Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2026-10-07
15:17:37 Uggla Reminder: upstream triage in ~15mn. (https://meet.google.com/zjr-rxus-hzj)
15:37:25 gmaan gibi_: updated it as per comment, please check https://review.opendev.org/c/openstack/nova/+/1006351
15:55:27 gibi_ gmaan: thanks, I cannot promise I get to it this week. I have to drop for today and tomorrow I need to use my time to read the cross cell live migration spec, and firday I'm off. But I will get back to it hopefully next week
15:57:45 opendevreview Stephen Finucane proposed openstack/nova master: Remove test for local ServiceAuthTokenWrapper https://review.opendev.org/c/openstack/nova/+/1009297
15:57:57 stephenfin melwitt: ^ easier than I thought: we can straight up remove it
15:59:07 stephenfin sean-k-mooney: you probably want to look at that also since you reviewed https://review.opendev.org/c/openstack/nova/+/1002540. We need it to unblock the ksa bump in upper-constraints
15:59:13 melwitt ahhh lol ... that was me again too. geesh. ok, well at least that makes the removal straightforward
16:00:22 stephenfin https://i.kym-cdn.com/entries/icons/original/000/037/468/cover5.jpg
16:01:33 sean-k-mooney ah ok that makes sense
16:01:57 sean-k-mooney i guess the way we were mockign did not catch the removal of our local versoin
16:02:44 sean-k-mooney im kind of surperise that this is broken by the new release if im honest
16:03:14 sean-k-mooney did
16:03:16 sean-k-mooney user_auth.get_endpoint_data.assert_called_once_with(
16:03:18 sean-k-mooney mock.sentinel.session)
16:03:29 sean-k-mooney stop passing because we didnt need to call it to fill in the data
16:05:43 melwitt I think it's due to this change https://review.opendev.org/c/openstack/keystoneauth/+/1002219/1/keystoneauth1/service_token.py
16:07:34 melwitt although it seems that was released a while ago
16:07:49 sean-k-mooney mostly likely i jsut dont see how that broke the test
16:08:13 sean-k-mooney the asset behvior we vhae shoudl have been the saem regardess of it was our wrapper or not
16:11:16 melwitt I was thinking the above change added **kwargs to the call which made it no longer match the assert
16:12:01 sean-k-mooney oh no it might be but i think its becasue its now caullign self.user_auth
16:12:21 melwitt but hm yeah the previous one was passing **kwargs too. nevermind
16:13:12 sean-k-mooney so we are creating a mock and passign that as the user aut when we get the service tone auth plugin
16:13:29 sean-k-mooney so i woudl expect our mock to be assigned to self.user_auth
16:13:49 sean-k-mooney and for that call to still cause or test to pass
16:14:03 sean-k-mooney stephenfin: do you have a link to the failure
16:14:09 sean-k-mooney i want to see which assert actully failed
16:14:20 stephenfin https://zuul.opendev.org/t/openstack/build/381a96a08d294adcbb4407e7bef15914
16:14:29 stephenfin also linked from a comment on the change
16:14:43 stephenfin I didn't put it in the commit message since those links expire after ~30 days
16:14:58 sean-k-mooney oh i see
16:15:01 sean-k-mooney the argument changes
16:15:15 sean-k-mooney so its still being called but with diffenrt kwargs
16:15:58 stephenfin yep
16:16:25 stephenfin but as I've noted in the review, whether it's called or not is immaterial to Nova, at least in the context of a unit test
16:18:27 melwitt sean-k-mooney: ok I think it's from this change rather https://review.opendev.org/c/openstack/keystoneauth/+/1003778/3/keystoneauth1/service_token.py
16:18:57 stephenfin melwitt: beat me to it
16:19:17 melwitt \o/ did one thing right today. eventually.
16:19:24 sean-k-mooney yes i was lookign at the refactorign that stephtn did on top
16:19:35 sean-k-mooney ya so that makes sense
16:19:48 sean-k-mooney i just wanted to narrow down what change broke it as yours shoudl not have
16:20:09 melwitt Mystery Solved ™
16:20:28 sean-k-mooney so we can delete this or we coud jsut replace called_once_with with called_once
16:20:31 sean-k-mooney i htink the delete
16:20:33 sean-k-mooney is fine
16:20:40 melwitt yeah, ++ diligence
16:20:41 sean-k-mooney as this si really internal behvior of the lib
16:21:16 sean-k-mooney we do care that the endpoing data is there but if nova is the thing catching that in its unit test its already too late
16:30:58 opendevreview Clif Houck proposed openstack/nova stable/2025.2: Parallelize per-node resource updates https://review.opendev.org/c/openstack/nova/+/1005118
16:50:11 shermanm hey, I wanted to note that I opened a few related issues with the ironic driver, but I was looking for some feedback on how best to approach functional tests for the regression given that there wasn't an existing fixture. I took a shot at it, the first reproducer is here https://review.opendev.org/c/openstack/nova/+/1008334 , and I've been using topic
16:50:11 shermanm `ironic-orphaned-nodes`
16:51:13 shermanm I didn't want to take things out of WIP and start spamming the CI without asking first :)
16:52:35 sean-k-mooney i did start on a basic set of fixture for some limited ironic functionl testing but i think i abandoned them
16:53:22 sean-k-mooney https://review.opendev.org/c/openstack/nova/+/980641/6
16:54:07 sean-k-mooney shermanm: so im not sure if any of that woudl be useful for you
16:56:21 gmaan gibi_: no worry, i understand
17:03:43 tkajinam can we merge this classifier update ? https://review.opendev.org/c/openstack/nova/+/1007898
17:03:51 tkajinam I should have covered that in the previous job update...
