Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-09-10
16:18:45 bauzas stephenfin: k thanks, I'll proactively add this in the prelude section
16:19:35 bauzas actually, I'll hold on it
16:27:30 openstackgerrit Sylvain Bauza proposed openstack/nova master: Victoria 22.0.0 prelude section https://review.opendev.org/751045
16:35:45 openstackgerrit Merged openstack/python-novaclient master: Update master for stable/victoria https://review.opendev.org/750935
16:39:34 bauzas gibi: https://review.opendev.org/751045
16:39:41 bauzas it's a start
16:39:54 gibi bauzas: thanks
16:40:05 gibi I will read it tomorrow
16:42:54 lyarwood \o
16:52:17 sean-k-mooney so ya there are races in these tests...
16:53:37 sean-k-mooney specificlaly looks like we are not waiting for the service to be forced_down before doing the evacuate
16:53:52 sean-k-mooney we are just calling the api then continuing
17:08:02 lyarwood sean-k-mooney: which tests are these?
17:08:59 openstackgerrit Merged openstack/nova master: docs: Remove resize, cold migration from vTPM limitations https://review.opendev.org/750675
17:10:37 lyarwood oh gate/test_evacuate.sh doesn't wait until the service is down for the first round of evacuations
17:15:49 openstackgerrit Merged openstack/nova master: functional: Add 'start_compute' helper https://review.opendev.org/746943
17:15:55 openstackgerrit Merged openstack/nova master: functional: Use tempdir for CONF.instances_path https://review.opendev.org/746944
17:16:03 openstackgerrit Merged openstack/nova master: functional: Clean up PCI tests https://review.opendev.org/746945
17:16:04 openstackgerrit Luyao Zhong proposed openstack/nova master: Rollback when reset running live migration https://review.opendev.org/747899
17:16:04 openstackgerrit Luyao Zhong proposed openstack/nova master: Improve orphans tracking in resource tracker https://review.opendev.org/745282
17:16:14 openstackgerrit Merged openstack/nova master: functional: Add cold migration tests for PCI servers https://review.opendev.org/746946
17:21:16 stephenfin sean-k-mooney: did you get that patch reworked?
17:24:03 sean-k-mooney stephenfin: no
17:24:30 sean-k-mooney lyarwood: on sec ill provid the link
17:24:54 sean-k-mooney stephenfin: some of the func test are not correct
17:25:13 sean-k-mooney stephenfin: they are not waiting for palcment to be updated after we set teh comptue service down
17:26:16 sean-k-mooney lyarwood: stephenfin https://github.com/openstack/nova/blob/master/nova/tests/functional/regressions/test_bug_1815153.py#L108
17:26:32 sean-k-mooney we put the service down
17:26:56 sean-k-mooney but form the debug logs i am not seing us update placment before we then do the migrate or eveacuate in the next test
17:27:46 sean-k-mooney https://review.opendev.org/#/c/654596/8/nova/compute/api.py@5156 should have made the api call update placment
17:28:09 sean-k-mooney and the tests are using the latest microverion
17:29:00 lyarwood sean-k-mooney: that's not due to something with the placement fixture?
17:29:08 sean-k-mooney it might be
17:29:12 sean-k-mooney im going to mock _update_compute_provider_status
17:29:17 sean-k-mooney and see if its called
17:29:31 sean-k-mooney but i dont think it is being called
17:29:45 sean-k-mooney well i dont see a http request to placment to update the traits
17:29:46 openstackgerrit Lee Yarwood proposed openstack/nova master: test_evacuate: Wait until subnode is disabled before starting tests https://review.opendev.org/751065
17:30:35 lyarwood anyway ^ is what I thought you were talking about, we should be waiting there as well tbh
17:31:17 sean-k-mooney ya so i was debating just pooling for the api to show the service down and sleeping
17:31:58 sean-k-mooney i tried runing the periodic task to have update resouce provider un too but it didnt help
17:32:08 sean-k-mooney i think since the compute service is stopped that is why
17:32:21 sean-k-mooney since we both stop it and set force down
17:33:26 sean-k-mooney lyarwood: but yes waiting there also makes sense
17:33:41 sean-k-mooney lyarwood: we have similar logic in whitebox now for when we update the config
17:35:37 openstackgerrit Lee Yarwood proposed openstack/nova master: test_evacuate: Wait until subnode is disabled before starting tests https://review.opendev.org/751065
17:44:40 openstackgerrit Merged openstack/python-novaclient master: Add Python3 wallaby unit tests https://review.opendev.org/750937
17:45:41 openstackgerrit Stephen Finucane proposed openstack/nova master: Support SRIOV interface attach and detach https://review.opendev.org/740995
17:45:41 openstackgerrit Stephen Finucane proposed openstack/nova master: Make PCI claim NUMA aware during live migration https://review.opendev.org/748453
17:49:49 sean-k-mooney oh this is just doing self.rpcapi.set_host_enabled(context, service.host, enabled)
17:50:10 sean-k-mooney so the api is relying on the compute node to update it? that cant be right
18:49:21 openstackgerrit Merged openstack/nova master: Move confirm resize under semaphore https://review.opendev.org/747745
18:52:32 openstackgerrit Merged openstack/nova stable/ussuri: Reject resize operation for accelerator https://review.opendev.org/749701
18:52:41 openstackgerrit Merged openstack/nova master: Undeprecate the vmwareapi driver https://review.opendev.org/742407
18:52:56 openstackgerrit Merged openstack/nova stable/ussuri: func: Introduce a server_expected_state kwarg to InstanceHelperMixin._live_migrate https://review.opendev.org/746527
18:53:06 openstackgerrit Merged openstack/nova stable/ussuri: Add regression test for bug 1879787 https://review.opendev.org/746528
18:53:06 openstack bug 1879787 in OpenStack Compute (nova) "post_live_migration does not handle Neutron errors" [Medium,Fix released] https://launchpad.net/bugs/1879787 - Assigned to Artom Lifshitz (notartom)
19:11:07 openstackgerrit Lee Yarwood proposed openstack/nova master: test_evacuate: Wait until subnode is down before starting tests https://review.opendev.org/751065
19:15:19 sean-k-mooney lyarwood: oh right force down changes the state field
19:26:32 lyarwood sean-k-mooney: yup my bad
19:26:38 lyarwood sean-k-mooney: should be correct this time
19:26:50 lyarwood sean-k-mooney: was going to hack around with the evacuate job anyway this evening
19:33:20 openstackgerrit Merged openstack/nova stable/ussuri: post live migration: don't call Neutron needlessly https://review.opendev.org/750374
21:34:45 sean-k-mooney lyarwood: so ya AssertionError: Expected '_update_compute_provider_status' to have been called once. Called 0 times.
