| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2026-10-12 | |||
| 11:34:05 | gibi | bauzas: do we know why the graceful shutdon slowness is not hitting the graceful shutdown job at every run? The patch suggest we never stop the event thread and therefore never exit before the full graceful shutdown timeout, that sonds like something we should hit all the time. Why it is then not blocking the gate? | |
| 11:36:10 | sean-k-mooney | stephenfin: my only reall issue with that is it makes it pretty hard to tell at a clance if its correct | |
| 11:38:24 | stephenfin | Yeah, I had the same concern, but I wanted to avoid hardcoding since there's a chance this will change again | |
| 11:38:43 | sean-k-mooney | well that kind of my concern | |
| 11:38:46 | stephenfin | The other approach I could do is just to drop the test. It's really testing functionlity of alembic/sqlalchemy | |
| 11:38:53 | stephenfin | *functionality | |
| 11:39:01 | sean-k-mooney | that the internal will change again and we will need to replace render_as_string with somethign else | |
| 11:44:02 | stephenfin | Yeah, okay, let me rework this to just drop the test so. There's nothing nova-specific in here anymore https://github.com/openstack/nova/blob/2b5fedace12dba41600bec10ca73476dbcfc7d10/nova/db/migration.py#L39-L51 | |
| 11:48:34 | opendevreview | Stephen Finucane proposed openstack/nova master: tests: Remove DB URL render test https://review.opendev.org/c/openstack/nova/+/1009877 | |
| 11:53:18 | stephenfin | sean-k-mooney: ^ | |
| 11:57:38 | gibi | bauzas: gmaan: I reviewed https://review.opendev.org/c/openstack/nova/+/1005022 and I think this is a valid bug. But I'm very curious why our testing did not caught it as compute never exiting before the full graceful timeout? | |
| 11:58:22 | gibi | I mean caught it in the devstack based graceful testing as in functional test we almost never enable libvirt event handling | |
| 11:59:12 | opendevreview | HyungJoon Kim proposed openstack/nova-specs master: Add spec for booting from Cinder backup-based image https://review.opendev.org/c/openstack/nova-specs/+/1009014 | |
| 11:59:55 | opendevreview | Stephen Finucane proposed openstack/nova master: Replace use of datetime.timezone.UTC https://review.opendev.org/c/openstack/nova/+/989515 | |
| 12:00:01 | opendevreview | Stephen Finucane proposed openstack/nova master: trivial: Fix outstanding flake8 issues https://review.opendev.org/c/openstack/nova/+/999714 | |
| 12:22:40 | opendevreview | Thibaut Démaret proposed openstack/nova-specs master: Spec for default SSD rotation rate on instance disks https://review.opendev.org/c/openstack/nova-specs/+/979696 | |
| 12:49:22 | opendevreview | Stephen Finucane proposed openstack/nova master: trivial: Fix outstanding flake8 issues https://review.opendev.org/c/openstack/nova/+/999714 | |
| 12:49:38 | opendevreview | Stephen Finucane proposed openstack/nova master: Add ruff-check https://review.opendev.org/c/openstack/nova/+/974441 | |
| 12:49:53 | opendevreview | Stephen Finucane proposed openstack/nova master: Enable G (flake8-logging-format) rules https://review.opendev.org/c/openstack/nova/+/998557 | |
| 12:50:50 | bauzas | sorry folks, I was AWOL, gibi : gtk | |
| 13:10:08 | bauzas | gibi: I don't know why we don't hit this libvirt event thread issue every time, this sounds like a performance issue only happening on slow nodes I'd say | |
| 13:11:18 | bauzas | I've noticed some job failures on my change only occurring due to long times (like a volume resize), which leads me thinking we may have a CI nodepool performance problem on some nodes | |
| 13:14:42 | gibi | the patch you pointed at showing a bug that should affect all nova-compute services with libvirt virt driver | |
| 13:15:31 | gibi | in tests where we shut down such service | |
| 13:15:43 | gibi | which I think we do in the graceful shutdown job | |
| 13:15:47 | bauzas | honestly, I'm also lost in translation, it should have been a blocking issue | |
