22:00:14 <alaski> #startmeeting nova_cells
22:00:15 <openstack> Meeting started Wed Mar 18 22:00:14 2015 UTC and is due to finish in 60 minutes.  The chair is alaski. Information about MeetBot at http://wiki.debian.org/MeetBot.
22:00:16 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
22:00:18 <openstack> The meeting name has been set to 'nova_cells'
22:00:25 <melwitt> o/
22:00:51 <alaski> awesome, someone showed up :)
22:01:03 <melwitt> hehe. bauzas might be around too
22:01:18 <bauzas> \o
22:01:30 <alaski> nice
22:01:31 <bauzas> there are 24 hours on a day
22:01:54 <bauzas> why sleeping a third of it ? :)
22:01:56 <alaski> I'm noticing that nobody who would benefit from this timeslot is here though
22:02:30 <melwitt> yeah, I recall you mentioned maybe moving these earlier an hour which I'd be cool with
22:02:46 <bauzas> I just checked about Australian times
22:02:52 <melwitt> or maybe doing the morning one every week
22:03:00 <bauzas> and I saw that daylight times are on winter for us
22:03:19 <alaski> melwitt: yeah, I'd like to move to the morning one if possible
22:03:40 <bauzas> so it means that when we're coming to a DST (ie. +1), Australian times are going to normal time (ie. -1)
22:04:05 <alaski> I'll reach out to tonyb since he's expressed interest and see how he feels about that
22:04:52 <alaski> bauzas: I'm  not sure I understand that
22:04:59 <alaski> but we'll sort it out
22:05:09 <alaski> #topic Open Discussion
22:05:53 <alaski> from my side I've been getting some good reviews and code has been merging
22:06:06 <alaski> and that should wrap up today/tomorrow
22:06:22 <alaski> then I can start to focus on L
22:06:38 <alaski> and maybe actually flesh out some meeting agendas
22:06:47 <bauzas> cool
22:07:04 <melwitt> nice
22:07:22 <bauzas> alaski: yeah, FF is tomorrow
22:07:37 <bauzas> alaski: but sounds like we'll have most of the series merged
22:07:40 <melwitt> I've been in the depths of the tempest tests still. it looks like we'll be stuck with around 40 failures for the whitelist at best. slow going as I'm still learning things I didn't know about cells even after this long. I found a couple of small bugs that I'll put patches up for, but they don't put much dent in the failure count
22:08:13 <bauzas> melwitt: do you think we should whitelist the left failing tests ?
22:08:29 <alaski> bauzas: yes, what's up looks on track to be merged
22:08:36 <bauzas> melwitt: because atm, the job is not so useful, albeit for each patch
22:09:00 <melwitt> most of it is just some apis don't work with cells, can't route down. floating ips, resizes mostly don't work at all (flavornotfound in cell), keypair not found, shelve is doing something weird I need to dig into more (creating two snapshots instead of one)
22:09:21 <melwitt> bauzas: yeah, those hypervisor api ones are never going to pass
22:09:54 <bauzas> melwitt: I had no time to look at your problem yet, but do you think we should instead provide new Tempest tests for cells ?
22:10:09 <bauzas> melwitt: and do run the ones which are not good ?
22:10:14 <bauzas> do *not*
22:10:45 <melwitt> bauzas: I thought about that, probably a good idea. if we can create some cells specific tests that call the hypervisors api with cell_name, we would expect it to work
22:11:40 <alaski> worth looking into
22:11:44 <melwitt> for the resize thing, the problem is at the topmost api level, we don't lookup flavor from the top db. so once we're down in the cell we can't get it
22:12:06 <melwitt> unlike servers create, where the flavor is immediately pulled in api/openstack/compute/servers.py
22:12:09 <melwitt> and passed down
22:12:15 <alaski> melwitt: is this for a flavor created by tempest?
22:12:19 <bauzas> melwitt: IIU, the problem is that we're asking for names not having cells details ?
