Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2026-07-13
09:11:48 elodilles so the command is just copies over the deliverables details, as is
09:12:23 sean-k-mooney[m] ya its more a case fo we are not grate at going in and updating the metadata when we change how we work wiht things
09:12:33 sean-k-mooney[m] like reconfiguring placment to not list storyborad
09:14:09 sean-k-mooney[m] we move that back to launchpad when it was taken back under the nova umbrella years ago
09:15:06 elodilles sean-k-mooney[m]: about changing release model: independent release model has its disadvantage as not having stable branches, which means there is no possibility to do stable releases, which is painful for example in case python version supports are dropped in the meanwhile, etc.
09:15:54 sean-k-mooney[m] elodilles: well for python-novaclient we have frozen the shell and pyhton binding
09:16:16 sean-k-mooney[m] so we were not ment to have new release each cycle in general
09:16:40 sean-k-mooney[m] we already have feature partity in osc and we only support new microverions in the sdk
09:17:51 elodilles sean-k-mooney[m]: in theory it is OK, but i'm sure there would be a point when an 'urgent stable release' would be needed o:) (been there with oslo ;))
09:18:17 elodilles sorry, i need to go AFK for a while now
09:18:23 sean-k-mooney[m] perhas although it will hopefully get returied in the next couple of release
09:18:34 sean-k-mooney[m] elodilles: no worries o/
09:47:07 opendevreview Artem Vasilyev proposed openstack/nova master: Fix race between resize and update_available_resource for dedicated CPUs https://review.opendev.org/c/openstack/nova/+/984048
09:47:25 opendevreview Artem Vasilyev proposed openstack/nova master: Add reproducer for bug #2147776 https://review.opendev.org/c/openstack/nova/+/984007
11:30:43 opendevreview Bhavana proposed openstack/nova master: Add custom GPU PCI weigher for scheduler optimization Add Minimal Displacement Solver for MIG GPU Repartitioning https://review.opendev.org/c/openstack/nova/+/996058
11:49:55 tkajinam could anyone review and merge https://review.opendev.org/c/openstack/nova/+/996880 ?
11:50:05 tkajinam ^^^ doc update after migration of cursive to oslo's namespace
12:11:03 gibi sean-k-mooney[m]: bauzas: ( elodilles): 16.0.0b1 placement is released and our tox targest are using it as they are asking for >= 9.0.0b1 and that triggers beta version usage in pip
12:11:29 bauzas ?
12:11:56 gibi bauzas: just YI so that the placement release went out and it works in our test env without further version bump
12:12:09 bauzas ah ok
12:12:15 bauzas so no issue
12:12:18 gibi no issue
12:13:56 sean-k-mooney gibi: what did we actully need that for by the way i missed that this morning
12:14:09 sean-k-mooney was it for the change to the fuctional fixture
12:14:20 sean-k-mooney to allow the db to use files
12:14:31 sean-k-mooney to unblock threaded mode ?
12:14:47 gibi yes it was fixture change on the placement side to be able to pass connection URL for the DB fixture in placement
12:14:53 sean-k-mooney ack
12:15:45 sean-k-mooney tox-sibling woudl have enabeld required-project to use master or any pinned checkout but unless we wanted to do that permently
12:15:49 sean-k-mooney doing the release is cleaner
12:16:05 sean-k-mooney the factg we dont use tox-siblings is why depend on only works for devstack today
12:16:20 gibi I did not find an example of tox-siblings and release was faster :)
12:16:24 elodilles sean-k-mooney[m]: is it possible that this patch needs to be backported to all stable branches? https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/955714
12:17:08 elodilles sean-k-mooney[m]: as this test case is failing on nova's stable/2025.1 nova-ceph-multistore job: https://review.opendev.org/c/openstack/nova/+/995958
12:17:14 sean-k-mooney no it can be but its not required
12:17:23 elodilles (tempest is branchless)
12:17:36 sean-k-mooney there is a better change to backport i think
12:17:56 sean-k-mooney well so we have a "fix" for nova
12:18:04 sean-k-mooney backporting that may be better
12:18:39 sean-k-mooney the fix is adding a config option for current hardcode value so in addtion to that we woudl ahve to tweak the job to set that
12:19:10 sean-k-mooney elodilles: https://review.opendev.org/c/openstack/nova/+/969538
12:19:15 sean-k-mooney so if we backport that
12:19:40 sean-k-mooney oh mel set the defautl to 2
12:19:42 sean-k-mooney images_address_space_limit
12:19:52 sean-k-mooney so that should jsut work
12:20:06 sean-k-mooney sicne she alrady doubled the hardcoded value for the new config default
12:20:38 opendevreview sean mooney proposed openstack/nova stable/2026.1: Make QEMU_IMG_LIMITS process limits configurable https://review.opendev.org/c/openstack/nova/+/997027
12:22:51 opendevreview sean mooney proposed openstack/nova stable/2025.2: Make QEMU_IMG_LIMITS process limits configurable https://review.opendev.org/c/openstack/nova/+/997028
12:23:14 opendevreview sean mooney proposed openstack/nova stable/2025.1: Make QEMU_IMG_LIMITS process limits configurable https://review.opendev.org/c/openstack/nova/+/997029
12:23:24 sean-k-mooney elodilles: so ^ in theory should work
12:23:39 elodilles sean-k-mooney: so to be clear, this patch is needed to make this test case stable: test_extend_attached_encrypted_volume_luksv1
12:23:42 elodilles right?
