21:00:07 <alaski> #startmeeting nova_cells
21:00:08 <openstack> Meeting started Wed Jun 10 21:00:07 2015 UTC and is due to finish in 60 minutes.  The chair is alaski. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:09 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:11 <openstack> The meeting name has been set to 'nova_cells'
21:00:29 <alaski> o/
21:00:31 <bauzas> \o
21:00:33 <melwitt> o/
21:00:38 <belmoreira> o/
21:00:43 <dansmith> o/
21:00:59 <alaski> #topic Tempest testing
21:01:15 <alaski> melwitt: the floor is yours
21:01:38 <melwitt> nothing new this week, logstash link for the job is http://goo.gl/b7R8wq
21:02:04 <melwitt> looks to be in the same expected state, the patch for the undelete is still up at https://review.openstack.org/176518
21:02:43 <melwitt> the other race condition patch https://review.openstack.org/188126 merged last night so that's cool
21:02:43 <alaski> there was some discussion at the end of that one
21:02:49 <alaski> oh, nice
21:03:03 <alaski> are you good on https://review.openstack.org/176518, or still unsure about it
21:03:32 <bauzas> well
21:04:19 <bauzas> I'm +1 because the trade-off would look like more benefits than debt
21:04:42 <melwitt> I guess overall, we can't get the same behavior it had, I think. it intended to create an instance row if there wasn't one found (because read_deleted="yes")
21:05:05 <melwitt> it intended to ignore InstanceInfoCacheNotFound. now we have instance.save() doing a lot of things inside it, which can raise NotFound
21:05:28 <bauzas> melwitt: yup that's my point
21:05:52 <bauzas> melwitt: it could still be raising a NotFound, but I think it's okay
21:06:14 <alaski> is a create going to undelete something it shouldn't?
21:06:23 <bauzas> because there is a big except exc.NotFound:
21:06:28 <alaski> if not then it seems fine as is
21:08:03 <melwitt> I think my concern is around, if we get NotFound for something other than InstanceNotFound, we would instance.create(). I don't know if maybe it should be like catch InstanceNotFound, create if that, else if it's other NotFound, pass
21:08:52 <melwitt> I don't know if that can really happen, I haven't seen it
21:09:04 <bauzas> mmm
21:09:12 <alaski> I like that approach
21:09:57 <melwitt> okay, I can do that.
21:10:21 <alaski> cool, I just backed off to a +1 for now
21:10:30 <bauzas> fair
21:11:26 <melwitt> I'll update it in a few. that's all I have on the testing, once this patch merges I'll be watching for a drop in job failures, and if that continues X amount of time we'll make it vote?
21:11:26 <alaski> I noticed on the logstash link that failures are down to about 5 a day or less for the last 4 days
21:11:37 <alaski> melwitt: that's the plan
21:11:44 <melwitt> cool
21:12:23 <alaski> thanks for the update
21:12:33 <alaski> #topic Specs
21:12:42 <alaski> https://review.openstack.org/#/c/190209/
21:13:01 <alaski> I wrote up a devref change walking through how some things will work with cells v2
21:13:19 <alaski> and included relevant specs for changes needed, and pointed out where specs are missing
21:13:32 <bauzas> that's a DNM ?
21:13:54 <alaski> it's subject to change based on spec feedback
21:13:55 <bauzas> alaski: are you still -W it ?
21:14:01 <bauzas> alaski: oh ok
21:14:22 <alaski> if it seems useful I'm not against it merging, but that wasn't my intention
21:15:06 <alaski> it was more an exercise for me to see the gaps in specs, and to point people to when they ask why a spec is proposing a thing
21:15:35 <bauzas> fair point
21:16:08 <melwitt> cool, good to have
21:16:47 <alaski> and it's a good medium for feedback which is why I chose it over an etherpad
21:17:37 <alaski> belmoreira: any updates on the specs you have been working on?
21:18:01 <belmoreira> alaski: not yet in that spec
21:18:25 <belmoreira> today we proposed: https://review.openstack.org/#/c/190147/
21:18:48 <belmoreira> I added it to the meeting agenda
21:19:08 <alaski> ahh, needed to refresh
21:19:51 <alaski> I added it to the priorities etherpad for visibility
21:20:10 <bauzas> ah
21:20:16 <bauzas> flavors, mmm ? :)
21:20:22 <belmoreira> great
21:21:07 <alaski> now that I have a list of gaps I'm going to go crazy proposing specs soon
21:21:15 <alaski> heads up
21:21:21 <alaski> anything else on specs?
21:22:00 <alaski> #topic Open discussion
21:22:19 <alaski> anyone have a topic for open discussion?
21:22:40 * bauzas was off for the last 2 days so pass
21:23:18 <alaski> the only spec approved right now is the host_mapping one so that limits the amount of coding for the moment
21:23:32 <alaski> there's also the persist request spec one I guess, but that's blocked atm I think
21:24:14 <alaski> so that just leaves specs to write, and docs
21:24:29 <alaski> that's it from me
21:24:32 <dansmith> I did some review on one today
21:24:35 <dansmith> the flavors one
21:24:53 <alaski> nice
21:24:53 <dansmith> it punted on migration, but it seems like we might as well include that in the spec since it's critical
21:25:30 <alaski> I did set a precedent of splitting the table creation and migration
21:25:48 <belmoreira> dansmith: yes, it's a fair point. The migration plan should be added
21:25:51 <alaski> but now that the database is in place that may not be the right path
21:26:47 <dansmith> well
21:27:28 <dansmith> in this case I think it's important because we'll need to migrate them before we can switch over,
21:27:35 <dansmith> where the instance mapping is kindof a cache at this point
21:27:56 <dansmith> just seems like this one should be "move flavors from here to there"
21:28:20 <alaski> yeah, I agree with that
21:28:31 <alaski> there was a lot more about how to create the database in the first specs adding tables
21:28:43 <alaski> now that it's there, the interesting bit is moving data
21:28:52 <bauzas> agreed
21:28:55 <belmoreira> yes, I agree
21:29:33 <dansmith> oh, also
21:29:39 <alaski> cool.  I'll also take a look at the spec, but probably tomorrow
21:29:42 <dansmith> I started on the patch to make devstack (and then grenade) create the api db
21:29:51 <dansmith> it was failing on something, I haven't looked why,
21:29:56 <dansmith> but at least that is started
21:30:18 <dansmith> https://review.openstack.org/190289
21:30:32 <dansmith> gotta have that in devstack and then a change to grenade
21:31:09 <alaski> very nice
21:31:24 <bauzas> cool indeed
21:31:25 <dansmith> oh, heh I just failed bashate
21:31:28 <dansmith> I think I can handle that :D
21:32:49 <alaski> it's great to see more reviews up
21:32:55 <alaski> anything else from anyone?
21:33:32 <alaski> #endmeeting