22:12:20 <melwitt> alaski: yep
22:12:36 <melwitt> tempest creates some flavors then proceeds to use those flavors for a lot of stuff
22:12:37 <bauzas> melwitt: ie. a name which is not cell-related?
22:12:54 <melwitt> so, for those we would expect them to pass if they use a canned already there flavor
22:13:04 <alaski> melwitt: right
22:13:05 <melwitt> another candidate for cells-special tests
22:13:42 <alaski> there was a spec at one point to have the flavor api sync to cells, but it sort of faded away since there are some challenges with that
22:14:05 <bauzas> alaski: yeah
22:14:05 <alaski> if it's possible to test that with the standard flavors that would be great
22:14:11 <melwitt> yeah, I tried that for a long while and it was pretty much impossible IMO
22:14:46 <bauzas> melwitt: right, we should instead work on cells V2
22:16:14 <melwitt> yeah. I'll look at the idea of some new tempest tests
22:16:34 <melwitt> for whatever reason I didn't think that's a thing we could do, special cells tests
22:16:56 <alaski> it's not ideal really
22:17:08 <alaski> but I think it depends on how much modification is necessary
22:17:20 <melwitt> yeah, I hope there's an easy way we could prevent them from running for other tempest runs and just cells
22:18:22 <alaski> that would be good, if possible
22:18:50 <alaski> we may want to try to get the job green with blacklisting some tests first, and then work on getting them running
22:19:17 <melwitt> yeah, I'm for that. the only ones failing are the hypervisors ones that are still going to fail after bauzas fixes merge
22:19:19 <alaski> with the new Depends-On flag I think we can test that much more easily
22:19:35 <melwitt> because of the lack of routing info
22:19:54 <alaski> melwitt: oh nice.  so these failures are on tests that aren't currently running?
22:20:14 <melwitt> alaski: no, they are running
22:20:34 <bauzas> alaski: Depends-On flag ?
22:20:37 <alaski> sorry, I meant the non hypervisor ones
22:20:42 <bauzas> alaski: re: Tempest ?
22:21:08 <melwitt> alaski: oh, right. yeah
22:21:21 <alaski> bauzas: I mean that it should allow us to put up a fix and a test at the same time, which we couldn't do before
22:21:39 <melwitt> so these are the things failing after bauzas fixes http://logs.openstack.org/06/160506/7/check/check-tempest-dsvm-cells/8394946/logs/testr_results.html.gz
22:22:12 <alaski> melwitt: awesome, that's not too bad
22:22:14 <melwitt> I guess one is a tenant networks test which probably needs to be blacklisted too
22:22:19 <bauzas> alaski: oh ok
22:22:55 <alaski> melwitt: do you want to put up a patch to blacklist those?
22:23:09 <melwitt> alaski: sure
22:23:21 <alaski> cool
22:23:33 <alaski> if we could get this job green and voting in Kilo that would be a huge win
22:24:02 <alaski> then we can tackle some of the tests, since things like resize would be good to have tested
22:24:37 <melwitt> yeah, agree
22:25:43 <alaski> anything else to discuss?
22:25:50 * alaski promisise to have more of an agenda soon
22:25:57 <bauzas> :)
22:26:42 <alaski> bauzas: soon we'll be back to your favorite topic, scheduling :)
22:26:53 <melwitt> hehe
22:27:18 <bauzas> alaski: eh, that's my first love, agreed :)
22:27:38 <bauzas> alaski: but now, I'm liking to enjoy other stuff like cells ;)
22:28:09 <bauzas> but I'll buy popcorn, for sure
22:28:20 <alaski> cool.  you and melwitt are becoming some of the few people to understand it
22:29:08 <alaski> if nothing else, I'll end here
22:29:10 <melwitt> I'm glad to learn it finally and glad we'll be having a v2
22:29:25 <alaski> I'll try to reach out to a few folks and see if this meeting time is worth keeping
22:29:30 <bauzas> sure
22:29:30 <alaski> melwitt: me too :)
22:29:35 <bauzas> +1
22:29:53 <alaski> thanks you two!
22:29:58 <alaski> #endmeeting