| 13:21:15 | opendevreview | ribaudr proposed openstack/nova master: Add reproducer for bug 2037585: trunk port deletion fails https://review.opendev.org/c/openstack/nova/+/1010008 | |
| 13:21:16 | opendevreview | ribaudr proposed openstack/nova master: Preserve trunk parent ports when deallocating instance networking https://review.opendev.org/c/openstack/nova/+/1010009 | |
| 13:33:00 | opendevreview | Lucas HENRY proposed openstack/nova master: Start the manager graceful shutdown once the RPC server drained https://review.opendev.org/c/openstack/nova/+/1009164 | |
| 13:33:01 | opendevreview | Lucas HENRY proposed openstack/nova master: libvirt: Stop the event threads in Host.cleanup() https://review.opendev.org/c/openstack/nova/+/1010011 | |
| 14:18:54 | opendevreview | Balazs Gibizer proposed openstack/nova master: Reproduce cross cell move losing trusted certs https://review.opendev.org/c/openstack/nova/+/1010021 | |
| 14:46:03 | gmaan | gibi: ack, will check it | |
| 14:50:40 | gibi | gmaan: thanks | |
| 15:12:34 | mhen | dansmith: assuming you are "danms" on the PTG etherpad, I added some responses to your questions in the volume encryption cross session section for tomorrow to hopefully clarify a bit more. | |
| 15:58:55 | gouthamr | Uggla: hola, running late in the os-agentic room | |
| 16:01:24 | Uggla | gouthamr, we have a 5mn to 10mn break. I hope you'll be able to join right after. | |
| 16:01:44 | gouthamr | woah, that helps me :) | |
| 16:07:45 | Uggla | gouthamr I hope :) | |
| 16:44:17 | bauzas | gibi: see, I'm tempted to think more and more that we have some not performant nodepools https://review.opendev.org/c/openstack/nova/+/1005727/5 | |
| 16:52:30 | opendevreview | YushoYamaguchi proposed openstack/nova-specs master: Disable guest CPU flags per flavor https://review.opendev.org/c/openstack/nova-specs/+/1010041 | |
| 18:54:18 | tkajinam | could anyone review this backport https://review.opendev.org/c/openstack/nova/+/990096 (of regression test) to unblock the following actual fix ? | |
| 19:01:40 | sean-k-mooney | tkajinam: done | |
| 19:01:47 | tkajinam | sean-k-mooney, thx ! | |
| 19:33:43 | opendevreview | Merged openstack/nova stable/2026.1: Reproduce bug/2152697 https://review.opendev.org/c/openstack/nova/+/990096 | |
| 21:06:20 | opendevreview | Merged openstack/nova stable/2026.1: Fix live migration with pinned VM and iothreads https://review.opendev.org/c/openstack/nova/+/990097 | |
| 21:06:30 | opendevreview | Merged openstack/nova master: Guard against deleting a concurrent request's BDM on timeout https://review.opendev.org/c/openstack/nova/+/1005727 | |
| 21:39:39 | opendevreview | Arnaud Morin proposed openstack/nova master: Add info log when destination hypervisor is too old https://review.opendev.org/c/openstack/nova/+/907972 | |
| 22:13:06 | opendevreview | Arnaud Morin proposed openstack/nova master: Only set device_id at port creation with a service token https://review.opendev.org/c/openstack/nova/+/861172 | |
| 22:16:20 | opendevreview | Arnaud Morin proposed openstack/nova master: Only set device_id at port creation with a service token https://review.opendev.org/c/openstack/nova/+/861172 | |
| 23:38:10 | opendevreview | Takashi Natsume proposed openstack/nova master: Update contributor guide for 2027.1 Indri https://review.opendev.org/c/openstack/nova/+/1008569 | |
| #openstack-nova - 2026-10-13 | |||
| 02:25:11 | opendevreview | Takashi Kajinami proposed openstack/nova master: Catch ValueError from from_image_ref in rescue https://review.opendev.org/c/openstack/nova/+/1000969 | |
| 03:39:42 | opendevreview | Rocky proposed openstack/nova master: Drop placement connections inherited across fork https://review.opendev.org/c/openstack/nova/+/1010106 | |
| 04:37:28 | opendevreview | taha hosseini proposed openstack/nova master: libvirt: assign virtio storage devices to guest IOThread https://review.opendev.org/c/openstack/nova/+/1008546 | |
