Posted	Nick	Remark
#openstack-nova - 2020-03-24
11:33:16	elod	lyarwood: well, yes, other teams don't even bother with conflict lines either. and it's not a big thing, i know. maybe this forces people to backport things when the patch on newer branch has merged o:)
11:34:22	elod	fot me it seemed mriedem kept the rules quite strictly so I try to do the same :D
11:34:29	lyarwood	yup that's cool
11:34:51	lyarwood	I wanted to revisit the SHA issue at some point but until then feel free to enforce if you want to
11:37:47	elod	lyarwood: ok, no problem, I'll edit the SHA as that is the only thing that's needed :)
11:44:18	openstackgerrit	Elod Illes proposed openstack/nova stable/pike: nova-live-migration: Wait for n-cpu services to come up after configuring Ceph https://review.opendev.org/713036
11:51:38	lyarwood	elod: many thanks
11:52:48	elod	lyarwood: thanks, too :)
12:01:01	lyarwood	gibi: https://review.opendev.org/#/c/696834/ - would you mind taking a look at this today please?
12:11:06	kevinz	stephenfin: lyarwood: would you mind to re-visit this one: https://review.opendev.org/#/c/709494/? Thanks
12:17:33	alex_xu	dansmith: query another thing, https://review.opendev.org/#/c/713354/1/nova/objects/fields.py@277, as I understand, we can't change a field. But we won't support mixed policy in the middle of upgrade, we only enable it after upgrade finished. so can we change a field like that? just do a version object bump.
12:19:12	openstackgerrit	Stephen Finucane proposed openstack/nova master: libvirt: Use domain capabilities to get supported device models https://review.opendev.org/666915
12:19:13	openstackgerrit	Stephen Finucane proposed openstack/nova master: Add transform_image_metadata request filter https://review.opendev.org/665775
12:20:15	stephenfin	gibi: Tests added ^
12:28:30	luyao	alex_xu: https://review.opendev.org/#/c/687856 comments addressed, thanks for review
12:29:29	luyao	stephenfin: could you review vpmem live migration https://review.opendev.org/#/c/687856 if you get time?
12:30:46	stephenfin	kevinz: replied. one more thing needed, I think
12:31:12	stephenfin	luyao: I can, but I was waiting until alex_xu had most of his things addressed before :)
12:35:04	luyao	stephenfin: OK, thanks
12:36:06	openstackgerrit	Kevin Zhao proposed openstack/nova master: Add default cpu model for aarch64 https://review.opendev.org/709494
12:53:07	alex_xu	stephenfin: I think most of my things addressed...
13:33:52	gibi	lyarwood, stephenfin: ack on both, I'm on a call will check it after
13:35:14	openstackgerrit	John Garbutt proposed openstack/nova master: Add logic to enforce local api and db limits https://review.opendev.org/712139
14:04:17	stephenfin	lyarwood: do my comments on https://review.opendev.org/#/c/712607/ make sense?