17:04:03 sean-k-mooney for 3.14 sure
17:04:40 sean-k-mooney we already have tox jobs and if my match merged tempet
17:04:56 sean-k-mooney it will be enabeld by default in many of our tempest jobs shortly if its not alreay
17:05:24 tkajinam at least unit tests already runs on py314 and the job is voting
17:05:41 tkajinam and yeah we may get full functional test coverage once devstack jobs are moved to resolute
17:05:56 sean-k-mooney yep but we have had ubuntu 26.04 testing in devstack for several months now
17:06:21 sean-k-mooney and canonicall ship 2026.1 on python 3.14 as well so you knwo it work in reality
17:06:27 sean-k-mooney so we shoudl declar it as such
17:14:10 sean-k-mooney shermanm: you have made a good start on extingn supprot to ironic based on a quick glance
17:25:48 stephenfin tkajinam: I am hoping to automate those classifier changes soon enough https://review.opendev.org/c/openstack/releases/+/993557
17:25:52 stephenfin jfyi
17:26:02 stephenfin for now, I've sent that on its way
17:34:39 tkajinam stephenfin, yeah I saw that
17:35:08 tkajinam I was wondering if we can split it out to a separate file somehow to avoid running full CI... though I'm unsure if we can do that really for pyproject.toml
17:35:27 tkajinam it's a bit annoying that just adding a single line without functional impact triggers heavy ci
17:35:46 tkajinam though it might eventually mean that we should drop these classifiers, as you suggested in the past
18:13:31 opendevreview Merged openstack/nova master: Remove test for local ServiceAuthTokenWrapper https://review.opendev.org/c/openstack/nova/+/1009297
18:15:12 dansmith Uggla: cross-cell migration is scheduled for wednesday but gibi_ says he'll have competition for brainpower.. I think we need (100% of) gibi for those conversations, FYI
18:23:55 gmaan Uggla: added comment in ths bug and mark it invalid for nova and opened for openstack client https://bugs.launchpad.net/python-openstackclient/+bug/2060812
18:28:01 gmaan dansmith: if you have time, this is ready (stop hanging tests in cleanup of executor shutdown) which we chatted about in IRC a few weeks before https://review.opendev.org/c/openstack/nova/+/1006351
20:28:05 opendevreview Damian Dąbrowski proposed openstack/nova master: libvirt: refresh frontend QoS on volume extend https://review.opendev.org/c/openstack/nova/+/982733
#openstack-nova - 2026-10-08
02:09:33 opendevreview Ghanshyam Maan proposed openstack/nova master: Reduce the unnecessary executors shutdown logs in test jobs https://review.opendev.org/c/openstack/nova/+/1004160
02:59:54 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
07:35:31 Uggla dansmith, I’m aware gibi_ might not be fully available on Wednesday. Unfortunately, I couldn’t find a better slot that worked with everyone’s constraints.
07:53:19 ykarel Uggla, Hi can you check release patches for os-vif https://review.opendev.org/c/openstack/releases/+/1008445 and https://review.opendev.org/q/topic:%22os-vif-tap-ipv6-release%22
07:55:44 gibi_ Uggla: dansmith: I will join Wednesday, and I will do my best to preserve energy for that session but life is a bit more complicated that day :)
08:31:36 bauzas looks like the gate is boring https://review.opendev.org/c/openstack/nova/+/1005727/5
08:53:48 bauzas gibi: sambork_: remind me bells, do we have a current open bug where nova misses some Cinder volume-reimage due to a threading concern ?
08:54:11 bauzas I paid little attention to the threadpool issues we got recently, except for functional tests
08:55:13 sambork Sorry, I'm not sure about that
08:57:19 bauzas nw
09:01:07 cw0306-lee[m] "Hi! I want to suggest two..." <- > <@cw0306-lee:matrix.org> Hi! I want to suggest two specs for nova and placement.... (full message at <https://matrix.org/oftc/media/v1/media/download/AZTmW8CPC_jlduQwzBYF65WcAvjnEp26ZxieBB0RtaYAPgGB7vR-SLypK1muVrvpqSh6oNyeUNFNXUP3Wpjm5KlCehYte6EwAG1hdHJpeC5vcmcvYWdPZXR3ZHdYWXpScGdYR2VnUkFpdEx0>)
09:01:56 gibi bauzas: I'm not aware of that
09:02:02 bauzas ack
09:02:23 bauzas my first triage of the last failure was incorrect, I'll add a better gerrit comment
09:03:06 bauzas good news is that I failed three times with three different causes, none of them being related to threading apparently :)
09:03:32 gibi bauzas: you might remember this https://review.opendev.org/c/openstack/nova/+/966446 that prevented the libvirt events to arrive. but your last failure was due to slow re-image and nova timing out waiting for the event from cinder not from libvirt
09:04:01 gibi btw re-image succeeded but after novas timeout
09:08:40 bauzas yeah that's what I figured
09:09:01 bauzas cinder took a while to reimage, nova timed out
10:33:00 frickler fyi just saw a gate failure with two rebuild tests in reqs gate, didn't look into the details, but might be related? https://zuul.opendev.org/t/openstack/build/69448ba8006646e2857f42c6449434e3
11:21:18 gibi frickler: at least the symptomps are different, in your case the ssh to the guest failed after nova declared the rebuild successfull, in bauzas case nova failed to finish the rebuild

Earlier   Later