21:35:22 sean-k-mooney also i dont think that would wokr anyway but i think there is something up with the fixture
21:35:46 sean-k-mooney self.api.put_service(
21:35:49 sean-k-mooney source_compute_id, {'forced_down': 'true'})
21:36:02 sean-k-mooney is not causing that to run
21:36:05 sean-k-mooney and it should
21:37:06 sean-k-mooney it might be an issue with the OSAPIFixture
22:54:28 openstackgerrit Merged openstack/nova master: Support SRIOV interface attach and detach https://review.opendev.org/740995
23:10:27 sean-k-mooney lyarwood: by the way i figured out what the issue with the test is i think
23:11:32 sean-k-mooney the test is seting the force donw flag
23:11:40 sean-k-mooney but its not disabling the service
23:11:42 sean-k-mooney https://github.com/openstack/nova/blob/c57d52e1978210fb3260bf8ea442237d548db576/nova/tests/functional/regressions/test_bug_1815153.py#L105-L107
23:13:09 sean-k-mooney service_update is only chekcing for disabled
23:13:10 sean-k-mooney https://github.com/openstack/nova/blob/master/nova/compute/api.py#L5820
23:15:10 sean-k-mooney the compute api also does a call to the compute manager to actully do the placment update
23:15:14 sean-k-mooney https://github.com/openstack/nova/blob/509c01e86d7f1168cf3fc963608b557f65f47f74/nova/compute/api.py#L5798
23:15:21 sean-k-mooney so that also wont work in this case.
23:21:19 sean-k-mooney so beacuse of this else https://github.com/openstack/nova/blob/6f6b63ee5f7ae180d5f81e61f04332f7aa14f9be/nova/scheduler/filters/compute_filter.py#L43-L47 the ComputeFilter and the compute_status_filter prefilter is not equivlent since the former check both for disable and up were as the later only check for disabled
23:22:59 sean-k-mooney so if i want to disable the ComputeFilter by default and use the prefilter instead i need to change teh api and periodic task to also chekc the state vaule not just the status.
23:23:14 sean-k-mooney or rahter check if the serivce is disable or down
23:23:34 sean-k-mooney and add COMPUTE_STATUS_DISABLED in both cases
23:23:45 sean-k-mooney or maybe add a COMPUTE_STATUS_DOWN trait too
23:24:00 sean-k-mooney if we dont want to over load COMPUTE_STATUS_DISABLED
23:24:43 sean-k-mooney so with that in mind i might undo the deprecation of the compute filter for victoria and just deprecte the az filter since the placment prefilter for that has the same behavior
23:25:04 sean-k-mooney and then fix the compute prefilter behavior in wallaby
23:25:33 sean-k-mooney stephenfin: gibi when ye are awake does ^ make sense to ye
23:26:55 sean-k-mooney also sriov attach/detach megered :) that is good
#openstack-nova - 2020-09-11
00:34:14 luyao stephenfin: Hi, recheck succeeds at '
00:34:15 luyao Rollback when reset running live migration' https://review.opendev.org/#/c/747899/, thanks for review
01:30:55 openstackgerrit zhufl proposed openstack/nova master: Fix invalid assert_has_calls https://review.opendev.org/751143
02:24:53 openstackgerrit Eric Xie proposed openstack/nova master: Fix error that cannot overwrite policy rule for 'forced_host' https://review.opendev.org/751152
02:31:37 openstackgerrit Merged openstack/nova master: Track error migrations in resource tracker https://review.opendev.org/745281
02:56:43 openstackgerrit Rui Zang proposed openstack/nova master: DO NOT REVIEW https://review.opendev.org/751156
07:11:43 gibi sean-k-mooney: sounds interesting. I will have to dig into it a bit
07:39:00 bauzas good Friday Nova
07:39:26 bauzas gibi: I have a specific concern on the relnote for the vmwareapi driver undeprecation but I forgot to tell it during the meeting yesterday
07:39:29 bauzas gibi: https://review.opendev.org/#/c/742407/5/releasenotes/notes/undeprecate-vmware-victoria-2eaf5d877733f8d9.yaml

Earlier   Later