Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-01-11
20:34:40 edleafe jaypipes: I'm getting it in some of my tests.
20:35:03 openstackgerrit Claudiu Belu proposed openstack/nova: Adds Hyper-V OVS ViF driver https://review.openstack.org/140045
20:35:06 edleafe jaypipes: let me see if there is somewhere else it is getting changed
20:36:47 mriedem EmilienM: well the thing is, placement and cells v2 were optional in newton,
20:37:00 mriedem so we had to explicitly enable those in specific jobs in newton to test them out
20:37:29 mriedem EmilienM: which takes the forethought to know you have to toggle those bits, we're missing that as part of our process right now for non-devstack/grenade things
20:47:17 mriedem diana_clarke: i'll approve the doc update once https://review.openstack.org/#/c/418590/ is merged
20:48:07 diana_clarke mriedem: thanks, I probably should have has a depends on
20:48:13 diana_clarke had
20:48:14 openstackgerrit Eric Fried proposed openstack/nova: Introduce base PowerVM driver https://review.openstack.org/391288
20:49:03 openstackgerrit Matt Riedemann proposed openstack/nova: docs - Connect to placement service & retries https://review.openstack.org/419123
20:49:09 mriedem diana_clarke: done
20:50:23 openstackgerrit Eric Fried proposed openstack/nova: WIP(2): Introduce base PowerVM driver (disk drv) https://review.openstack.org/409401
20:50:36 edleafe jaypipes: found it. _check_capacity_exceeded() creates new RP objects, and these are the ones whose generation gets updated instead of the originals
20:50:39 edleafe jaypipes: https://github.com/openstack/nova/blob/master/nova/objects/resource_provider.py#L1078-L1081
20:51:33 openstackgerrit Eric Fried proposed openstack/nova: WIP(3): Introduce base PowerVM driver (console) https://review.openstack.org/409402
20:51:43 dansmith diana_clarke: mriedem: thanks
20:52:44 EmilienM diana_clarke: fyi, depends-on works for inter-projects connection, but for the same project, you better to use git rebase
20:52:58 mriedem dansmith: dash doesn't support multiple changes for the -c option does it...?
20:53:18 mriedem EmilienM: it's fine either way in this case
20:53:39 EmilienM right, just fyi again
20:54:13 dansmith mriedem: dunno, I never use that
20:54:19 dansmith mriedem: patches accepted :)
20:54:21 openstackgerrit Eric Fried proposed openstack/nova: WIP(4): Introduce base PowerVM driver (cfg drv) https://review.openstack.org/409404
20:54:33 diana_clarke EmilienM: Oh right. I did know that, but I just drank some flu meds so I'm a bit groggy ;)
20:54:48 EmilienM diana_clarke: oops :-)
20:54:54 mdrabe dansmith: Is something enforcing having cells v2 in ocata?
20:55:41 mriedem mdrabe: yes, a db schema migration
20:55:41 dansmith mdrabe: various bits of cellsv2 have been required each release recently
20:55:43 dansmith mdrabe: there is more in ocata sure
20:55:57 mriedem mdrabe: you can't upgrade the api db schema without doing some setup first
20:56:33 openstackgerrit Jay Pipes proposed openstack/nova: placement: adds REST API for nested providers https://review.openstack.org/384807
20:56:34 openstackgerrit Jay Pipes proposed openstack/nova: pci: add uuid field to PciDevice object https://review.openstack.org/414768
20:56:34 openstackgerrit Jay Pipes proposed openstack/nova: placement: allow filter providers in tree https://review.openstack.org/377215
20:56:35 openstackgerrit Jay Pipes proposed openstack/nova: placement: SRIOV PF devices as child providers https://review.openstack.org/414769
20:56:36 openstackgerrit Jay Pipes proposed openstack/nova: placement: update client to set parent provider https://review.openstack.org/385693
20:56:37 openstackgerrit Jay Pipes proposed openstack/nova: placement: add nested resource providers https://review.openstack.org/377138
20:58:39 jaypipes edleafe: le sigh... yeah, line 1169 needs to be changed to:
20:58:41 mdrabe mriedem: What if you're starting fresh on ocata?
20:58:47 jaypipes rp.generation = _increment...
20:59:03 openstackgerrit Lee Yarwood proposed openstack/nova: libvirt: utils.get_disk return the target dev of the root device https://review.openstack.org/402716
20:59:05 openstackgerrit Lee Yarwood proposed openstack/nova: libvirt: Make guest.BlockDevice only use the target dev of a disk https://review.openstack.org/402163
20:59:05 openstackgerrit Lee Yarwood proposed openstack/nova: libvirt: Use the mirror element to detect when to pivot https://review.openstack.org/399735
20:59:07 mriedem mdrabe: same deal
20:59:17 mriedem mdrabe: you still have to have a cell0 cell mapping record in the api db
20:59:28 mriedem actually...
