottda: hi :)
17:01:16 <jgriffith> ildikov: :)  Thought you had a conflict
17:01:26 <ildikov> I have a meeting in parallel, so if you feel like facilitating this one a bit that would be great :)
17:01:32 <jgriffith> Or are you time slicing your meeting processor
17:01:33 <jgriffith> :)
17:02:09 <jgriffith> Ok... so I'll make this sort of easy if there are no objections.
17:02:21 <patrickeast> objection!
17:02:26 <jgriffith> anybody have any updates or questions they'd like to bring up?
17:02:32 <jgriffith> patrickeast: darn you!
17:02:35 <ildikov> jgriffith: I'm a girl, I should be able to figure this out ;)
17:02:51 <ildikov> jgriffith: any updates on your side?
17:03:05 <ildikov> tempest or Nova stuff?
17:03:21 <jgriffith> ildikov: I have nothing new to report
17:04:25 <ildikov> jgriffith: :(
17:04:40 <ildikov> jgriffith: I saw you +1'd my patch, which depends on yours
17:04:48 <jgriffith> ildikov: turn that frown upside down
17:04:54 <ildikov> I can work on adding a few more tests if you're not working on that
17:04:54 <jgriffith> yes, thanks for fixing that
17:05:18 <ildikov> no probs, I was playing with it earlier
17:05:28 <jgriffith> ildikov: so honestly once we decided this was O work it's certainly not on the front burner for me
17:06:23 <ildikov> jgriffith: I'm not sure I'm following you now
17:06:30 <hemna> mep
17:06:41 * mriedem lurks
17:07:01 <jgriffith> ildikov: the attach/detach changes
17:07:11 <hemna> I haven't had time to test it :(
17:07:17 <jgriffith> ildikov: given that they're not going to go in for N
17:07:20 <hemna> too many things coming at the same time right now.
17:07:33 <jgriffith> hemna: that's how it goes... no worries
17:07:45 <mriedem> yup my time is constrained this week, i'm on vacation next week and then it's n-3 after that
17:07:46 <johnthetubaguy> it would be great to get a spec to write up our plans pre summit
17:07:46 <ildikov> jgriffith: I thought for Cinder we can still consider it, but then I missed a few bits
17:07:46 <jgriffith> I will get the latest update pushed to the review this week
17:07:55 <ildikov> jgriffith: and then get things done in Nova as well
17:08:17 <ildikov> jgriffith: as it's a new API in Cinder with microversions it should not break anything until no one is using it
17:08:20 <jgriffith> ildikov: so my point was that things for the current release are priority
17:08:49 <ildikov> jgriffith: if there are other things that are urgent and no bandwidth for this that's another topic
17:08:55 <jgriffith> ildikov: we already all agreed (I thought) that this work was not going to go in during the N release.  Unless I horribly misunderstood something?
17:09:08 <johnthetubaguy> so I have an aside here
17:09:10 <ildikov> jgriffith: we might do a task list and let people pick up things if they have time
17:09:19 <jgriffith> johnthetubaguy: go for it
17:09:26 <johnthetubaguy> turns out we having a very similar conversation about neutron ports and neutron attachments
17:09:33 <johnthetubaguy> mentioned that to jgriffith earlier
17:09:49 <johnthetubaguy> there are plans to get live-migrate into a state machine during Ocata
17:10:05 <johnthetubaguy> it would be great if we end up with similar patterns in the end
17:10:32 <scottda> johnthetubaguy: Is there a spec for the live-migrate state machine?
17:10:40 <johnthetubaguy> there are slight differences around the shared cinder connection, and neutron needing to know the active port, but its very similar
17:11:10 <johnthetubaguy> scottda: the full one I think is WIP, I have notes on the neutron thing here: https://review.openstack.org/#/c/353982/
17:11:53 <johnthetubaguy> scottda: its a little bit too rough and ready to be useful yet, but just trying to capture stuff before the neutron midycle later in the week
17:11:59 <scottda> Cool, thanks. #link https://review.openstack.org/#/c/353982/
17:12:33 <scottda> johnthetubaguy: sure. I'll keep an eye in it..
