| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2026-05-04 | |||
| 16:08:45 | Uggla | #info I have updated Launchpad and the above doc. Please ping me if you spot something missing. | |
| 16:08:51 | Uggla | #info I have added all the BP/specs that are proposed for 2026.2 cycle. (23 BP/Spec proposed so far) | |
| 16:09:04 | Uggla | #info Starting: https://etherpad.opendev.org/p/nova-2026.2-status#L16 interesting bugs to review. | |
| 16:09:18 | Uggla | Proposals for changes to nova-core https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/EI44ZQG6UHIBNH3O26KHUVHIE576SCOC | |
| 16:09:31 | Uggla | ^ cores please don't forget to vote | |
| 16:11:12 | Uggla | So lots of stuff to review. Clearly, I doubt we will manage to deal with at least half of the BP. | |
| 16:11:37 | Uggla | #topic Stable Branches | |
| 16:11:49 | elodilles | thanks | |
| 16:11:56 | elodilles | #info 2024.2 Dalmatian moved to EOL last week ( https://review.opendev.org/c/openstack/releases/+/984964 ) | |
| 16:12:04 | Uggla | \o/ | |
| 16:12:09 | elodilles | #info see final state of 2024.2 Dalmatian under tag '2024.2-eol' | |
| 16:12:16 | elodilles | and | |
| 16:12:19 | elodilles | #info please do not backport patches to stable/2024.2 anymore (branch will be deleted soon) | |
| 16:12:52 | elodilles | so now we have only 3 stable branches again: stable/2026.1, stable/2025.2, stable/2025.1 | |
| 16:13:00 | elodilles | #info placement's stable/2025.1 gate is blocked, workaround: https://review.opendev.org/c/openstack/placement/+/980761 | |
| 16:13:05 | elodilles | #info rest of the stable gates should be OK | |
| 16:13:17 | elodilles | at least i'm not aware of any other gate issue | |
| 16:13:24 | elodilles | #info stable branch status / gate failures tracking etherpad: https://etherpad.opendev.org/p/nova-stable-branch-ci | |
| 16:13:35 | elodilles | and that's all from me about stable | |
| 16:13:55 | Uggla | thx elodilles | |
| 16:13:59 | elodilles | np | |
| 16:14:13 | Uggla | #topic vmwareapi 3rd-party CI efforts Highlights | |
| 16:14:22 | Uggla | fwiesel anything for us this week ? | |
| 16:15:23 | fwiesel | Hi, there was a misconfiguration from my side. I fix that now, and the pipeline should work again | |
| 16:15:42 | fwiesel | I am also looking into speeding it up a bit, as the backlog became quite large. | |
| 16:15:45 | fwiesel | That is it from my side. | |
| 16:15:56 | Uggla | thx fwiesel | |
| 16:16:21 | Uggla | #topic Gibi's news about eventlet removal | |
| 16:16:24 | gibi | o/ | |
| 16:16:28 | Uggla | gibi, sambork anything for us ? | |
| 16:16:31 | gibi | sambork: do you want to take it? | |
| 16:17:50 | gibi | so | |
| 16:17:50 | gibi | no worries I can take it today | |
| 16:18:18 | sambork | I can update partially (still don't take it fully), so we finally support thread mode in all cli and service last week patch testing cli was merged | |
| 16:18:40 | Uggla | \o/ | |
| 16:20:55 | sambork | I also investigate issue that sean found with python 3.14 and eventlet so reason why we see pickling error in novncproxy tests is monkey patching __socket = __import__('socket') which cause error when websockify call spawn | |
| 16:21:15 | sambork | I think gibi have more update with poisoning of eventlet and tests | |
| 16:22:10 | gibi | on the last one I think we monkey patch socket in other services too, so it is stringe why only cause this pickling issue the console proxy. | |
| 16:22:33 | gibi | I know we explicitly fork there while oslo.service does the fork for the rest of the services | |
| 16:22:58 | gibi | so maybe when oslo.service forks there is a fallback of a fix for this pickling issues | |
| 16:23:43 | gibi | anyhow we can look at it in more details | |
| 16:23:55 | sambork | ack, will do then. Thanks! | |
| 16:23:56 | gmaan | yeah, if not picklable then fallback is 'fork' | |
| 16:23:58 | gibi | I guess the goal is still to make py3.14 work in eventlet mode too | |
| 16:24:27 | gibi | gmaan: yeah I mean we do spawn and fallback to fork | |
| 16:24:36 | gmaan | yeah | |
| 16:25:30 | gibi | anyhow on my side I have a patch poisoning the eventlet import when we start in native threading mode and it should be green now as we fixed oslo.privsep, the last outstanding code that imports eventlet unconditionally | |
| 16:25:42 | gibi | https://review.opendev.org/c/openstack/nova/+/986282?usp=search | |
| 16:26:11 | gibi | (CI is stil running to prove me wrong) | |
| 16:26:58 | gibi | and Ashish has two patches in the functional test series that can be reviewed | |
| 16:27:09 | gibi | stating here https://review.opendev.org/c/openstack/nova/+/979850/8 | |
| 16:27:17 | gibi | that is it from me. | |
| 16:27:38 | Uggla | thx gibi and sambork | |
| 16:27:50 | Uggla | skipping Layos topic because he is not available this week. | |
