| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-08-25 | |||
| 17:12:26 | sean-k-mooney | im surprised that it got that far by the way | |
| 17:12:31 | sean-k-mooney | dansmith: no that is a bug | |
| 17:12:45 | sean-k-mooney | which ooo worked around by hardcodeing cidner to call nova admin api endpoint directly | |
| 17:12:57 | dansmith | is this the swap volume api or something else? | |
| 17:13:08 | lbragstad | fwiw - this is what i'm doing to recreate this https://review.opendev.org/c/openstack/tempest/+/805452 | |
| 17:13:17 | sean-k-mooney | i tought it was resize for nfs but ill have to go look it up | |
| 17:13:38 | sean-k-mooney | lbragstad: by the way i was expecting a failure before that when nova called neutron to bind the port | |
| 17:13:46 | lbragstad | yeah - me too | |
| 17:13:52 | sean-k-mooney | we do that as a admin | |
| 17:13:58 | dansmith | okay just saying.. the swap volume api has always been user-visible, but problematic.. people have used it for non-cinder tasks before and we discussed whether or not they should be able to.. but they are (or were) | |
| 17:14:36 | sean-k-mooney | dansmith: im pretty sure the issue i was refering too was for online volume resize whne cinder calls nova back to tell it its done | |
| 17:14:41 | sean-k-mooney | dansmith: ill see if i can find it | |
| 17:15:16 | dansmith | okay, there's some volume op that uses swap_volume for that sort of thing (cinder calling to nova) | |
| 17:15:20 | dansmith | but might be different | |
| 17:16:10 | sean-k-mooney | dansmith: i dont have the ooo one to hand but this is the osa dup https://bugs.launchpad.net/openstack-ansible/+bug/1902914 and redhat kcs...https://access.redhat.com/solutions/3675991 | |
| 17:17:01 | dansmith | okay I expect this is different | |
| 17:17:33 | sean-k-mooney | yes lbragstad current issue is different and expected | |
| 17:18:23 | sean-k-mooney | well expected in that each of the serviceis will need to know how to call each of the other services with the correct scopes | |
| 17:19:51 | sean-k-mooney | lbragstad: you could try booting an instance with no network as a simpler starting point | |
| 17:20:20 | sean-k-mooney | lbragstad: that would limit the interaction ot nova, keystone, glance and placment | |
| 17:20:47 | sean-k-mooney | placment would be the only thing that used any admin creds in that flwo also | |
| 17:20:59 | sean-k-mooney | we would use the user token for the glance image | |
| 17:22:19 | sean-k-mooney | im not 100% shoure what you would need to add but i think it would be something like network=none to your self.create_test_server( call | |
| 17:22:40 | lbragstad | yeah - that makes sense | |
| 17:24:13 | gmann | lbragstad: for system scope enabled. tempest does not create network | |
| 17:24:41 | gmann | lbragstad: that is skipped when create network was project scoped in neutron but that is changed i think | |
| 17:24:55 | lbragstad | ah - interesting | |
| 17:25:25 | gmann | lbragstad: | |
| 17:25:27 | gmann | lbragstad: https://review.opendev.org/c/openstack/tempest/+/798130 | |
| 17:26:21 | lbragstad | gmann is that still the case with https://review.opendev.org/c/openstack/tempest/+/805452/4/tempest/api/compute/admin/test_servers.py ? | |
| 17:26:26 | gmann | lbragstad: for network created for network, we need to modify that also to pass the project_id in neutron request for system scope | |
| 17:26:58 | gmann | lbragstad: yes. we have not changed that for ^^. and no network for system scoped cred | |
| 17:28:45 | gmann | I can work on that tomorrow to pass projetc_id as slaweq mentioned in that review | |
| 17:42:47 | lbragstad | gmann ok - that sounds good, i'm trying to track all of this stuff as a find it, but it's likely going to be a lot of sprawl | |
| 17:43:02 | lbragstad | stepping out for about 30 minutes | |
| 17:45:45 | opendevreview | melanie witt proposed openstack/placement master: Refactor consumer type methods for readability https://review.opendev.org/c/openstack/placement/+/806035 | |
