Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-01-11
15:05:05 gibi mriedem: no problem, I was not prepared but I could run it
15:10:30 efried Hi cores, would anyone be willing to look at the first two patches for the PowerVM integration bp? https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/powervm-nova-it-compute-driver
15:11:00 efried stephenfin gibi bauzas perhaps if you have time?
15:11:43 stephenfin efried: I've been meaning to review the OVS VIF one, but I'll have to leave the storage patches to someone else
15:11:45 bauzas efried: I'm still rushing on my feature branch :/
15:11:52 bauzas efried: blame mdbooth
15:12:09 mdbooth bauzas: Sorry :P
15:12:52 bauzas mdbooth: no worries, you just owe me a beer :p
15:12:58 bauzas et voila!
15:13:16 bauzas interesting, creating a new mdev even if you have a kernel race works
15:13:25 bauzas that's awesome news
15:13:37 bauzas since the mdev is *really* created, you can ask libvirt to use it
15:13:46 bauzas even if libvirt isn't able to show it per its API
15:13:53 bauzas amazing
15:14:10 bauzas so, one big caveat isn't really a big deal and not requiring a separate change
15:14:14 bauzas just docs
15:14:22 bauzas huzzah
15:14:24 efried stephenfin Thanks, anything helps. TBH, there's not a lot of storage (or network) understanding required for these. The guts are in the pypowervm lib; these patches are really just translating the nova-ese to powervm-ese.
15:19:14 mriedem stephenfin: do you see anything obviously wrong with https://review.openstack.org/#/c/532689/ ? i ran it last night through the experimental queue but results were never posted back on the patch
15:20:05 mriedem i can see it queued up in the experimental queue in zuul right now
15:22:24 mriedem gmann: https://review.openstack.org/#/c/330406/85/nova/api/openstack/compute/migrations.py@119
15:22:32 mriedem we have 2.0 and 2.1 all over the API, it doesn't seem to make a difference
15:24:07 openstackgerrit sahid proposed openstack/nova master: hardware: only take into account small pages https://review.openstack.org/532168
15:26:57 hongbin mrodden: mlavalle : want to make a call on the full substring or partial substring debate https://review.openstack.org/#/c/521683/ ?
15:27:18 mlavalle hongbin: I will later today. Thanks for the heads up
15:29:15 gibi efried: opened the patches, but it won't fitt to my today's schedule
15:29:27 efried gibi Thanks in advance
15:30:01 mriedem hongbin: full substring
15:30:10 hongbin mlavalle: mriedem : ack
15:32:24 kashyap Does anyone know what provides this one? I'm on CentOS 7.4, and DevStack fails with:
15:32:28 kashyap $ rpm -qf `which generate-subunit`
15:32:30 kashyap file /usr/bin/generate-subunit is not owned by any package
15:32:43 kashyap (Err the failure message was: "./stack.sh: line 488: generate-subunit: command not found")
15:33:32 mriedem ask in qa
15:33:49 kashyap Yep, checking there
15:35:43 kashyap mriedem: Hehe, I found a bug with your comment (#5) -- https://bugs.launchpad.net/devstack/+bug/1547379
15:35:47 openstack Launchpad bug 1547379 in devstack "./stack.sh: line 488: generate-subunit: command not found" [Undecided,Invalid]
15:36:09 kashyap It's indeed a mistake in my local.conf!
15:36:17 mriedem exellent
15:36:20 mriedem *excellent event
15:36:21 mriedem *gdi
15:36:32 kashyap (Because I copied my F27 local.conf to the old and sloth-like CentOS)
15:40:23 openstackgerrit Eric Berglund proposed openstack/nova master: PowerVM Driver: vSCSI volume driver https://review.openstack.org/526094
15:44:14 stephenfin mriedem: I can't see anything obvious, no. I'm guessing it's just zuul funkiness
15:44:37 stephenfin There's been a lot of that this week
15:48:41 mriedem stvnoyes: ildikov: jgriffith: about 10pm last night i thought about this https://etherpad.openstack.org/p/multi-attach-volume-queens - see L48
15:48:55 mriedem stephenfin: yeah, thanks for looking. have'nt merged any patches in 2 days...
15:50:38 ildikov mriedem: does that apply to live_migrate as well?
15:51:11 openstackgerrit Eric Berglund proposed openstack/nova master: PowerVM Driver: vSCSI volume driver https://review.openstack.org/526094
15:51:14 mriedem ildikov: no
15:51:26 mriedem ildikov: with live migration, it's the same volume attached on both the source and dest nodes,
15:51:39 mriedem and it's using the same disk config xml which already has the shareable flag set for the source guest xml
15:51:55 ildikov ok, I wasn't sure about the config on the dest
15:51:58 mriedem i think this is purely a swap volume wrinkle
15:52:21 mriedem so i think i need to handle that in between the libvirt patch and https://review.openstack.org/#/c/531244/
15:52:28 mriedem since https://review.openstack.org/#/c/531244/ bumps the rpc version
15:52:54 stvnoyes mriedem: at this point in Queens, what about explicitly not allowing swap volume if the volume is MA'd? Then we can test swap more exhaustively for rocky...
