Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2025-06-26
12:12:46 sahid have you already suffered issue like Live Migration failure: <mac address>
12:13:15 sahid I'm not able to find what could be the issue and unfortunatly the exception is logged like that
12:13:48 sahid except Exception as e:
12:13:48 sahid with excutils.save_and_reraise_exception():
12:13:48 sahid LOG.error("Live Migration failure: %s", e, instance=instance)
12:14:05 sahid it s in live migration operation
12:28:08 opendevreview Balazs Gibizer proposed openstack/nova master: Remove unnecessary test module level code https://review.opendev.org/c/openstack/nova/+/953436
12:47:33 opendevreview Balazs Gibizer proposed openstack/nova master: Allow to start unit test without eventlet https://review.opendev.org/c/openstack/nova/+/953436
12:56:58 opendevreview Balazs Gibizer proposed openstack/nova master: Allow to start unit test without eventlet https://review.opendev.org/c/openstack/nova/+/953436
12:58:58 stephenfin gmaan: Got a follow-up for you to re-approve here, when you've time https://review.opendev.org/c/openstack/nova/+/952213/
13:10:14 stephenfin sean-k-mooney: Oh, I also addressed your comment on https://review.opendev.org/c/openstack/nova/+/952281/ here https://review.opendev.org/c/openstack/nova/+/953281/
13:10:28 stephenfin and thanks for the review 🙏
13:26:18 sean-k-mooney stephenfin: cool +2w on teh organl patch and +2 on the fixup patch
13:28:12 sean-k-mooney sahid: was it for an sriov port? i know that there have been issues with sriov ports adn attach/detach if there are more then one port with the same mac. havign a vm with multipel prot with the same mac is sort of unsupproted.
13:28:15 sean-k-mooney its not
13:28:37 sean-k-mooney but it was never expiclty added as a feature and kind fo bolted on as bugs were reported
13:28:56 sean-k-mooney it woudl never happen with nova networks and it cant happen in a signel neutron network
13:29:43 sean-k-mooney so nova never really blocked it. over time as neturon becausem the default we got bug reports that usign the mac to attach/dtach the prots failed fi you had a vm with 2 prots htat had the same mac
13:30:19 sean-k-mooney sahid: we coudl also hith that internally when live migrating vms with sriov vnic-type=direct ports
13:30:47 sean-k-mooney sahid: that is the only mac related live migration failure that i recall
13:31:52 opendevreview Merged openstack/nova master: api: Add response body schemas for server topology API https://review.opendev.org/c/openstack/nova/+/940300
13:37:23 sahid sean-k-mooney: thanks for this, we don't use srio port, but it's more or less realted to vms AVI loadbalancer
13:53:26 sahid it's strange, when i'm doing a port list for given vm that is buggy, there is 2 ports attached, and when looking at the xml, 5 interfaces are indicated
13:53:59 sahid sean-k-mooney: have you already noticed something like that, where xml is not well cleaned?
15:03:35 opendevreview Pierre Riteau proposed openstack/nova stable/2024.2: Add repoducer test for bug 2074219 https://review.opendev.org/c/openstack/nova/+/953468
15:35:44 opendevreview Pierre Riteau proposed openstack/nova stable/2024.2: Fix detaching devices by alias with mdevs https://review.opendev.org/c/openstack/nova/+/953474
15:40:04 opendevreview Balazs Gibizer proposed openstack/nova master: Run unit test with threading mode https://review.opendev.org/c/openstack/nova/+/953475
16:44:53 gmaan stephenfin: one question about doc generation, if we allow rdp in schema then how we reflect that it is gone in latest version https://review.opendev.org/c/openstack/nova/+/952213/comment/e644f810_7e7b50c2/
17:24:53 stephenfin gmaan: that depends on whether you're talking about the API perspective (i.e. clients) or the docs (i.e. humans/AI scrapers)
17:25:47 stephenfin from the API/client perspective, we're already handling that by returning HTTP 400 (Bad Request) if you attempt to use it on a newer version of Nova (or an earlier version not using HyperV) so nothing needs to change there, IMO
17:27:03 stephenfin gmaan: From the latter perspective, we'd likely handle that by either (a) indicating it in the 'description' text or (b) using a custom property that os-api-ref or its replacement can pick up
17:29:29 gmaan stephenfin: yeah, we should update the description saying it is gone but my main concern in this change is how to indicate that now via response schema
17:30:42 stephenfin gmaan: beyond what we're doing here? https://review.opendev.org/c/openstack/nova/+/952213/4/nova/api/openstack/compute/remote_consoles.py#161
17:31:38 stephenfin With this patch, we're moving validation (and rejection) from the schema into the method itself. The end result is the same though.