12:24:35 sean-k-mooney yes
12:24:59 elodilles ACK, sorry, i'm a bit slow today double checking things o:)
12:25:07 sean-k-mooney the luks hearder make 1g cinder voluem need more the 1G of aresss space for qemu img to map and or process them
12:25:33 sean-k-mooney the patch set the default for the new config option to 2g
12:25:39 elodilles though i'm not really fond of changing config value on stable branches only to make test cases pass :/
12:26:11 sean-k-mooney well the other option is no luks on ubuntu 24.04
12:26:26 sean-k-mooney because without the patch its hardcoded to 1
12:26:47 elodilles yes, that's what i'm saying o:)
12:26:55 elodilles s/what/why
12:27:10 sean-k-mooney so this was really cause by a chagne in qemu
12:27:33 sean-k-mooney and how qemu-img internally manges its concurancy
12:27:52 elodilles so we need 2G to make the test pass, but that changes the value for already deployed environments from 1G to 2G if i understand correctly
12:27:55 sean-k-mooney the current change mimics what cinder and ironic do
12:28:38 sean-k-mooney elodilles: so i think the real long tim fix is to remove the adress space based limit and move to RSS (resedet set size) or drop the limit entrily
12:29:08 sean-k-mooney it exits to prvent patalogical behvior in qemu-img form triggerign the OOM killer
12:30:13 elodilles i see
12:30:50 sean-k-mooney which form a defence in dept pont of view i get but adressspace was never the right thing to limit on in my view RSS was but even then we dont sandbox all the other clis we interact with so it feels kind of arbiaty that we did limit qemu-img
12:33:28 elodilles sean-k-mooney: still, i'm a bit hesitant to backport this patch. i'd perhaps tempest-regex-exclude for this single test case on stable branches (as this test case was disabled before the revert has landed on tempest)
12:34:09 sean-k-mooney well this test used to be sable on the stable branchs
12:34:20 sean-k-mooney it was broken by a qemu release
12:34:51 sean-k-mooney so sure we can disbale it but i think we expect to backport the bugfix regardless fo the ci
12:35:14 sean-k-mooney this sued to work on stable and it was regressed by a ubuntu package release
12:35:23 elodilles ah, i see
12:35:25 sean-k-mooney so we want to be abel to suprpot existing users
12:37:12 sean-k-mooney elodilles: this affect a very specific subset of deployemtn because it depens on teh qemu+librados/ceph clinet release in use. and you have to have encypted volumes as well.
12:37:37 sean-k-mooney anyway lets get some ci results adn see if the backprot even works
12:38:43 sean-k-mooney elodilles: we could also backport swaping the ceph jobs to debian 12 but chaning the disto seams more invasive to me
12:39:07 sean-k-mooney althouhg it would only be on 2025.1
12:39:25 sean-k-mooney 2025.2+ already were using debian for ceph
12:40:52 sean-k-mooney elodilles: https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/997035 we can proably dicusss that in the qa channel seperatly
12:44:00 elodilles sean-k-mooney: thanks for the explanation. i've commented on the patch my concern o:)
12:45:47 sean-k-mooney we could set it to the same value in the backport and tweak it in the job or as you said just exclude the test form the relevent jobs
12:48:36 elodilles i'm not fully against backporting the patch as is, if stable cores agree to do so o:) but maybe the above options are less invasive ^^^
13:12:50 opendevreview Joan Gilabert proposed openstack/nova master: Add libvirt driver MDEV ARQ support for Cyborg https://review.opendev.org/c/openstack/nova/+/995840
13:33:48 opendevreview Merged openstack/nova master: doc: Update cursive repository path https://review.opendev.org/c/openstack/nova/+/996880
15:27:29 gmaan gibi: thanks for placement release, I had to do same workaround of testing nova cross jobs with placement master when enforce_scope removal caused the issue. I tired to release it but then dropped the idea to let it releases during its own time.
15:27:46 gmaan gibi: on 2160224, I will check today bug and the change.
15:35:37 opendevreview Joan Gilabert proposed openstack/nova master: Add libvirt driver MDEV ARQ support for Cyborg https://review.opendev.org/c/openstack/nova/+/995840
15:35:37 opendevreview Joan Gilabert proposed openstack/nova master: Report OWNER_NOVA trait on resource providers https://review.opendev.org/c/openstack/nova/+/994299
15:35:56 gibi gmaan: thanks!
15:36:07 gibi it turned out that releasing a beta was easy
15:39:01 gibi FYI nova meeting starts in 20 mintes here. (I will be your guide today)
16:00:49 opendevmeet The meeting name has been set to 'nova'
16:00:49 opendevmeet Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:49 opendevmeet Meeting started Mon Jul 13 16:00:49 2026 UTC and is due to finish in 60 minutes. The chair is gibi. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:49 gibi #startmeeting nova
16:00:59 gibi o/

Earlier   Later