15:53:47 mriedem stvnoyes: because that means you'd need a new microversion for swap volume with multiattach support in rocky, and that would mean cinder (using nova client) would have to know to send that microversoin if swapping with a MA volume
15:54:04 mriedem stvnoyes: what needs to happen is like a 3 line code change i think
15:54:07 ildikov mriedem: I'm good with your solution
15:54:11 stvnoyes ah ok
15:54:21 mriedem if new_volume['multattach']: new_connection_info['multiattach'] = True
15:54:39 ildikov yep
15:54:41 mriedem thinking about a tempest test scenario to tickle this...
15:55:07 mriedem attach non-multiattach volume A, swap to multiattach volume B, then attach B to a 2nd serer
15:55:08 mriedem *server
15:55:31 mriedem if we didn't fix the swap volume piece, the 2nd attach for B on the 2nd server should be wonky
15:55:36 ildikov mriedem: do we track if the old volume was multiattach, but the new one is not?
15:55:51 mriedem ildikov: no. we can figure that out if needed, but i'm not sure if we care.
15:56:22 mriedem we did say in the cinder spec that you can't retype an in-use volume to change it's multiattach value
15:56:52 ildikov but with swap it's a new volume?
15:57:31 ildikov anyway, if the shareable flag cannot appear in the xml by accident for a non-multiattach volume, then good
15:57:43 openstackgerrit Sylvain Bauza proposed openstack/nova master: libvirt: create vGPU for instance https://review.openstack.org/528832
15:58:29 mriedem i'm not entirely sure how the various retype and volume migration stuff works on the cinder side
15:58:39 mriedem retype is like a resize in nova right? new flavor/volume type
15:58:53 mriedem volume migration is not a change in volume type, but it migrates the volume from one backend to another of the same type?
15:59:30 mriedem anywho, meeting time
15:59:36 ildikov yeah, I guess you're right, I always confuse myself with the different types of swap...
16:10:28 bauzas mdbooth: updated based on your comments
16:18:10 openstackgerrit Matt Riedemann proposed openstack/nova master: Address nits in change I7e01f95d7173d9217f76e838b3ea71555151ef56 https://review.openstack.org/532898
16:27:18 openstackgerrit Matt Riedemann proposed openstack/nova-specs master: Amend the migration paging spec for uuid in server migrations response https://review.openstack.org/532904
16:28:14 mriedem nova-multiattach job failed with NODE_FAILURE in the experimental queue :(
16:28:18 mdbooth bauzas: Hey, can you try a couple of things on your mdev box and document the results: create more mdevs than permitted, and create an mdev with an invalid name.
16:28:41 bauzas mdbooth: I'm just trying the former :)
16:29:55 mdbooth bauzas: The BFL around that function is fine by me. How do you plan to address all the other places which call _get_guest_xml?
16:31:30 ildikov mriedem: sadness :(
16:36:19 mnaser from what i understand, once an instance is scheduled in a cell, it will forever remain in that cell (rebuilds, live migrations, etc?)
16:36:27 openstackgerrit Eric Fried proposed openstack/nova master: Track associated sharing RPs in report client https://review.openstack.org/526539
16:36:27 openstackgerrit Eric Fried proposed openstack/nova master: Track provider traits in report client https://review.openstack.org/521686
16:36:28 openstackgerrit Eric Fried proposed openstack/nova master: ProviderTree.populate_from_iterable https://review.openstack.org/520756
16:36:28 openstackgerrit Eric Fried proposed openstack/nova master: Raise on API errors getting aggregates/traits https://review.openstack.org/526540
16:36:29 openstackgerrit Eric Fried proposed openstack/nova master: Raise conflict exception on RP create 409 https://review.openstack.org/532563
16:36:29 openstackgerrit Eric Fried proposed openstack/nova master: Track tree-associated providers in report client https://review.openstack.org/526541
16:36:30 openstackgerrit Eric Fried proposed openstack/nova master: WIP: Scheduler[Report]Client.get_provider_tree https://review.openstack.org/521098
16:36:30 openstackgerrit Eric Fried proposed openstack/nova master: SchedulerReportClient.set_traits_for_provider https://review.openstack.org/532564
16:36:31 openstackgerrit Eric Fried proposed openstack/nova master: WIP: Use update_provider_tree from resource tracker https://review.openstack.org/520246
16:36:31 openstackgerrit Eric Fried proposed openstack/nova master: WIP: ComputeDriver.update_provider_tree() https://review.openstack.org/521187
16:36:32 openstackgerrit Eric Fried proposed openstack/nova master: Fix nits in update_provider_tree series https://review.openstack.org/531260
16:36:59 bauzas mdbooth: yup, in separate change
16:37:08 efried gibi If you're still around, would you please re+W https://review.openstack.org/521686 (just rebased)

Earlier   Later