17:31:52 gmaan I am thinking either we need to version (add different schema for each releases if changed) schema based on releases like we do for microversions by adding the info in current api-ref description
17:32:35 gmaan stephenfin: code behavior wise it is all good that is why i am +1. I am thinking what is usage of showing it in schema and how to handle the case where we can say it is gone after caracal
17:33:06 stephenfin So to be clear, you're worried about what the OpenAPI-derived replacement for api-ref will look like?
17:33:13 gmaan exactly
17:33:44 gmaan I do not want new OpenAPI-derived replacement show rdp as valid value
17:33:52 gmaan if it consume response schema as it is
17:34:19 stephenfin Well it is valid *if* you're using pre-Caracal and Hyper-V
17:34:46 stephenfin So we'll still document it, but in the description we'll say "rdp doesn't work since Caracal etc. etc.
17:34:57 stephenfin just like we do currently in the api-ref
17:36:49 stephenfin All of these schemas - query, request body, and response body - are going to have additional metadata layered on top in terms of the description field. We're not doing that yet because these reviews are already numerous and difficult enough, but the future api-ref will have to have all the same context the current one does in order to be useful
17:38:45 gmaan yeah, if openAPI will use the same content of field description and use schema as example show then we should update about rdp field here https://github.com/openstack/nova/blob/master/api-ref/source/parameters.yaml#L5986-L5999
17:39:21 gmaan so that if your change showing it in schema then updating current description about rdp should be done
17:40:05 gmaan and if new content for field description are generated ( from schema or so) by openapi then we need to manually review it and your comment in schema works fine
17:42:03 gmaan stephenfin: basically my request is to update the current api-ref description also if you are adding it in schema
17:42:10 stephenfin oh!
17:42:15 stephenfin I agree. I can do that
17:42:25 gmaan cool
21:28:37 opendevreview Stephen Finucane proposed openstack/nova master: api: Address issues with remote consoles APIs https://review.opendev.org/c/openstack/nova/+/952213
23:36:02 mikal sean-k-mooney: Just so you know, I've seen your review comments about release notes. I wrote some, but in the process noticed that the extra specs aren't working quite how I expected so I am now debugging those.
#openstack-nova - 2025-06-27
00:39:31 sean-k-mooney mikal: ack. im going ot sleep now but ill take another look tomrow or next week
00:49:02 opendevreview Cyril Roelandt proposed openstack/os-vif master: Zuul: do not use USE_PYTHON3 https://review.opendev.org/c/openstack/os-vif/+/953510
03:37:30 mikal sean-k-mooney: no worries. The good news it was me forgetting you need a very recent libvirt / qemu for virtio sound to actually work.
03:55:03 opendevreview Michael Still proposed openstack/nova master: libvirt: Add objects and notifications for sound model. https://review.opendev.org/c/openstack/nova/+/926126
03:55:04 opendevreview Michael Still proposed openstack/nova master: Implement sound model extra spec for libvirt. https://review.opendev.org/c/openstack/nova/+/940770
03:55:04 opendevreview Michael Still proposed openstack/nova master: libvirt: Add objects and notifications for USB controller model. https://review.opendev.org/c/openstack/nova/+/927354
03:55:05 opendevreview Michael Still proposed openstack/nova master: Implement USB controller extra spec for libvirt. https://review.opendev.org/c/openstack/nova/+/950643
07:11:13 gibi sean-k-mooney: when you are up, dansmith is +2 on https://review.opendev.org/c/openstack/nova/+/952966 can I get a +A?