| 17:46:35 | melwitt | sean-k-mooney, lyarwood: fixups for consumer types ^ | |
| 18:12:46 | sean-k-mooney | melwitt: im kind of +0.5 https://review.opendev.org/c/openstack/placement/+/806035/1/placement/objects/usage.py#91 | |
| 18:14:58 | melwitt | sean-k-mooney: thanks, I will reword it | |
| 18:22:45 | opendevreview | melanie witt proposed openstack/placement master: Refactor consumer type methods for readability https://review.opendev.org/c/openstack/placement/+/806035 | |
| 18:26:45 | sean-k-mooney | melwitt: by the way im just looking at the db tests i assume somehting in here https://github.com/openstack/nova/blob/master/nova/tests/unit/db/main/test_migrations.py that uses the OpportunisticDBTestMixin would be what you want for the alembic urls | |
| 18:30:34 | melwitt | sean-k-mooney: ah, cool. I have used that mixin once before I think. fwiw I wasn't clear on whether we need a database to cover some test inputs or if we only need the config parser behind it. whatever works | |
| 18:31:18 | melwitt | want to be able to do at least a couple of cases, one with the encoded chars and one with a password for example with a literal "%" in it | |
| 18:31:41 | melwitt | or I guess that could just be one case by putting both together in one example | |
| 18:32:28 | sean-k-mooney | ok im just going thought the different fixture we have aviable to us | |
| 18:32:30 | sean-k-mooney | https://github.com/openstack/oslo.db/blob/22b44ee18b1585bc2943a29ddaa18051cb1344ed/oslo_db/sqlalchemy/test_fixtures.py#L399 | |
| 18:32:46 | sean-k-mooney | we have the fix ture and the mixins | |
| 18:32:55 | melwitt | ah ok | |
| 18:33:39 | sean-k-mooney | the adhock db fixture allows us to pass a url https://github.com/openstack/oslo.db/blob/22b44ee18b1585bc2943a29ddaa18051cb1344ed/oslo_db/sqlalchemy/test_fixtures.py#L273 | |
| 18:34:30 | sean-k-mooney | the opertunistic mixin can be combind with any of the fixture i think | |
| 18:35:32 | sean-k-mooney | we need to modify the engin url if we want to test this right before tuning the db sync | |
| 18:36:56 | melwitt | meanwhile, the consumer types patches are failing on that dang server delete fail for AllocationDeleteFailed conflict 😑 | |
| 18:37:04 | sean-k-mooney | melwitt: i have updated stephens exisitng test case already just trying to see if theyer is a good way to also add the extra tests you where hoping for | |
| 18:37:46 | sean-k-mooney | if only there was a patch we could merge for that :) | |
| 18:37:52 | sean-k-mooney | thats furstrating | |
| 18:38:32 | melwitt | sean-k-mooney: ok, cool. lmk if you need a hand with it, I can try to help | |
| 20:01:31 | opendevreview | sean mooney proposed openstack/nova master: db: Handle parameters in DB strings https://review.opendev.org/c/openstack/nova/+/805663 | |
| 20:01:33 | sean-k-mooney | melwitt: ok that did not work but i found another way to test this | |
| 20:02:04 | sean-k-mooney | i realised after i fot the adhoc fixture working that would not test any of our config loading ectra | |
| 20:02:19 | sean-k-mooney | which is what we actully wanted to test | |
| 20:02:47 | sean-k-mooney | well initalising the alembic config form our config | |
| 20:10:26 | opendevreview | Artom Lifshitz proposed openstack/nova master: WIP: Update PCI requests in request spec on resize https://review.opendev.org/c/openstack/nova/+/806049 | |
| 20:15:41 | melwitt | sean-k-mooney: sweet! just added some comments | |
| 22:18:12 | opendevreview | Merged openstack/nova stable/wallaby: Allow X-OpenStack-Nova-API-Version header in CORS https://review.opendev.org/c/openstack/nova/+/796860 | |
| 22:18:15 | opendevreview | Merged openstack/placement master: Add consumer_types migration, database and object changes https://review.opendev.org/c/openstack/placement/+/669170 | |
| 22:18:30 | opendevreview | Merged openstack/placement master: Microversion 1.38: API support for consumer types https://review.opendev.org/c/openstack/placement/+/679441 | |
| 22:18:38 | opendevreview | Merged openstack/placement master: Switch ConsumerType to use an AttributeCache https://review.opendev.org/c/openstack/placement/+/679486 | |
