Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-12-06
12:08:24 lyarwood vishalmanchanda: and then rescan within the instance for it to pickup the increased block device size
12:10:57 vishalmanchanda lyarwood: ok thanks for the help. I will check and let you know.
12:11:10 vishalmanchanda Have a good day:)
12:20:29 lyarwood np
12:43:44 openstackgerrit Merged openstack/nova master: Remove unused rootwrap filters https://review.opendev.org/697134
13:11:57 efried gibi: Do we still need to keep tests where allocation_request_version is less than the latest?
13:15:55 gibi efried: good question. I think those are created in the scheduler and scheduler is updated along with the controller services first
13:16:07 gibi so that should be always the latest
13:17:16 gibi but I remember there was some controvery along this line
13:17:31 efried e.g. this one https://opendev.org/openstack/nova/src/branch/master/nova/tests/unit/conductor/tasks/test_migrate.py#L730
13:17:44 efried Is it possible to actually get 1.19 there?
13:19:07 gibi efried: no I don't think so.
13:19:23 efried I'll just change it to 1.34 and make the allocation requests modern.
13:19:37 gibi efried: that test checks the old compute service version and that is still valid test but you can change the request version to latest
13:19:37 efried There's a bunch of other tests in that suite that have 1.19 hardcoded
13:20:03 efried but that may just be because it never failed before so wasn't noticed :)
13:45:41 openstack Launchpad bug 1855314 in OpenStack Compute (nova) "Typo in documentation of "host aggregates in nova"" [Undecided,New]
13:45:41 mriedem easy docs bug if someone wants to throw up a fix https://bugs.launchpad.net/nova/+bug/1855314
13:54:59 mriedem please to be getting core reviews on these stable stein backports that have a +2 already https://review.opendev.org/#/q/topic:bug/1852610+(status:open+OR+status:merged)+branch:stable/stein
13:56:59 openstackgerrit Boris Bobrov proposed openstack/nova master: Switch to uses_virtio to enable iommu driver for AMD SEV https://review.opendev.org/696697
13:57:00 openstackgerrit Boris Bobrov proposed openstack/nova master: Create a controller for qga when SEV is used https://review.opendev.org/693072
13:57:00 openstackgerrit Boris Bobrov proposed openstack/nova master: Also enable iommu for virtio controllers and video in libvirt https://review.opendev.org/684825
14:11:29 kashyap Probably on PTO
14:16:41 mriedem efried: questions in https://review.opendev.org/#/c/696418/ - also wondering about a reno
14:34:41 openstackgerrit Matt Riedemann proposed openstack/nova master: Remove dict compat from populate_filter_properties https://review.opendev.org/697669
14:41:25 openstackgerrit Matt Riedemann proposed openstack/nova master: Remove now invalid cells v1 comments from conductor code https://review.opendev.org/697670
14:47:49 openstackgerrit Kashyap Chamarthy proposed openstack/nova master: libvirt: Remove MIN_QEMU_FILE_BACKED_DISCARD_VERSION https://review.opendev.org/695915
14:47:49 openstackgerrit Kashyap Chamarthy proposed openstack/nova master: libvirt: Remove MIN_LIBVIRT_KVM_AARCH64_VERSION https://review.opendev.org/695901
14:47:50 openstackgerrit Kashyap Chamarthy proposed openstack/nova master: libvirt: Remove MIN_{LIBVIRT,QEMU}_FILE_BACKED_VERSION https://review.opendev.org/695942
15:12:54 melwitt kashyap: would you mind revisiting this review pls? https://review.opendev.org/589769 author has updated after your comments
15:18:39 openstackgerrit Balazs Gibizer proposed openstack/nova master: Fup for I63c1109dcdb9132cdbc41010654c5fdb31a4fe31 https://review.opendev.org/697678
15:37:09 gibi stephenfin: what kind of generic exception type you would like to see in https://review.opendev.org/#/c/696574/5/nova/objects/instance.py@1238 ? ValueError? for both cases?
