Wednesday, 2026-05-13

*** ykarel_ is now known as ykarel05:04
opendevreviewTakashi Kajinami proposed openstack/nova-specs master: libvirt: AMD SEV-SNP support  https://review.opendev.org/c/openstack/nova-specs/+/98337607:16
opendevreviewTakashi Kajinami proposed openstack/nova-specs master: libvirt: AMD SEV-SNP support  https://review.opendev.org/c/openstack/nova-specs/+/98337607:18
opendevreviewDong Ma proposed openstack/nova master: Keep cached VIF until host-side unplug succeeds  https://review.opendev.org/c/openstack/nova/+/98839007:45
opendevreviewDong Ma proposed openstack/nova master: Keep cached VIF until host-side unplug succeeds  https://review.opendev.org/c/openstack/nova/+/98839007:48
sean-k-mooney[m][gibi](https://matrix.to/#/@_oftc_gibi:matrix.org) i have a working kernel module that does not require a special vm and can emulate multiple PFs each with up to 7 VFs. the pfs and vfs are enumerable in lspci qnd libvirt, get assigned a seperate iommu group and emulate a simple tty loopback device that works in a cirros qemu instance. so with this we could fully test pci passthough and sriov 07:59
sean-k-mooney[m]im going to look at adding the kernel to our zuul instance and creating a new cyborg extra repo to pacgakege the module and provide a set of playbooks to enable building a minimal kernel with the module08:01
sean-k-mooney[m]nova alt config or nova next should ideally jusr be able to enable the pre playbook and it would handel the build and reboot into the 7.1 kernel. im also considering if a dkms module for ubuntu 26.04 would also be a valid apporch08:03
gibisean-k-mooney: sounds cool08:40
sean-k-mooney[m]i think i can make the devices livemigratable too in the future08:42
sean-k-mooney[m]to make the device fully work i needed to provide a vfio variant driver like is used for nvdia vgpus08:43
sean-k-mooney[m]you can pass the device to a vm with the standard vfio-pci08:43
sean-k-mooney[m]and it will be visible to the guest but reads and writes to bar0 dont work08:44
sean-k-mooney[m]so we can test managed true or false08:44
sean-k-mooney[m]bit false is needed for it to fully work in the guest08:44
sean-k-mooney[m]im sure i could emulate orther device thia way too now that i have a working framework08:45
fricklergibi: was the pyroute2 issue only sporadic? I'm wondering whether we should do more rechecks on the test change or whether we should actually to try the reqs bump now that the CI is all green09:11
sean-k-mooney[m][frickler](https://matrix.to/#/@_oftc_frickler:matrix.org) i belive yes it was nondetermistic09:30
sean-k-mooney[m]that why we were originally able to bump it becase it worked well enough for the tempest test to pass in the requirements repo09:32
fricklerok, I'll rebase my patch and run some more iterations10:19
gibifrickler: indeed it is intermittent. And only a subset of the occurence of the log "This event loop is already running" actually causing tempest failure11:35
gibifrickler: I think we have another complication. I read back the os-vif bug https://bugs.launchpad.net/os-vif/+bug/2109396 and when the above log can cause tempest failure is during testing of interface detach. We discussed yesterday that this is potentially an eventlet + asyncio issue and we only have nova-alt-configuration left running with eventlet mode we need that job to test interface detach 11:40
gibiotherwise we miss the potential issue with pyroute211:41
gibiUnfortunately I don't see such detach interface test result in the nova-alt-configuration in previous run During interface detach os-vif fails with:11:41
gibiUnfortunately I don't see such detach interface test result in the nova-alt-configuration in previous run https://0c34cad25d42d1a19e0a-af98ebf7fb9f80d504693b6d78e0e17e.ssl.cf5.rackcdn.com/openstack/8ba07fdc336b43b9855a6abd3ac25cd1/testr_results.html11:42
gibiso I think we lost test coverage when we moved our eventlet testing to nova-alt-configuration11:42
gibimaybe too much11:42
gibias an immediate fix we can add tempest.api.compute.servers.test_attach_interfaces.AttachInterfacesTestJSON to nova-alt-configuraton. But probably we should mirror the tests running in nova-next to nova-alt-configuration11:44
sean-k-mooneynova-alt-config was orgianly intened to test livemgirtion primarlly12:26
sean-k-mooneybut we coudl add in other test into it12:26
sean-k-mooneygibi: its relitivly fast today so expanding the testing more espcially in light of the eventlet role i tnow plays makes sense12:26
fricklerok, I'll wait for that testing to be added, then12:40
opendevreviewTakashi Kajinami proposed openstack/placement master: doc: Use systemctl in Ubuntu  https://review.opendev.org/c/openstack/placement/+/98845713:42
opendevreviewAnton Iacobaeus proposed openstack/nova-specs master: Intel TDX support in libvirt driver  https://review.opendev.org/c/openstack/nova-specs/+/97960813:44
opendevreviewTakashi Kajinami proposed openstack/placement master: Ubuntu: Use systemctl to restart services  https://review.opendev.org/c/openstack/placement/+/98845713:44
*** haleyb is now known as haleyb|away14:20
opendevreviewDong Ma proposed openstack/nova master: Keep cached VIF until host-side unplug succeeds  https://review.opendev.org/c/openstack/nova/+/98839014:36
UgglaI know you're busy and there’s no immediate rush, but if both of you could take a look at the patches I’ve submitted, that would be great. 982675: Add reproducer for bug #2111320: soft-deleted services in Instance.services | https://review.opendev.org/c/openstack/nova/+/982675 (Uggla) && 982676: Filter soft-deleted services from16:32
UgglaInstance.services ORM relationship | https://review.opendev.org/c/openstack/nova/+/982676 (Uggla)16:32
opendevreviewMerged openstack/placement master: Ubuntu: Use systemctl to restart services  https://review.opendev.org/c/openstack/placement/+/98845716:32
opendevreviewElod Illes proposed openstack/nova master: Do not log metadata proxy shared secret  https://review.opendev.org/c/openstack/nova/+/98850116:48
opendevreviewribaudr proposed openstack/nova master: Clean up stale comments in ironic virt driver  https://review.opendev.org/c/openstack/nova/+/98850417:01
opendevreviewClif Houck proposed openstack/nova master: perf(ironic): eliminate O(N²) ProviderTree deepcopy at startup (B6)  https://review.opendev.org/c/openstack/nova/+/98067617:09
opendevreviewmelanie witt proposed openstack/nova master: Fix swap disk creation skipped on NFS during cold migration  https://review.opendev.org/c/openstack/nova/+/98854723:21
*** hemna4 is now known as hemna23:53
*** erlon4 is now known as erlon23:57

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!