| #openstack-nova - 2021-08-26 | |||
| 00:35:26 | opendevreview | Merged openstack/nova stable/train: Raise InstanceMappingNotFound if StaleDataError is encountered https://review.opendev.org/c/openstack/nova/+/777254 | |
| 02:45:33 | jameshyde | is this a known bug? every time after live migrating a VM from one hypervisor (source) to another hypervisor (target), nova-compute service is still up and running, but nova-conductor from controller keeps saying this: Failed to compute_task_migrate_server: Compute service of hostname_here is unavailable at this time.: nova.exception.ComputeServiceUnavailable: Compute service of hostname_here is | |
| 02:46:25 | jameshyde | it complains about nova compute service on source host | |
| 02:47:31 | jameshyde | restarting nova-compute works around this issue, but this requiring restarting nova-compute every time after vm live migration away from source host. | |
| 06:57:05 | opendevreview | Jorhson Deng proposed openstack/nova master: recheck the attachment_id after the reschedule successful https://review.opendev.org/c/openstack/nova/+/796209 | |
| 07:28:11 | kevinz | sean-k-mooney: thanks for the update, will ping them next week. | |
| 08:11:58 | lyarwood | jamesdenton: That sounds like a bug if the source service is marked as up and active | |
| 08:12:27 | lyarwood | jameshyde sorry | |
| 09:59:31 | opendevreview | sean mooney proposed openstack/nova master: db: Handle parameters in DB strings https://review.opendev.org/c/openstack/nova/+/805663 | |
| 10:00:16 | sean-k-mooney | lyarwood: it proably just the bug where rabbit mq loses i thin the topic on the compute recive queue | |
| 10:00:54 | opendevreview | Stephen Finucane proposed openstack/nova master: api: Add support for 'hostname' parameter https://review.opendev.org/c/openstack/nova/+/778550 | |
| 10:00:55 | opendevreview | Stephen Finucane proposed openstack/nova master: tests: Speed up 'servers' API tests https://review.opendev.org/c/openstack/nova/+/778732 | |
| 10:00:55 | opendevreview | Stephen Finucane proposed openstack/nova master: policy: Deprecate field from 'os-extended-server-attributes' policy https://review.opendev.org/c/openstack/nova/+/806131 | |
| 10:00:57 | sean-k-mooney | the one where the compute agent can send rpcs like the heartbeat but not recive them until you recreate the queue | |
| 10:01:22 | sean-k-mooney | eight that or the hostname has change to the instace.host does not match the current host | |
| 10:01:57 | sean-k-mooney | both would break the conductors ablity to manage vms on the host but the agent would be up | |
| 10:02:01 | lyarwood | I was assuming it came from here and the servicegroup API tbh | |
| 10:02:02 | lyarwood | https://github.com/openstack/nova/blob/2a78626a85954997d35f5fe62c50de297e2ca92d/nova/conductor/tasks/live_migrate.py#L284-L288 | |
| 10:02:12 | lyarwood | so assuming that's using the db driver | |
| 10:02:53 | sean-k-mooney | well the host in that case comes form instnace.host | |
| 10:03:06 | sean-k-mooney | so if the host name changed on the node that would not fine it properly | |
| 10:03:31 | lyarwood | yup right | |
| 10:05:45 | sean-k-mooney | lyarwood: melwitt by the way just updated the db url test assertin gthe content of the alembic conf and equating it to the the input string | |
| 10:06:03 | lyarwood | ah awesome missed that above | |
| 10:06:35 | sean-k-mooney | just did | |
| 10:06:43 | sean-k-mooney | altought its complaing about failing to merge? | |
| 10:06:58 | sean-k-mooney | oh we both did | |
| 10:07:48 | sean-k-mooney | right pep8... | |
| 10:08:08 | sean-k-mooney | for some reason i dont have pre-commmit in this once copy of nova | |
| 10:08:15 | lyarwood | failed to merge was just because you pushed a new ps of the nova change | |
| 10:08:25 | sean-k-mooney | oh ok | |
| 10:11:50 | sean-k-mooney | oh melwitt was right about the glbal needing to be reset | |
| 10:12:32 | sean-k-mooney | depending on order the test breaks ill fix that and push it again | |