17:00:29 #startmeeting nova_cells 17:00:30 Meeting started Wed Jun 3 17:00:29 2015 UTC and is due to finish in 60 minutes. The chair is alaski. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:32 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:00:34 The meeting name has been set to 'nova_cells' 17:00:46 Anyone around for the cells meeting? 17:00:49 \o 17:00:51 o/ 17:00:54 o/ 17:00:54 o/ 17:00:55 o/ 17:01:13 cool 17:01:20 #topic Tempest testing 17:01:33 I'm going to put melwitt on the spot and defer to her for updates here 17:01:40 :) 17:02:05 okay, the cells job seems in the same state as it has been, link here http://goo.gl/b7R8wq 17:02:50 so it's good there's nothing new happening. I went through the failures from yesterday and they all have to do with tests that delete a server, and I have identified a race where instances can be "undeleted" causing some tests to fail or time out 17:03:20 I have this patch up for the fix https://review.openstack.org/#/c/176518/ that fixes it sort of indirectly, details in the bug it closes 17:04:11 nice, I will review that again in a bit 17:04:11 melwitt: seeing a -W, can I review it ? 17:04:20 I will remove the -W today, I was running a test all night that runs one of the failure-prone tests in a loop to make sure it doesn't make any races with it worse 17:04:27 melwitt: ok 17:04:54 I finally got it to where it could run all night without failing, so I'll put up a couple of more patches for those races that came out after undelete became impossible 17:05:00 alaski: which target do you try to reach for having the cells job voting ? 17:05:28 the UnexpectedTaskStateError is I think unrelated, but is a trace we need to clean up 17:05:39 alaski: considering another observation period once melwitt's patch lands, I guess ? 17:05:54 melwitt: okay, so there's still a bit more work 17:06:14 bauzas: that's a good question 17:06:18 there is also this bug that I *think* doesn't fail tests (I'm not sure) but is a clean up https://bugs.launchpad.net/nova/+bug/1448302 17:06:18 Launchpad bug 1448302 in OpenStack Compute (nova) "cells: intermittent KeyError when deleting instance metadata" [Low,Confirmed] - Assigned to melanie witt (melwitt) 17:06:38 alaski: I just wonder about any benefits that we could have 17:06:47 bauzas: I would like to see us close to the overall failure rate for non cells, but I don't know that that is right now 17:07:01 alaski: agreed, melwitt's patch is important 17:07:04 it's another race with update/delete server metadata I think but I didn't spend much time on it 17:07:14 alaski: my only wonder is how we can prevent any tunnel effect 17:07:19 melwitt: would it help to have someone else look into that? 17:07:20 alaski: ie. fixing all the cells bugs 17:07:48 alaski: yeah, I think bauzas said he might want to look at it, so feel free 17:07:51 alaski: melwitt: my brain certainly needs some recreation, so I would be pleased to help 17:07:54 bauzas: sorry, tunnel effect? 17:07:57 great 17:08:09 melwitt: 17:08:20 cool 17:09:02 belmoreira: sorry, something probably french 17:09:07 so, I guess what I'm thinking is watch the job after the patches land. that I expect to see very few failures after that. but you never know 17:09:15 belmoreira: the idea of trying to boil the ocean I guess 17:09:27 realistically I think if we fix the list servers negative test failure and the random test failures from unexpected task state errors I think we'll be where we need to be 17:10:16 yeah, okay. I'll have all the patches ready today, now that I'm done testing it 17:10:26 cool 17:10:43 I think that's all I have about the tempest testing 17:10:52 awesome, great work 17:10:56 melwitt: so you don't expect https://bugs.launchpad.net/nova/+bug/1448302 to break tests ? 17:10:56 Launchpad bug 1448302 in OpenStack Compute (nova) "cells: intermittent KeyError when deleting instance metadata" [Low,Confirmed] - Assigned to melanie witt (melwitt) 17:11:01 thanks for the update 17:11:08 ok, let's move on 17:11:21 bauzas: honestly I don't know, I didn't spend enough time to be sure. often it doesn't fail them 17:11:38 #topic Specs 17:12:04 The host mapping spec merged during the summit 17:12:21 and https://review.openstack.org/#/c/141486/ is ready for review 17:12:41 and the followup to that one, though I suspect it will need some work 17:13:19 I have a few more to write, but have still been organizing a few things after the break last week 17:13:43 but there is some work available if anyone is interested, in implementing the host mapping spec 17:14:22 and that's all I have on that, so if no comments we can move on 17:14:34 alaski: we are looking into a spec to have flavor tables in the api DB 17:14:45 belmoreira: ahh, excellenbt 17:14:49 bah, excellent 17:15:45 please ping me when that's up, or add me to the review 17:15:53 alaski: that spec seems pretty trivial to be done, probably we should propose some new contributors for it ? 17:16:01 and add it to the agenda for mention, or I can do that 17:16:11 alaski: thinking about the "mentoring" stuff you know 17:16:26 bauzas: I would love for somebody to pick up that work 17:16:38 alaski: we know we have some quickwins that new contributors could help, and that one seems also a good opportunity 17:16:45 alaski, bauzas: we can do it 17:16:50 belmoreira: \o/ 17:16:58 alaski, bauza, belmoreira one thing to be confirmed about that spec.. 17:17:07 belmoreira: I can help you on the implementation but I don't have the bandwidth for implementing it 17:17:18 we assumed that flavor will reside in cell.. i hope it's okay 17:17:25 belmoreira: so you can hassle me on that one 17:17:47 belmoreira: to be clear, I was talking about cell-host-mapping :) 17:18:07 vineetmenon: what do you mean? I thought we were discussing moving flavor to the api db 17:18:11 so out of the cell 17:18:37 alaski, ooh ya.. sorry. flavors in api 17:19:13 vineetmenon: yes, that's okay. I think there's agreement that that's where they should be 17:19:20 +1 17:19:35 does anyone have the spec number... I can't find it now... 17:19:39 belmoreira: still happy with https://review.openstack.org/#/c/182715/2/specs/liberty/approved/cells-host-mapping.rst,cm ? 17:19:48 https://review.openstack.org/#/c/182715/ 17:20:13 alaski: thanks 17:20:23 eh :) 17:21:15 belmoreira: so, there is an API model, a migration script to write and a NovaObject to provide 17:21:24 belmoreira: to make sure we're on the same page, you're saying that you can help with that spec? 17:21:32 +1 17:21:38 alaski: yes 17:21:42 great, thanks 17:21:47 belmoreira: awesome 17:22:03 anything else on specs? 17:22:33 #topic Open Discussion 17:22:43 anything else to discuss generally? 17:23:24 what is blocking https://review.openstack.org/#/c/136490/ ? 17:24:07 belmoreira: I just need to go through it again and refresh it 17:25:10 my goal is that by the next meeting I will have refreshed all specs, and written more or at least have a list of specs I think need to be proposed 17:26:00 anything else? 17:26:40 thanks everyone! 17:26:45 #endmeeting