| 06:02:33 | opendevreview | taha hosseini proposed openstack/nova master: libvirt: assign virtio storage devices to guest IOThread https://review.opendev.org/c/openstack/nova/+/1008546 | |
| 06:38:53 | opendevreview | Merged openstack/nova master: Convert [libvirt] nfs_mount_options to list https://review.opendev.org/c/openstack/nova/+/986258 | |
| 09:14:11 | opendevreview | Maximilian Stinsky proposed openstack/nova stable/2026.1: Remove experimental warning from UnifiedLimitsDriver https://review.opendev.org/c/openstack/nova/+/1010135 | |
| 09:14:39 | opendevreview | Maximilian Stinsky proposed openstack/nova stable/2025.2: Remove experimental warning from UnifiedLimitsDriver https://review.opendev.org/c/openstack/nova/+/1010136 | |
| 09:14:58 | opendevreview | Maximilian Stinsky proposed openstack/nova stable/2025.1: Remove experimental warning from UnifiedLimitsDriver https://review.opendev.org/c/openstack/nova/+/1010137 | |
| 11:17:32 | stephenfin | bauzas: gibi: Just another poke for https://review.opendev.org/c/openstack/nova/+/1009877/ It's tiny, and unblocks the SQLA bump in upper-constraints | |
| 11:17:53 | stephenfin | (ping ye specifically because EU tz) | |
| 11:22:18 | opendevreview | taha hosseini proposed openstack/nova master: libvirt: assign virtio storage devices to guest IOThread https://review.opendev.org/c/openstack/nova/+/1008546 | |
| 11:40:34 | opendevreview | Maximilian Stinsky proposed openstack/nova stable/2026.1: Fix KeyError: 'body' in os-volumes_boot API proxy https://review.opendev.org/c/openstack/nova/+/1010162 | |
| 12:04:49 | opendevreview | Moritz Tanner proposed openstack/nova master: Implement extend_volume for local devices https://review.opendev.org/c/openstack/nova/+/878763 | |
| 12:37:50 | opendevreview | Merged openstack/nova master: Fix KeyError: 'body' in os-volumes_boot API proxy https://review.opendev.org/c/openstack/nova/+/1009982 | |
| 12:50:46 | gibi | stephenfin: does sqlite :memory: still works in the new SQLAlchemy that start quoting ":"? | |
| 12:52:00 | gibi | I mean is this just a test assert issue or is it affect sqlite URL usage? | |
| 12:53:19 | stephenfin | it must since oslo.db tests it and has a tips job that runs against `main` of SQLA | |
| 12:53:59 | gibi | cool then | |
| 12:54:52 | stephenfin | we also don't use memory in production, and I'm not sure we use it in our tests (though if we do, that's more proof it does still work since the nova tests are passing in https://review.opendev.org/c/openstack/requirements/+/1008776 with that change) | |
| 12:55:07 | stephenfin | thanks for taking a look at that 🙏 | |
| 13:02:38 | dansmith | mhen: I think we should scrub my confusion from your ptg topic this morning.. I (obviously) pattern-matched that to be about encrypted images given our recent discussions and it would probably be good to avoid confusing others. okay with that? | |
| 13:03:22 | mhen | sure | |
| 13:04:00 | gibi | stephenfin: I think the default is :memory: if no url is provided. In nova at least in the funtional test we switched to file based sqlite due to threading complications | |
| 13:21:24 | opendevreview | Nathan proposed openstack/nova master: libvirt: Attach disk.config.rescue in stable device rescue https://review.opendev.org/c/openstack/nova/+/1009444 | |
| 13:42:53 | opendevreview | Ahmed Syed proposed openstack/nova master: Tolerate transient InstanceActionNotFound on instance actions https://review.opendev.org/c/openstack/nova/+/1010180 | |
| 13:49:06 | opendevreview | Ahmed Syed proposed openstack/nova master: Tolerate transient InstanceActionNotFound on instance actions https://review.opendev.org/c/openstack/nova/+/1010180 | |
| 13:54:15 | opendevreview | Ahmed Syed proposed openstack/nova master: Tolerate transient InstanceActionNotFound on instance actions https://review.opendev.org/c/openstack/nova/+/1010180 | |
| 13:55:41 | opendevreview | Ahmed Syed proposed openstack/nova master: Tolerate transient InstanceActionNotFound on instance actions https://review.opendev.org/c/openstack/nova/+/1010180 | |