20:59:31 dansmith mriedem: the blocker doesn't enforce that it's done first,
20:59:34 mriedem i think the db schema migration ends early
20:59:41 dansmith but you still need that
20:59:43 mriedem if there are no flavors in the api db then it's a noop
21:00:19 mriedem mdrabe: we're also working on this nova-status upgrade check CLI for checking this stuff at any time prior to install/upgrade
21:00:27 mriedem just haven't integrated that yet into the grenade tooling we have
21:06:56 mdrabe mriedem: So say I didn't have any cells in my deployment down the road, things are certainly not going to work right? The cells requirement isn't just for want?
21:07:12 openstackgerrit Jay Pipes proposed openstack/nova: pci: add uuid field to PciDevice object https://review.openstack.org/414768
21:07:13 openstackgerrit Jay Pipes proposed openstack/nova: placement: update client to set parent provider https://review.openstack.org/385693
21:07:13 openstackgerrit Jay Pipes proposed openstack/nova: placement: SRIOV PF devices as child providers https://review.openstack.org/414769
21:07:31 dansmith mdrabe: you can have a cell of one, of course, if that's what you mean
21:07:48 dansmith mdrabe: everyone will be transitioning to a single-cell deployment all the time
21:08:08 mdrabe dansmith: But if I had none like we could in newton that's not gonna fly?
21:08:48 dansmith mdrabe: I'm not sure what you mean.. "one" cell is just a single row in the database -- nothing else is different
21:09:58 dansmith mdrabe: meaning, moving from a newton deployment to an ocata one with just one cell is just adding a row to the database, which says your existing stuff is "cell 1"
21:10:26 dansmith mdrabe: no extra services or configuration like normal cells (i.e. cellsv1) requires
21:11:04 mdrabe dansmith: But if I didn't have a single cell could I still deploy?
21:11:20 dansmith mdrabe: no, there's no such thing as "no cells" anymore
21:12:53 dansmith mdrabe: well, you could deploy with no cells, but you could never create any instances.. :P
21:14:08 mdrabe because the instance has to be built to a specific cell
21:14:55 dansmith yes
21:16:25 openstackgerrit Ed Leafe proposed openstack/nova: Fix missing RP generation update https://review.openstack.org/419137
21:16:44 edleafe jaypipes: The fix ^^
21:17:21 jaypipes edleafe: cheers man
21:17:28 jaypipes mriedem: I'm good with https://review.openstack.org/#/c/140045/ now.
21:17:34 jaypipes feel free to wallaby.
21:19:00 jaypipes edleafe, reviewed.
21:20:23 edleafe jaypipes: the functional test that was failing was in the later patch that removed Allocation.create()
21:20:34 openstackgerrit Vasyl Saienko proposed openstack/nova: Support Ironic interface attach/detach in nova virt https://review.openstack.org/364413
21:21:03 edleafe jaypipes: I'll see what I can add that would be close
21:21:51 jaypipes edleafe: cool, thanks Ed.
21:21:58 cdent more stuff borked?
21:22:38 edleafe cdent: one effect of removing Allocation.create() is that AllocationList.create_all() is much more stringent in its checking
21:23:15 edleafe cdent: and that uncovered a slew of ResourceProvider references that were stale as far as their generation was concerned
21:23:55 cdent in?
21:24:34 edleafe cdent: see https://review.openstack.org/419137 for what I fixed
21:24:41 cdent cool thanks
21:25:51 edleafe cdent: the tests in nova/tests/functional/db/test_resource_provider.py were creating allocations before there was any inventory
21:26:13 edleafe cdent: Allocation.create() allowed that. AllocationList.create_all() doesn't.
21:29:46 cdent gyres with gyres or something
21:34:32 openstackgerrit Jay Pipes proposed openstack/nova: pci: add uuid field to PciDevice object https://review.openstack.org/414768
21:34:33 openstackgerrit Jay Pipes proposed openstack/nova: placement: SRIOV PF devices as child providers https://review.openstack.org/414769
21:36:00 openstackgerrit Claudiu Belu proposed openstack/nova: Hyper-V: Adds vNUMA implementation https://review.openstack.org/282407
21:42:47 mriedem claudiub|2: i was just about done commenting on that one :)
21:43:17 dansmith mriedem: melwitt: don't mind me, just bragging: https://review.openstack.org/#/c/418141/
21:43:45 claudiub|2 ha, oups. :D
21:44:47 openstackgerrit Ed Leafe proposed openstack/nova: Fix missing RP generation update https://review.openstack.org/419137
21:44:50 mriedem claudiub|2: comments posted
21:44:58 mriedem dansmith: nice
21:45:10 mriedem dansmith: i can see your nippies from here
21:45:11 melwitt hah, yay
21:45:29 mriedem i guess that should have been in a PM
21:45:42 dansmith haha
21:46:11 openstackgerrit Vasyl Saienko proposed openstack/nova: Support Ironic interface attach/detach in nova virt https://review.openstack.org/364413

Earlier   Later