21:00:08 <dansmith> #startmeeting nova_cells
21:00:09 <openstack> Meeting started Wed Jun 14 21:00:08 2017 UTC and is due to finish in 60 minutes.  The chair is dansmith. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:10 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:12 <openstack> The meeting name has been set to 'nova_cells'
21:00:15 <melwitt> o/
21:00:17 <dansmith> who dat
21:00:20 <dtp> HI
21:00:31 <dansmith> dtp: inside voice please
21:00:46 <dtp> _hi_
21:00:52 <dansmith> better
21:00:55 * melwitt casts summon mriedem
21:00:56 <mriedem> o/
21:00:57 <dansmith> no mriedem
21:00:58 <dansmith> heh
21:01:05 <dtp> cool!
21:01:07 <mriedem> saving throw failed
21:01:10 <dansmith> #topic bugs/testing
21:01:13 <melwitt> haha
21:01:29 <dansmith> mriedem: this is your section of the meeting
21:01:41 <mriedem> oh god
21:01:48 <dansmith> I have this fix up: https://review.openstack.org/#/c/473931/
21:01:58 <mriedem> that's the only recent one that comes to mind
21:02:01 <dansmith> which fixes ironic's devstack plugin when we're in multicell
21:02:06 <dansmith> woot, that's progress
21:02:24 <melwitt> I shall review it
21:02:37 <dansmith> I shall be appreciative
21:02:49 <melwitt> so that's not a bug though? there's no bug reference in the commit message
21:02:52 <dansmith> nothing else on fire?
21:03:01 <dansmith> melwitt: it is a bug but we found it pre-merge so nobody filed anything
21:03:08 <melwitt> oh k
21:03:10 <dansmith> melwitt: and it's really just a straggler from the bp
21:03:47 <mriedem> right we aren't going to backport this
21:03:50 <mriedem> so don't really need a bug
21:03:50 <dansmith> no other bugs.. going once ...
21:03:53 <dansmith> yeah
21:04:10 <melwitt> cool
21:04:29 <dansmith> #topic open wounds
21:04:50 <melwitt> quota battle damage
21:04:53 <dansmith> melwitt's stuff is still up,
21:04:56 <dansmith> I +Wd the bottom change recently and the next one has a -1 on it
21:05:44 <melwitt> the next one is updated already, mriedem said he'd +2 it :)
21:05:50 <dansmith> oh cool
21:05:56 <mriedem> i will after +2ing this stats thing
21:06:11 <melwitt> yeah, he's burning through these and it's awesome
21:06:21 <dansmith> melwitt: we call that "having given up"
21:06:32 <melwitt> >:| hrmph!
21:06:43 <mriedem> i -1 each one :)
21:06:50 <mriedem> so i haven't given up yet
21:06:52 <dansmith> well, it's still resulting in you getting patches merged, so ..
21:06:55 <melwitt> yeah, I mean it's great for one's stats
21:06:56 <mriedem> i'll give up when it comes to reviewing placement changes
21:07:05 <dansmith> heh you and me both
21:07:15 <melwitt> nested resource providers anyone? anyone?
21:07:15 <dansmith> anything else up for review?
21:07:33 <dansmith> bueller?
21:07:39 <melwitt> the placement project/user stuff is up and actively reviewed
21:07:45 <dansmith> oh right
21:07:46 <melwitt> but I think we all knew that
21:07:49 <mriedem> i still have this host api one
21:08:03 <mriedem> https://review.openstack.org/#/c/461532/
21:08:04 <melwitt> yeah I need to review the host api ones
21:08:31 <mriedem> https://review.openstack.org/#/c/464280/ is also basically ready for review; the -2 is because i have to get the os-hypervisors API changes done and then we're going to lock-step those i think
21:08:37 <dansmith> I just starrred it
21:08:37 <mriedem> like we did for 2.36 in newton
21:08:57 <mriedem> so we -2 all other microversions until those get in, at some point
21:09:05 <mriedem> the os-hypervisors changes are turning out to be a major pain in my ass
21:09:49 <dansmith> yeah
21:10:09 <dansmith> anything else up?
21:10:10 <mriedem> if it weren't a total api rewrite,
21:10:14 <mriedem> then it woudn't be so bad
21:10:18 <dansmith> how come dtp gets away with not having open reviews?
21:10:26 <mriedem> he's an "operator"
21:10:30 <dansmith> oh
21:10:41 <dtp> i dunno about that..
21:11:30 <dansmith> okay well, anyway... anything else up we should mention here?
21:12:16 <dansmith> okay then
21:12:21 <dansmith> #topic open discussion
21:12:26 <dansmith> discuss, openly
21:12:44 <melwitt> I wanted to openly mention I resurrected the old console objects patches,
21:13:10 <dansmith> I openly hear what you're saying
21:13:29 <melwitt> and they mostly worked, but the noVNC console isn't working as the code expects. so I'm digging into that more. but AFAICT so far, it's not something that's going to end up working (putting server uuid in the access url),
21:14:24 <melwitt> and the effort might have to be redesigned in the worst case. I'm not 100% sure what's going on yet bc I don't know much about noVNC and how our websockets proxy works
21:14:52 <dansmith> okay I remember discussing this but it has all fallen out of my head
21:15:10 <melwitt> the idea was that we could put server uuid in the access url we give out to make things work with cells v2
21:15:19 <dansmith> this is the thing where the websocket proxy needs to know what cell to search for a token in?
21:15:31 <melwitt> and that it would be backward compatible with old proxies, but what I see in tempest is that for noVNC it is not backward compatible
21:15:35 <melwitt> yeah
21:15:48 <mriedem> melwitt: thorst and markus_z might be able to help
21:15:51 <dansmith> and this is because we assume a global auth service thing?
21:15:56 <mriedem> i think thorst's team added the vnc tests to tempest
21:16:02 <mriedem> and markus has done a bunch of console crap
21:16:05 <dansmith> mriedem: yeah
21:16:19 <melwitt> mriedem: cool, thanks. I think I'll soon be needing to ask some questions
21:16:38 <melwitt> the problem is we thought old proxies would ignore the server uuid added to the url
21:16:52 <melwitt> but noVNC isn't ignoring it and it's blowing up when the server uuid is added to the url
21:17:00 <dansmith> we should also consider if there's a reasonable way to shard this on cells too, like make each cell run its own proxy and just make the thing that returns the user point directly at the right one
21:17:04 <melwitt> because of the way it forwards the token as a header cookie
21:18:03 <melwitt> hm, okay. will think on that
21:18:30 <melwitt> it sounds like that could work. would eliminate the need for instance uuid in url
21:18:35 <dansmith> I would think sharding that service per cell would probably be good anyway,
21:18:52 <dansmith> since that could get overwhelmed by tons of people with open consoles
21:18:58 * melwitt nods
21:19:00 <dansmith> but yeah, think on it and feel free to ping me if you want to dive deep on it
21:19:07 <melwitt> cool, thank you
21:19:20 <dansmith> okay what else?
21:19:22 <mriedem> +2 on https://review.openstack.org/#/c/446241/ now so it's on dansmith
21:19:24 <mriedem> oh wait,
21:19:27 <mriedem> author, can't +W
21:19:34 <dansmith> yeah
21:19:42 <mriedem> https://review.openstack.org/#/c/473931/ +2 now it's on melwitt :)
21:19:46 <dansmith> jay'll do it
21:20:13 <melwitt> mriedem: looking at it now
21:20:20 <mriedem> nothing else from me
21:20:24 <dansmith> aight cool
21:20:32 * dansmith waits a few more seconds
21:21:06 <dansmith> #endmeeting