| 14:34:35 | opendevreview | Maximilian Stinsky proposed openstack/nova stable/2026.2: Fix KeyError: 'body' in os-volumes_boot API proxy https://review.opendev.org/c/openstack/nova/+/1010193 | |
| 14:35:24 | opendevreview | Maximilian Stinsky proposed openstack/nova stable/2026.2: Remove experimental warning from UnifiedLimitsDriver https://review.opendev.org/c/openstack/nova/+/1010194 | |
| 14:36:41 | opendevreview | Merged openstack/nova master: tests: Remove DB URL render test https://review.opendev.org/c/openstack/nova/+/1009877 | |
| 14:41:27 | opendevreview | Balazs Gibizer proposed openstack/nova master: Reproduce cross cell move losing trusted certs https://review.opendev.org/c/openstack/nova/+/1010021 | |
| 14:41:27 | opendevreview | Balazs Gibizer proposed openstack/nova master: Reject shelve and cross-cell moves with trusted certs https://review.opendev.org/c/openstack/nova/+/1010197 | |
| 14:57:37 | opendevreview | Merged openstack/nova master: Replace use of datetime.timezone.UTC https://review.opendev.org/c/openstack/nova/+/989515 | |
| 15:14:58 | opendevreview | Gabin Lanore proposed openstack/nova master: libvirt: stop the event dispatch loops on cleanup https://review.opendev.org/c/openstack/nova/+/1005022 | |
| 15:14:59 | opendevreview | Gabin Lanore proposed openstack/nova master: Reproduce bug #2170719 https://review.opendev.org/c/openstack/nova/+/1010210 | |
| 15:18:49 | sean-k-mooney | gibi to answer a question you asked, there is a usecase for grouping the cybrog devices in a single swich even if the vm does not have a numa toplogy but if we were to make that change anyway we would detail it in the spec. on the nova side i think we would want to default to 1 pci switch per virutal numa node in the guest and then decied if cybrog devices go into there own or | |
| 15:18:51 | sean-k-mooney | get spread between them. again topic to be detiled in any future spec | |
| 15:23:08 | gibi | sean-k-mooney: ack, thanks. I feel like here are a big bunch of incoming complexity that I'm a bit reluctant to blindly accept as something nova needs to take. At least the livemigration side of this complexity feels similar to how nova started deciding if two hosts' cpu are compatible for live migration and then libvirt implemented that eventually and we still not using the better libvirt | |
| 15:23:14 | gibi | implementation properly ever since | |
| 16:00:46 | opendevreview | Balazs Gibizer proposed openstack/nova master: Reproduce trusted cert bugs in cross-cell move and shelve https://review.opendev.org/c/openstack/nova/+/1010021 | |
| 16:00:47 | opendevreview | Balazs Gibizer proposed openstack/nova master: Reject shelve and cross-cell moves with trusted certs https://review.opendev.org/c/openstack/nova/+/1010197 | |
| 16:32:03 | gmaan | gibi: sambork: somehow i missed the eventlet sessions, but read the etherpad. I replied to the gibi question about 'nova testing without installing eventlet', its hard to "not install" it in devstack based job but we can uninatll it explicitly https://etherpad.opendev.org/p/oct2026-ptg-eventlet-removal#L117 | |
| 16:33:46 | sambork | sounds reasonable, and easy to do | |
| 16:36:52 | opendevreview | YushoYamaguchi proposed openstack/nova-specs master: Configurable virtio-net multiqueue count via flavor and image https://review.opendev.org/c/openstack/nova-specs/+/1009977 | |
| 17:14:10 | gibi | gmaan: uninstalling it today will probably uninstall oslo.service as oslo.service hard depends on eventlet today.(or is pip dumber than that?) So we need to make the oslo.service -> eventlet dependency soft | |
| 17:17:19 | gmaan | gibi: ohk, i thought oslo.services are ready with no-eventlet code path but if not then yeah that is first thing to do | |
| 17:21:01 | gibi | they have two backend, we need their eventlet backend to exists | |
| 17:21:41 | gibi | but in threading mode we does not depend on the eventlet backend of oslo.service | |
| 17:21:54 | gibi | so we should be able to install oslo.service without the eventlet dependency | |