Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-02-16
19:33:34 dansmith the biggest thing would be going back to the api database again, but that shouldn't be affected I think right?
19:33:56 dansmith melwitt: we can still cache them even if we target them early
19:34:31 melwitt yeah. we'd have to find a place for that, maybe nova.context is not the best place is another thing belliott didn't like about it
19:34:55 dansmith yeah, I don't really get those concerns, but..
19:34:56 melwitt dansmith: yeah going back to api database isn't affected. those can use the global api db context managers
19:35:18 dansmith yeah, so, maybe I'm just being overly concerned about doing the target hard and early
19:37:26 melwitt yeah, I don't think we've thought a lot about targeting a context and just letting it carry the db/mq connections around
19:37:42 melwitt it might work fine
19:38:05 dansmith let me just do the early full target thing and see what explodes
19:38:12 melwitt sure
19:43:55 openstackgerrit Artom Lifshitz proposed openstack/nova master: API support for tagged device attachment https://review.openstack.org/376021
19:43:55 openstackgerrit Artom Lifshitz proposed openstack/nova master: Delete disk metadata when detaching volume https://review.openstack.org/435082
19:46:09 mriedem dansmith: on https://review.openstack.org/#/c/433362/4/nova/api/openstack/compute/servers.py i wonder how much of that pattern is going to be everywhere
19:46:14 melwitt dansmith: I'll play around with making my mq connection patch cache early in the target_cell later
19:46:43 dansmith mriedem: lots of places
19:46:46 mriedem we could put some common code into nova/api/openstack/common, but that doesn't help with all of the target_cell stuff
19:47:18 dansmith mriedem: well, it does if it just targets the context permanently, which is what I'm trying now
19:47:41 mriedem why do this in the REST API handlers rather than the compute API class methods themselves?
19:48:21 dansmith mriedem: they have to look up the instance
19:48:38 dansmith mriedem: and we don't want the rest api looking up the mapping, then making compute/api do it again
19:48:44 mriedem oh yeah
19:51:08 mriedem dansmith: before i read scrollback i was thinking we could pass the mapping to every compute API method, like the context, and then we could put a decorator on the compute API methods to do the target cell thing for us
19:51:15 mriedem but i see melwitt doesn't like that
19:51:22 dansmith mriedem: _I_ don't like that
19:51:25 mriedem oh
19:51:30 melwitt we both don't like it
19:51:33 dansmith I asked so someone could say "fuck no"
19:51:43 melwitt heh
19:52:21 mriedem was that "(1:17:41 PM) dansmith: melwitt: yeah, well, glad we agree on that, I'd like to rule that out right here and now :)" ?
19:52:29 dansmith yeah
19:52:36 mriedem because you were also talking about stashing the mapping in the context or the instance at the same time
19:52:50 dansmith yeah, but I meant the param thing
19:53:04 jaypipes mriedem, dansmith, cdent, edleafe, johnthetubaguy, bauzas, melwitt: https://etherpad.openstack.org/p/decision-finding-shared-inventory
19:54:26 jaypipes mriedem, dansmith, cdent, edleafe, johnthetubaguy, bauzas, melwitt: feel free to vote and leave comments... adding that to the scheduler/placement session discussions at PTG
19:54:32 cdent jaypipes: thank you for writing that up. will digest properly tomorrow morning, unless I'm suddenly overcome with desire to think about it now (which is somewhat likely)
19:55:56 mriedem that bile yellow color hurts my eyes
19:56:38 jaypipes mriedem: then lucky for you you didn't see the shirt I was wearing this morning. cdent is still recovering.
19:56:45 cdent jaypipes: for completeness could you add a bit more info on why that particular question needs to be asked?
19:56:58 jaypipes cdent: yuppers.
19:56:59 cdent yeah, I'm going to have nightmares tonight. the combo is...
19:57:01 jaypipes cdent: will do.
19:57:07 mriedem were the nipples cut out?
19:57:23 mriedem because if so, sign me up
19:57:30 jaypipes mriedem: lol
19:57:52 mriedem dansmith: so it sounds like stashing the targeted context is the plan?
19:58:19 dansmith mriedem: potentially, until I find some reason to hate that more
19:58:27 cdent mriedem: great, thanks. I'm totally broken now.
20:00:05 mriedem i'm trying to think of an operatoin where we could have multiple instances in different cells
20:00:10 mriedem but i don't think that's possible right now
20:00:20 dansmith mriedem: it's not now for sure,
20:00:39 dansmith mriedem: and I was thinking of migrations of some sort, but that'll clearly require some special handling anyway
20:00:39 mriedem what apis take multiple instances anyway?