14:04:29	openstackgerrit	John Garbutt proposed openstack/nova master: Enforce api and db limits https://review.opendev.org/712142
14:05:02	dansmith	artom: sorry, just got to your base test patch.. fix that thing right quick and I can +2
14:05:34	artom	(Dammit, missed the panix possibility)
14:07:47	dansmith	alex_xu: we can but you have to do it in a specific way.. I commented
14:08:04	dansmith	alex_xu: the hint is the test_objects hash change with no version bump
14:08:16	dansmith	alex_xu: good job catching that
14:10:12	openstackgerrit	John Garbutt proposed openstack/nova master: Add logic to enforce local api and db limits https://review.opendev.org/712139
14:10:13	openstackgerrit	John Garbutt proposed openstack/nova master: Enforce api and db limits https://review.opendev.org/712142
14:11:14	openstackgerrit	Artom Lifshitz proposed openstack/nova master: func tests: move _run_periodics() into base class https://review.opendev.org/714228
14:11:15	openstackgerrit	Artom Lifshitz proposed openstack/nova master: tests: work around malformed serial XML https://review.opendev.org/705546
14:11:15	openstackgerrit	Artom Lifshitz proposed openstack/nova master: Functional tests for NUMA live migration https://review.opendev.org/672595
14:11:16	openstackgerrit	Artom Lifshitz proposed openstack/nova master: NUMA LM: Add func test for bug 1845146 https://review.opendev.org/687404
14:11:17	openstack	bug 1845146 in OpenStack Compute (nova) train "NUMA aware live migration failed when vCPU pin set" [High,Fix committed] https://launchpad.net/bugs/1845146 - Assigned to Dan Smith (danms)
14:11:23	artom	dansmith, ^^
14:14:14	artom	\o/
14:20:58	nightmare_unreal	gibi: what if user pass both instance-id and cell-id
14:20:58	openstackgerrit	Luyao Zhong proposed openstack/nova master: bug-fix: set do_cleanup always True for libvirt driver https://review.opendev.org/714593
14:20:59	openstackgerrit	Luyao Zhong proposed openstack/nova master: support live migration with vpmems https://review.opendev.org/687856
14:20:59	openstackgerrit	Luyao Zhong proposed openstack/nova master: Track orphan instances and error migrations in resource tracker https://review.opendev.org/714653
14:30:01	openstackgerrit	Huaqiang Wang proposed openstack/nova master: Take the instance dedicated CPU list from 'cpu_pinning' if possible https://review.opendev.org/713352
14:30:01	openstackgerrit	Huaqiang Wang proposed openstack/nova master: 'cpu_pinning_requested' property could be derived directly from cpu_policy https://review.opendev.org/713353
14:30:02	openstackgerrit	Huaqiang Wang proposed openstack/nova master: Introduce 'MIXED' CPU allocation policy for instance https://review.opendev.org/713354
14:30:02	openstackgerrit	Huaqiang Wang proposed openstack/nova master: Introduce the interface of creating 'MIXED' policy instance through 'PCPU' and 'VCPU' https://review.opendev.org/713355
14:30:03	openstackgerrit	Huaqiang Wang proposed openstack/nova master: metadata: export the vCPU IDs that are pinning on the host CPUs https://review.opendev.org/688936
14:30:03	openstackgerrit	Huaqiang Wang proposed openstack/nova master: libvirt: support to create instance with dedicated and shared CPUs https://review.opendev.org/714655
14:30:04	openstackgerrit	Huaqiang Wang proposed openstack/nova master: Remove 'InstanceNUMACell.cpu_pinning_requested' field https://review.opendev.org/714656
14:30:04	openstackgerrit	Huaqiang Wang proposed openstack/nova master: Refactor the code in checking available host CPUs https://review.opendev.org/714657
14:30:05	openstackgerrit	Huaqiang Wang proposed openstack/nova master: [WIP] An alternative way for keeping instance dedicated CPUs https://review.opendev.org/714658
14:37:25	openstackgerrit	Stephen Finucane proposed openstack/nova master: libvirt: Use domain capabilities to get supported device models https://review.opendev.org/666915
14:37:26	openstackgerrit	Stephen Finucane proposed openstack/nova master: Add transform_image_metadata request filter https://review.opendev.org/665775
14:40:40	openstackgerrit	Kevin Zhao proposed openstack/nova master: Add default cpu model for aarch64 https://review.opendev.org/709494
14:47:11	openstackgerrit	Kevin Zhao proposed openstack/nova master: Add default cpu model for aarch64 https://review.opendev.org/709494
14:51:34	gibi	nightmare_unreal: good point. You can make the two filters (instance_uuid and cell_id) mutually exclusive
14:52:06	gibi	I don't see any reason why the user want to filter for both at the same time
14:52:07	nightmare_unreal	yup I was thinking something like that in for loop
14:53:39	nightmare_unreal	gibi: I am not sure but can we add/append to cells ? for e.g. if instance_id is provided cells variable will have cell_id of that instance and along with cell_id is also passed, can we just append it in cells. it can make code more simple
14:56:06	huaqiang	hi stephenfin，in my new patches, I haven't add a 'cpu_policy' property to 'InstanceNUMATopology' object.