15:40:23 gibi stephenfin: I fine not to have BuildAbortException there directly I'm just not sure if you really want a blanked ValueError for both cases or something more specific like AmbiguousResourceProviderForPciRequest and UnexpectResourceProviderNameForPciRequest for the two cases?
15:40:37 gibi s/blanked/blanket/
15:50:37 gibi stephenfin: OK I think I going for the nicer named exceptions and you can express your opinion in the review :)
15:54:06 efried mriedem: Yeah, gibi, how do you feel about that? Should we start advertising (in docs etc) that you can use string suffixes in your flavors? I kinda feel like it's something we could do, but is certainly not necessary, so that could be a separate change.
15:55:59 gibi efried: do I understand correctly that the old numbered syntaxt still works?
15:56:06 efried absolutely
15:57:56 gibi efried: I think it is worth to add a sentence or two to the flavor doc. But that can be done in a follow up
15:58:06 efried yeah
15:58:10 efried mriedem: okay with that?
15:58:47 mriedem i'm fine with a fup
15:58:58 mriedem those existing tests don't need to change then right?
16:00:28 efried mriedem: they needed to change a little bit because some of the previously-bogus extra specs are now not-ignored.
16:00:41 efried but no, I didn't need to change as many as I did.
16:00:45 efried see response in the patch.
16:05:22 mriedem ack, will reply and +2 in a sec
16:05:32 mriedem waiting to push up my refactor of build_request_spec since you tickled it :)
16:05:48 mriedem these todos from bauzas have been wafting in here for years
16:07:26 efried heh. wafting.
16:10:18 donnyd So I am hoping I can get a pointer with the right direction to go. I am having an issue where VM's just get stuck in scheduling. No errors, no logs and the hypervisor reports its up. Never seen nova do this before, there is usually something that says I couldn't build a machine because X.
16:10:44 efried gibi: FYI I'm having trouble with this one "unit" test (not sure how it gets away with not being in the functional set): https://opendev.org/openstack/nova/src/branch/master/nova/tests/unit/conductor/test_conductor.py#L1046
16:10:58 efried donnyd: which logs have you inspected?
16:11:17 donnyd efried: all of them... conductor, scheduler, api, and compute
16:11:18 openstackgerrit Matt Riedemann proposed openstack/nova master: Make build_request_spec return an actual RequestSpec https://review.opendev.org/697686
16:11:30 donnyd the only thing I get that is close to related is an error in libvirt
16:11:53 donnyd virNetSocketReadWire : End of file while reading data: Input/output error
16:12:12 mriedem efried: done, +W'ed
16:12:17 efried thanks mriedem
16:12:26 efried donnyd: what release are you running?
16:12:48 donnyd this is for one of my older deployments on rocky
16:13:25 gibi efried: does that unit test still fails on your last mapping refactor patch? I can take a look
16:13:53 efried gibi: it does, yes. I think I have fixed all the others locally, can push that if it would help.
16:14:38 efried gibi: if you look at the results, there's an exception setting up the report client to talk to placement, which is odd -- if that's an issue now, it should have been an issue before.
16:14:46 efried But maybe it "didn't matter" before somehow
16:15:14 efried https://storage.gra1.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_e07/696992/6/check/openstack-tox-py36/e074f8e/testr_results.html.gz
16:16:14 openstackgerrit Matt Riedemann proposed openstack/nova master: Remove now invalid cells v1 comments from conductor code https://review.opendev.org/697670
16:16:27 openstackgerrit Balazs Gibizer proposed openstack/nova master: Move _update_pci_request_spec_with_allocated_interface_name https://review.opendev.org/696574
16:16:28 openstackgerrit Balazs Gibizer proposed openstack/nova master: Support live migration with qos ports https://review.opendev.org/695905
16:17:01 efried wait, maybe I'm pointing to the wrong test, hold daphone.
16:17:01 gibi efried: if you can push your local change the I will grab them and reproduce the remaining failure locally
16:17:09 efried ight, will do.