20:00:53 dansmith well, the events api, which is why I said what I did, but that one is already handles
20:00:55 dansmith *handled
20:01:03 mriedem ok
20:01:34 mriedem maybe we'll revive that spec to expose quiesce in the api for multiple instances, which could all be in different cells
20:01:36 mriedem :)
20:01:51 mriedem haha
20:06:17 openstackgerrit Sean Dague proposed openstack/nova master: Verify project_id when quotas are checked https://review.openstack.org/435010
20:11:56 mriedem sdague: your spec update https://review.openstack.org/#/c/435016/ looks good to me
20:12:04 openstackgerrit Yuming Ma proposed openstack/nova-specs master: Nova BFV by reserving volume https://review.openstack.org/417541
20:12:07 mriedem it does clarify the logic
20:13:54 sdague it seems to be working on the happy paths, I'll need to put some other real testing in place to test the rest of the cases
20:17:55 openstackgerrit Yuming Ma proposed openstack/nova-specs master: Nova BFV by reserving volume https://review.openstack.org/417541
20:39:08 openstackgerrit Yuming Ma proposed openstack/nova-specs master: Nova BFV by reserving volume https://review.openstack.org/417541
20:40:21 mriedem anyone aware of a way to basically squash a series of changes into a single patch using git format-patch?
20:40:31 mriedem without first doing a git rebase to squash them into a single commit
20:42:16 mriedem nvm thanks google http://stackoverflow.com/questions/616556/how-do-you-squash-commits-into-one-patch-with-git-format-patch
20:46:33 cdent edleafe: you the light blue on jay's etherpad? I think we're running into similar mental blocks :(
20:46:44 edleafe dat's me
20:51:34 mriedem nova meeting in 9 minutes
20:59:42 openstackgerrit Maciej Szankin proposed openstack/nova master: Tests: remove .testrepository/times.dbm in tox.ini https://review.openstack.org/435111
21:00:14 mriedem it's time
21:35:52 cfriesen mriedem: you had proposed that I send in a stable/newton release note for bug 1644248....what's the process for that given that there won't be an equivalent commit in master or stable/ocata?
21:35:52 openstack bug 1644248 in OpenStack Compute (nova) ocata "Nova incorrectly tracks live migration progress" [High,In progress] https://launchpad.net/bugs/1644248 - Assigned to Matt Riedemann (mriedem)
21:36:44 mriedem cfriesen: just commit a release note to stable/newton with the known issue listed
21:37:39 cfriesen mriedem: okay, can do
21:42:11 openstackgerrit Eric Fried proposed openstack/nova master: PowerVM Driver: spawn/destroy https://review.openstack.org/391288
21:59:10 openstackgerrit melanie witt proposed openstack/nova master: Count resources to check quota for cells https://review.openstack.org/416521
22:00:31 openstackgerrit Takashi NATSUME proposed openstack/python-novaclient master: Microversion 2.42 - Fix tag attribute disappearing https://review.openstack.org/429512
22:02:17 openstackgerrit Takashi NATSUME proposed openstack/python-novaclient master: Microversion 2.43: List/Show all server migration types https://review.openstack.org/430839
22:02:29 openstackgerrit Takashi NATSUME proposed openstack/python-novaclient master: Microversion 2.43 - Enable cold migration with target host https://review.openstack.org/406707
22:02:43 openstackgerrit Dan Smith proposed openstack/nova master: Make servers API use cell-targeted context https://review.openstack.org/433362
22:02:49 dansmith melwitt: surprisingly little drama with unit and functional ^
22:02:56 dansmith we'll see how devstack likes it
22:04:41 melwitt dansmith: yeah, that's almost no change to the unit and functional. huh.
22:05:05 openstackgerrit Eric Fried proposed openstack/nova master: PowerVM Driver: power_on/off and reboot https://review.openstack.org/427380
22:05:19 dansmith well, it kinda makes sense since I'm just making the set call a no-op for the simple cell fixture
22:05:22 dansmith although I'm surprised a bunch of functional tests didn't fail
22:05:26 dansmith but we'll see
22:06:19 melwitt yeah
22:06:20 mriedem bauzas: got it for you https://review.openstack.org/#/c/433855/
22:08:43 bauzas mriedem: heh
22:09:05 bauzas anyway, I promised a comment
22:12:02 openstackgerrit Sujitha proposed openstack/nova master: Change tags to default field in Instance object. https://review.openstack.org/415298
22:12:55 cfriesen dansmith: regarding https://review.openstack.org/#/c/265282/ if we remove the "name" and "id" from the flavor details, do we just display all of the flavor info when doing "nova list --fields flavor"? Currently it shows just the flavor ID.
22:21:44 zzfoo Hi, I'm trying to get live migration working. I have shared store setup and working with NFS though when i do a live migrate I get the following error in the nova-compute logs: error:
22:21:44 zzfoo Unable to acquire lock on `/opt/stack/data/nova/instances/locks/nova-fd2836e4eef36d5483a37a5000a42a72fb95ea7c` due to [Errno 37] No locks available
22:22:09 cfriesen does anyone know why I'm not getting an auto-generated Change-Id when creating a python-novaclient commit? Is this expected or did I mess something up?
22:22:25 zzfoo on the shared storage though that file has been created.

Earlier   Later