17:12:37 <johnthetubaguy> basically, I think port bindings and volume attachments are quite similar, and during live-migrate we have two of both during the live-migration
17:12:59 <johnthetubaguy> right now we only have one during the live-migration, evening though we kinda have two connections during live-migration
17:13:16 <johnthetubaguy> but anyways, hopefully we converge on something similar on both sides
17:13:22 <scottda> +1000 to doing things the same (as much as possible) for nova and cinder
17:13:36 <scottda> s/nova/neutron
17:13:39 <johnthetubaguy> there is obviously the shared problem around token expiry, etc
17:14:08 <johnthetubaguy> anyways, I am trying to keep an eye on both conversations to see what is the same / different
17:15:09 <johnthetubaguy> I think neutron has a little more interaction around what is active, and cinder has a little more state to share around the shared volume target stuff
17:15:09 <jgriffith> johnthetubaguy: I'll study the spec you linked and see if we can combine and get a cinder or a shared version written up
17:15:59 <johnthetubaguy> so that current spec *should* be close to a shared vision, its just missing some words for it to make sense on both sides
17:16:20 <johnthetubaguy> its probably worth waiting till next week, I am hopeful to have more details then
17:16:42 <ildikov> johnthetubaguy: thanks for sharing
17:17:00 <ildikov> johnthetubaguy: will this be finalised on the Neutron mid-cycle?
17:17:01 <johnthetubaguy> no problem
17:17:23 <johnthetubaguy> I suspect not, I am aiming to have this at the summit, ideally something cross project with us all together
17:17:44 <johnthetubaguy> I am really just hoping to get vague agreement at the midcycle
17:17:58 <johnthetubaguy> like to the level we have between Nova and Cinder right now
17:19:08 <ildikov> ah, ok
17:19:38 <ildikov> although if there's a rough agreement that should help with figuring out which direction to look at with the Cinder items
17:19:47 <johnthetubaguy> +1 hopefully
17:20:33 <ildikov> from Cinder perspective I would guess the question is more how to use the new calls jgriffith is proposing as opposed to change and adapt those, right?
17:21:24 <scottda> I think Cinder might need additional calls for special cases like live-migrate...
17:21:25 <jgriffith> ildikov: my take away was to modify them most likely to have a similar pattern between the two
17:21:35 <johnthetubaguy> I think it will end up being about using a similar flow, like Cinder owning attachment state, and Neutron owning binding state
17:21:40 <jgriffith> scottda: I've already got that FWIW
17:21:43 <johnthetubaguy> scottda: neutron needs more calls right now
17:21:45 <jgriffith> scottda: anyway
17:21:58 <scottda> jgriffith: Cool. I figured you'd already thought about it...
17:22:02 <johnthetubaguy> jgriffith: do you have this written up now?
17:22:09 <jgriffith> johnthetubaguy: of course not :)
17:22:17 <scottda> What, no spec?
17:22:26 * scottda ducks
17:22:29 <jgriffith> johnthetubaguy: the past few weeks have been..... umm... anyway
17:22:59 <jgriffith> johnthetubaguy: but like I said, I'll try and get that knocked out on planes the next week or so
17:23:05 <johnthetubaguy> jgriffith: yeah, gotcha, writing that up would be cool, so we can compare whats in our heads
17:23:06 <ildikov> jgriffith: your calls are pretty basic and in line with what johnthetubaguy is saying
17:23:10 <johnthetubaguy> jgriffith: sounds good
17:23:31 <jgriffith> ildikov: first look at the spec he referenced yes I think we're aligned
17:23:56 <johnthetubaguy> neutron doesn't have a reserve thing yet, but I am asking for one, etc
17:23:56 <ildikov> jgriffith: I would be sad if we would need to complicate your proposal
17:24:18 <ildikov> jgriffith: having some additional calls, like that safe_to_remove, etc are fine of course
17:24:29 <jgriffith> ildikov: I think we'll be alright
17:24:41 <johnthetubaguy> yeah, if that happens we are doing it wrong
17:24:44 <jgriffith> ildikov: I do need to write up a proper spec and get back to finishing the code though
17:25:02 <ildikov> jgriffith: and a spec would be nice to list work items, so if we can get some extra hands we can point them to the spec
17:25:33 <ildikov> johnthetubaguy: +1
17:26:19 <johnthetubaguy> I hope we can get some OSIC help with this, is enterprises really want stable live-migrate
17:26:29 <johnthetubaguy> s/is/as/
17:27:25 <ildikov> yeap, that's understandable
17:27:55 <johnthetubaguy> anyways, this is future stuff, did we want to cover the newton issues? did they get bashed out now?