16:17:35 openstackgerrit Eric Fried proposed openstack/nova master: Use provider mappings from Placement (mostly) https://review.opendev.org/696992
16:17:36 openstackgerrit Eric Fried proposed openstack/nova master: Restore test_minbw_allocation_placement in nova-next job https://review.opendev.org/697586
16:19:11 efried gibi: sorry, it's test_schedule_and_build_instances_fill_request_spec
16:19:19 gibi ack
16:22:08 efried bbiab
16:26:09 openstackgerrit Matt Riedemann proposed openstack/nova master: Remove dict compat from populate_filter_properties https://review.opendev.org/697669
16:26:09 openstackgerrit Matt Riedemann proposed openstack/nova master: Remove now invalid cells v1 comments from conductor code https://review.opendev.org/697670
16:26:10 openstackgerrit Matt Riedemann proposed openstack/nova master: Make build_request_spec return an actual RequestSpec https://review.opendev.org/697686
16:27:31 gibi efried_afk: I think stack trace in the log is missleading. At the end there is a dropped exception that seems relevant
16:27:34 gibi Original exception being dropped: ['Traceback (most recent call last):\n', ' File "nova/conductor/manager.py", line 1532, in schedule_and_build_instances\n scheduler_utils.fill_provider_mapping(request_spec, host)\n', ' File "nova/scheduler/utils.py", line 1234, in fill_provider_mapping\n mappings = jsonutils.loads(host_selection.allocation_request)[\'mappings\']\n', "KeyError: 'mappings'\n"]
16:32:03 gibi efried_afk: replied in the review. There is some a_c responses used by that test which does not have mapping key
16:37:34 openstackgerrit Matt Riedemann proposed openstack/nova master: Make build_request_spec return an actual RequestSpec https://review.opendev.org/697686
16:42:19 openstackgerrit Stephen Finucane proposed openstack/nova master: functional: Unify '_wait_until_deleted' implementations https://review.opendev.org/689181
16:42:20 openstackgerrit Stephen Finucane proposed openstack/nova master: functional: Remove 'get_invalid_image' https://review.opendev.org/697454
16:42:20 openstackgerrit Stephen Finucane proposed openstack/nova master: functional: Unify '_build_minimal_create_server_request' implementations https://review.opendev.org/695024
16:42:21 openstackgerrit Stephen Finucane proposed openstack/nova master: functional: Add unified '_build_server' helper function https://review.opendev.org/697537
16:42:21 openstackgerrit Stephen Finucane proposed openstack/nova master: functional: Make '_IntegratedTestBase' subclass 'InstanceHelperMixin' https://review.opendev.org/689182
16:42:22 openstackgerrit Stephen Finucane proposed openstack/nova master: functional: Add '_delete_server' to 'InstanceHelperMixin' https://review.opendev.org/697539
16:42:22 openstackgerrit Stephen Finucane proposed openstack/nova master: functional: Add unified '_(build|create)_flavor' helper functions https://review.opendev.org/697538
16:42:23 openstackgerrit Stephen Finucane proposed openstack/nova master: functional: Remove 'api' parameter https://review.opendev.org/697694
16:42:23 openstackgerrit Stephen Finucane proposed openstack/nova master: functional: Add '_create_server' helper https://review.opendev.org/697540
17:00:30 efried Thanks gibi, I figured that was probably it but I was having a hard time finding where those responses came from.
17:05:02 openstackgerrit Eric Fried proposed openstack/nova master: Restore test_minbw_allocation_placement in nova-next job https://review.opendev.org/697586
17:05:02 openstackgerrit Eric Fried proposed openstack/nova master: Use provider mappings from Placement (mostly) https://review.opendev.org/696992
17:05:08 efried gibi: that should be good to go now. Thanks for the help.
17:14:14 openstackgerrit Matt Riedemann proposed openstack/nova master: Support RequestSpec in populate_retry https://review.opendev.org/697697
17:25:14 mriedem we need to bump the conductor rpc api to 2.0 but i don't want to do that work
17:25:28 mriedem it would allow removing a lot of legacy filter_properties and request spec cruft from compute and conductor though

Earlier   Later