| 16:28:13 | Uggla | So new topic. | |
| 16:28:15 | Uggla | #topic Ironic review priorities | |
| 16:28:51 | cardoe | We've got a couple of patches around the performance that sean-k-mooney has been working on. | |
| 16:29:26 | cardoe | I'll let sean-k-mooney speak to those but there's also another one https://review.opendev.org/c/openstack/nova/+/985785 | |
| 16:30:49 | cardoe | We continue to see complaints about "Build of instance $UUID aborted: Failure prepping block device." as an error message. This is unique to Ironic because the network devices are setup in a different spot than the other drivers because of block devices over network. In the other driver paths there is a list of exceptions that's allowed to bubble up to the user which is not matched in this path. | |
| 16:30:58 | cardoe | https://opendev.org/openstack/nova/src/commit/fa64fa82bf22c7d1c947268ed388e4d428e8d408/nova/compute/manager.py#L3060 is the call site. | |
| 16:31:13 | cardoe | right now there's just one outstanding backport version at https://review.opendev.org/c/openstack/nova/+/986674 | |
| 16:31:33 | gibi | cardoe: https://review.opendev.org/c/openstack/nova/+/985785 seems very clean and only impact ironic. So if we could get an ironic dev to confirm that this is good for ironic I have no problem to +2 it from nova perspective | |
| 16:32:38 | cardoe | +1 from me | |
| 16:33:01 | cardoe | On the BP list we've got https://review.opendev.org/c/openstack/nova-specs/+/471815 for trunk support. | |
| 16:33:37 | cardoe | I do think for the "Failure prepping block device" issues that maybe we need a different approach in the driver. Like only trying to setup the networks for storage connections there and not all the networks. | |
| 16:33:51 | cardoe | Or mirroring the same exceptions that can bubble up to the user in both places. | |
| 16:34:06 | cardoe | Otherwise that's all I've got. | |
| 16:35:07 | Uggla | gibi, if I understand well Sean propose to use 955685 instead of mine 985785 | |
| 16:35:10 | gibi | cardoe: Uggla: can we resolve the conflict of who proposed the fix first for the num_instance thing? | |
| 16:35:30 | Uggla | So I was about to abandon it. | |
| 16:35:43 | gibi | fine by me, I cna review the earlier patch | |
| 16:35:49 | gibi | *can | |
| 16:36:24 | cardoe | Both have my +1. | |
| 16:36:28 | gibi | OK | |
| 16:38:04 | cardoe | We can take uggla's testing and squish it into the original author. Or whichever approach you'd like. I'll do the lift. | |
| 16:38:16 | Uggla | reading Sean I'm not sure which one he prefers. So I keep it open | |
| 16:39:26 | Uggla | cardoe, you also have this spec right: 471815: Expose vlan trunking in metadata/configdrive | https://review.opendev.org/c/openstack/nova-specs/+/471815 | |
| 16:39:33 | cardoe | yes | |
| 16:39:58 | cardoe | Happy to make updates if that's desired. Most of the feedback is around other parts of the existing spec that aren't necessarily clear. | |
| 16:42:04 | Uggla | cardoe what's the top priority for you atm ? | |
| 16:43:04 | cardoe | I think right now just figuring out how ironic folks can best work with the nova folks. I don't wanna make it harder for you folks to review or accept patches. So if we should be working on testing improvements we can do that. | |
| 16:43:33 | cardoe | From the operators I've asked the "Failure prepping block device" is one that's pretty annoying for them to handle. | |
| 16:46:12 | Uggla | ok, I think it helps to focus. | |
| 16:46:30 | Uggla | Am i wrong or there is not patch regarding "Failure prepping block device" yet ? | |
| 16:46:58 | cardoe | There's just a backport adding another exception into the list of allowed to be seen by the user. | |
| 16:47:35 | cardoe | So what I'm wondering is if it would be better to make a list of those that should be seen by the user and keep things in sync that way or if I need to look at approaching that call site differently. | |
| 16:47:45 | cardoe | It's a change to nova/compute/manager.py either way. | |
| 16:47:54 | Uggla | which is 986674: ironic: Abort build on fatal image validation errors | https://review.opendev.org/c/openstack/nova/+/986674 right ? | |
| 16:48:05 | cardoe | Yes | |
| 16:48:26 | cardoe | ah no | |
| 16:48:27 | cardoe | Sorry | |
| 16:48:38 | cardoe | That's to not try to build over and over if the image is bad. | |
| 16:49:09 | cardoe | I linked the wrong review to this. | |
| 16:49:36 | Uggla | anyway the above patch is in the pipe now thanks to gibi. | |
| 16:49:40 | cardoe | Thank you. | |
| 16:49:53 | cardoe | I'll come back with better details next week. I've held you all up long enough. | |
| 16:50:13 | Uggla | thanks cardoe. | |
| 16:50:23 | Uggla | moving on next topic | |
| 16:50:25 | Uggla | #topic Bug scrubbing | |
| 16:50:30 | Uggla | #info up to 191 (-9) | |