17:28:05 <ildikov> and we know it's "hacky" at a few places, so fixing it would make our lives easier as well
17:28:42 <mriedem> johnthetubaguy: the lock thing is being discussed in -cinder,
17:28:52 <mriedem> i haven't been following along, but imagine there will be a summary in the ML
17:29:09 <johnthetubaguy> mriedem: +1 on that
17:29:23 <ildikov> johnthetubaguy: the VM actions part can be figured out when we have a good attach/detach flow
17:30:14 <ildikov> johnthetubaguy: maybe we can try to at least list them in jgriffith's spec, but it might be too much writing for no good (enough) reason
17:30:25 <johnthetubaguy> ildikov: so my (personal) goal post summit is to have an answer for attach, detach and live-migrate, the others "should be easy".
17:30:31 <ildikov> I don't think a detailed list dough, just point out where we need eyes and then hands to fix the flow
17:31:21 <ildikov> johnthetubaguy: that's my thinking as well, attach/detach is getting close-ish to a PoC state just to see it working
17:33:05 <johnthetubaguy> anyways, feels like we are done for now, maybe?
17:34:44 <scottda> done
17:35:48 <ildikov> johnthetubaguy: if you could check this chain: https://review.openstack.org/#/c/335358/ ?
17:35:48 <ildikov> :)
17:36:21 <ildikov> johnthetubaguy: as it has test changes as well I would rather not keep it rebasing much as it can be painful
17:37:08 <johnthetubaguy> yeah, sorry, that is on my list, just not got to it yet
17:40:20 <ildikov> johnthetubaguy: ok, tnx
17:40:43 <ildikov> johnthetubaguy: the test patches already got review comments, so I hope they're fine
17:41:08 <ildikov> johnthetubaguy: the remove check_attach part can be tricky, I'm not sure hemna had some time to look at it
17:41:20 <hemna> I haven't yet :(
17:41:33 <ildikov> johnthetubaguy: mriedem added a few comments, I fixed those
17:42:01 <ildikov> hemna: no worries, I got the tests green now finally, I take it as a good sign :)
17:42:33 <mriedem> at this point in the cycle i'm not sure how much i want to land that change
17:42:38 <ildikov> hemna: although if you can take a look that would be great also, just to have more eyes on the new reserve_volume call
17:43:16 <ildikov> mriedem: well, it's still a bug fix
17:43:38 <hemna> ildikov, ok cool I'll try to take a look
17:43:42 <mriedem> it's a pretty complicated bug fix for a long standing latent issue though
17:43:49 <hemna> I'm stacked 3 deep at the moment..
17:44:02 <mriedem> and we've got 3 weeks before feature freeze and people tied up in reviewing priority blueprints still
17:44:20 <mriedem> i think i'm just trying to say, this is very low priority for me right now
17:45:42 <johnthetubaguy> mriedem: risk / benefit wise, I think you are right
17:45:48 <ildikov> mriedem: ok, I got that part, will try to annoy others as well then and we will see
17:48:02 <ildikov> johnthetubaguy: mriedem: somehow it would be great to sync efforts
17:49:01 <ildikov> johnthetubaguy: mriedem: but for now, could this land in the RC period as well or we're talking about it gets delayed until Ocata if does not land now?
17:49:22 <mriedem> it would certainly not land for RC
17:49:30 <mriedem> RC is like, stop ship bugs / regressions / upgrade issues
17:49:37 <mriedem> but RC is october
17:49:45 <mriedem> i think you mean sept
17:50:13 <mriedem> idk honestly right now, i haven't put the time into reviewing it in real depth
17:50:20 <mriedem> we have a lot of other stuff going on right now
17:50:24 <mriedem> trying to close out before FF
17:51:44 <ildikov> I meant the bug fixing period before the release
17:53:17 <ildikov> it would be great to land the mox --> mock test changes at least as that's the bigger chunk and should not make any harm, the remove check_attach patch tests are stable with those changes
17:53:32 <ildikov> but I'll see if I can get others to check that part
17:54:59 <mriedem> yeah that's maybe doable
17:55:23 <ildikov> ok, thanks
17:55:33 <ildikov> I don't have anything else for today
17:55:41 <ildikov> anything else form anyone?
17:56:44 <ildikov> oh, one other thing
17:56:50 <ildikov> I'm out next Monday
17:57:09 <ildikov> would there be anyone to run the meeting or let's skip?
17:58:59 <ildikov> jgriffith: scottda hemna mriedem johnthetubaguy ^^
17:59:11 <mriedem> skip
17:59:20 <mriedem> i won't be there either way
18:00:56 <ildikov> mriedem: ok, let's skip then, I'll send out a mail about it
18:01:09 <ildikov> thank you all!
18:01:09 <ildikov> #endmeeting