14:56:06	huaqiang	hi stephenfin，in my new patches, I haven't add a 'cpu_policy' property to 'InstanceNUMATopology' object.
14:57:21	huaqiang	If I do, I have to replace every 'topology.cpu_pinning_requested' line with 'if topology.cpu_policy in "DEDICATED or MIXED"'
14:58:11	huaqiang	I think 'topology.cpu_pinning_requested' and 'topology.cpu_sharing_requested' are more concise.
14:59:40	gibi	nightmare_unreal: there is not much reason to append to the cells variable if both instance_uuid and cell_id is provided. Because today if the instance_uuid is passed the it means that heal _only_ that instance.
14:59:58	gibi	so if you just append the extra cell specified by the cell_id then you still only heal one instance
15:00:12	gibi	not one instance + a whole other cell
15:00:21	nightmare_unreal	ah yes
15:08:19	stephenfin	huaqiang: I don't think that replacement is that much work (I've it done locally and it's quite concise). However, adding 'cpu_policy' to the 'InstanceNUMATopology' object is a lot of work though that can wait, yes.
15:10:15	sean-k-mooney	we can infer the policy from the cpu pinning info in anycase
15:11:19	sean-k-mooney	the cpu_policy is also somewhat incorrectly stored in the insance numa cell https://github.com/openstack/nova/blob/master/nova/objects/instance_numa.py#L59-L60
15:12:11	gibi	stephenfin: I'm +2 on the image-metadata-prefiltering
15:13:00	sean-k-mooney	it should be in the InstanceNUMATopology instead bug you could add a property that would return the value form the first cell and in a later version add a cpu_policy field and remove it form teh instance numa cell
15:13:04	stephenfin	sean-k-mooney: yeah, I'd a suggestion to duplicate it to the topology and slowly deprecate use of the cell field, but it's not really related to this
15:13:38	huaqiang	stephenfin: nice!
15:14:05	sean-k-mooney	right its not related to this but we shoudl eventually move it
15:15:52	huaqiang	stephenfin, can you share some detail? or you can submit the changes
15:16:05	huaqiang	then I do the work under your changes
15:16:14	stephenfin	for the replacement?
15:16:19	huaqiang	yes
15:16:24	stephenfin	sure
15:17:07	sean-k-mooney	by the way if you want to start using the value form the isntanceNUMATopology without bumping the verion fo the object you can add this
15:17:10	sean-k-mooney	@property
15:17:12	sean-k-mooney	def cpu_policy(self):
15:17:14	sean-k-mooney	return self.cells[0].cpu_policy if self.cells else "shared"
15:17:37	stephenfin	sean-k-mooney: that's what I have
15:17:54	openstackgerrit	Stephen Finucane proposed openstack/nova master: Remove future imports https://review.opendev.org/714675
15:17:55	openstackgerrit	Stephen Finucane proposed openstack/nova master: Use unittest.mock instead of third party mock https://review.opendev.org/714676
15:17:56	stephenfin	albeit with a check to make sure all the cells have the same value and error out if not (it shouldn't ever be triggered)
15:18:04	huaqiang	sean-k-mooney,stephenfin: thanks
15:18:06	sean-k-mooney	cool
15:18:20	sean-k-mooney	stephenfin: ya so that cant happen today unless you messed with the db
15:18:26	sean-k-mooney	so i did not bother with the check
15:22:26	openstackgerrit	jayaditya gupta proposed openstack/nova master: Support for nova-manage placement heal_allocations --cell https://review.opendev.org/714459
15:22:27	sean-k-mooney	ok ill be afk for about an hour or so. bbl
15:31:03	gibi	lyarwood, stephenfin: is this a bugfix or a feature? https://review.opendev.org/#/c/696834
15:32:19	stephenfin	gibi: It's probably closer to a feature. We couldn't do this before and now we can
15:32:57	stephenfin	Should there be a specless blueprint? I didn't think of that
15:34:44	stephenfin	alex_xu: Think you could look at this? https://review.opendev.org/#/c/666915/
15:37:24	gibi	stephenfin: would be nice to track it somehow. I'm OK to have a specless bp as the impl basically done and only affects the libvirt driver

Earlier   Later