07:11:21 gibi this is the py3.13 fix
07:19:41 opendevreview Kamil Sambor proposed openstack/nova master: Replace eventlet.event.Event with threading.Event https://review.opendev.org/c/openstack/nova/+/949754
07:25:42 gibi is it just me or CI is mosty shit due to etcd fails to start with 'listen tcp 65.17.193.12:2379: bind: cannot assign requested address'
07:26:41 gibi https://bugs.launchpad.net/devstack/+bug/2115338
07:29:09 gibi maybe it is already fixed ^^:
07:32:35 gibi sean-k-mooney: also if you can look back to https://review.opendev.org/c/openstack/nova/+/953121 Dan is +2 as well. This is the build_and_run_instance error handling patch
09:08:20 priteau Hello. One of our customer experienced this bug: https://bugs.launchpad.net/nova/+bug/2074219. They are on Caracal. It made me realise the fix had not been backported from Epoxy. I submitted backports: https://review.opendev.org/c/openstack/nova/+/953468 and https://review.opendev.org/c/openstack/nova/+/953474
09:08:37 priteau Backports for 2024.2 first, I will submit 2024.1 once approved.
09:17:47 gibi priteau: thanks for the backports I'm +2. Could you change the gerrit topic to bug/2074219 for better tracking?
09:18:06 gibi elodilles: ^^ some easy backport reviews :)
09:21:54 priteau Thanks, done
09:42:03 elodilles gibi: ACK, will review it today when I get there o/
09:54:20 opendevreview Balazs Gibizer proposed openstack/nova master: Run unit test with threading mode https://review.opendev.org/c/openstack/nova/+/953475
11:42:41 sean-k-mooney gibi: done for the 3.13 patch ill look at the other one next
11:43:22 gibi sean-k-mooney: thanks!
11:43:46 sean-k-mooney gibi: i should know this but its been a while. in my devstack env what is the simplest way to run tempest?
11:44:13 sean-k-mooney do you happen to knwo of the top of your head. i knwo that it preconfigures it for you and you can use tox
11:44:27 sean-k-mooney is that the best way or just using the tempest command in one of the venvs?
11:44:43 sean-k-mooney i know i have done both before
11:45:11 gibi cd /opt/stack/tempest/
11:45:23 gibi source .tox/tempest/bin/activate
11:45:24 gibi tempest run
11:45:45 sean-k-mooney ack thanks i was not sure if that was still thge correct thing or not
11:45:52 gibi that works for me
11:46:24 sean-k-mooney ya i know that there used to be some script for it before they moved things to tox envs for the jobs and i knwo it has the idea of "workspaces" or whatever
11:46:39 sean-k-mooney so i was not sure if i need to cd to a specific dir or something beyond the root of the repo
11:46:39 gibi I never dig into workspaces
11:47:02 sean-k-mooney cool ill give that a go and if i have issue ill just ask gmann later
11:47:13 gibi btw the majority of the unit tests now run with threading in zuul https://review.opendev.org/c/openstack/nova/+/953475 I will work on the exclude list now
11:47:22 gibi dansmith: ^^ fyu
11:47:22 sean-k-mooney +2w on the rpc decorator change too
11:47:27 gibi thanks
11:48:17 sean-k-mooney oh cool sodo you need to set NOVA_RAISE_ON_GREENLET_LEAK=True again explcitly
11:48:32 sean-k-mooney im fine with doing that but shoul that not be inherited form {[testenv]setenv}
11:48:36 gibi I'm not sure about that any more I will check it
11:48:46 sean-k-mooney oh
11:48:49 gibi I'm not sure the env list is merged
11:48:51 sean-k-mooney sorry its false by default
11:48:59 sean-k-mooney so we need to turn it on in teh unit tests

Earlier   Later