| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-10-05 | |||
| 12:44:07 | bauzas | ping me if you want :p | |
| 12:45:14 | bauzas | gibi: any urgent queries for me ? | |
| 12:46:58 | gibi | bauzas: hi. Nothing in my mind for Victoria. I've started reading open specs like https://review.opendev.org/#/c/742785/ | |
| 12:47:13 | bauzas | cool, I'll then repropose the routed-networks one | |
| 12:47:31 | bauzas | and I'll start looking at open specs | |
| 12:48:41 | gibi | i've proposed my own spec for W https://review.opendev.org/#/c/755477/ | |
| 12:51:44 | bauzas | my review ratio is super low, looks good a good start for improving my stats :) | |
| 13:04:42 | gibi | :) | |
| 13:06:02 | kashyap | bauzas: quality > quantity, no? :-) | |
| 13:06:46 | bauzas | kashyap: surely yes, but honestly, I still need to review for others | |
| 13:07:01 | kashyap | Sure, quality doesn't exclude reviewing for othres :) | |
| 13:07:18 | bauzas | that's what I'm paid^R^R^R^R expected to do | |
| 13:07:33 | bauzas | :) | |
| 13:08:40 | sean-k-mooney | gibi: oh the qos interface attach | |
| 13:09:10 | sean-k-mooney | gibi: that would be a good one to bring up with neutron in the cross project session | |
| 13:09:29 | sean-k-mooney | gibi: they had a FFE for the ablity to update a qos policy | |
| 13:09:37 | sean-k-mooney | for bandwidth | |
| 13:09:48 | sean-k-mooney | but it had a bunch of unhandeled edgecases | |
| 13:10:05 | sean-k-mooney | it might be workth discussing if there is a more holistic way to hanel it | |
| 13:10:57 | gibi | sean-k-mooney: qos interface attach also has some of the same edge cases | |
| 13:11:17 | gibi | namely that we only allocate bandwidth inplace | |
| 13:11:22 | sean-k-mooney | gibi: ya the main gap they had is if you remove the qos policy they could not add it again | |
| 13:11:27 | gibi | ohh | |
| 13:11:36 | sean-k-mooney | because there was no set of alloctions in the port profile | |
| 13:11:47 | sean-k-mooney | so they could not update placement | |
| 13:12:03 | sean-k-mooney | i was wondering if we coudl always populated it or something | |
| 13:12:58 | gibi | we cannot store zero allocation unfortunately | |
| 13:13:18 | stephenfin | sean-k-mooney: would you be able to sanity check https://review.opendev.org/#/c/746949/ today? | |
| 13:13:36 | stephenfin | I think lyarwood was waiting on you before closing out that SR-IOV functional test series | |
| 13:13:39 | sean-k-mooney | now is as good a time as any before i get bogged down in specs | |
| 13:15:21 | stephenfin | great, thanks :) | |
| 13:16:28 | openstackgerrit | Stephen Finucane proposed openstack/nova master: functional: Add tests for mixed CPU policy https://review.opendev.org/755852 | |
| 13:16:29 | openstackgerrit | Stephen Finucane proposed openstack/nova master: objects: Fix issue in exception type https://review.opendev.org/756069 | |
| 13:16:29 | openstackgerrit | Stephen Finucane proposed openstack/nova master: objects: Add 'InstanceNUMACell.min_threads' https://review.opendev.org/756070 | |
| 13:24:36 | sean-k-mooney | stephenfin: +1 on that some nits inline but i dont see any blockers | |
| 13:58:03 | openstackgerrit | Lee Yarwood proposed openstack/nova master: libvirt: Increase incremental and max sleep time during device detach https://review.opendev.org/755799 | |
| 13:58:03 | openstackgerrit | Lee Yarwood proposed openstack/nova master: fup: Remove unused arguments from detach_device_with_retry https://review.opendev.org/756079 | |
| 14:08:14 | openstackgerrit | Merged openstack/nova master: libvirt: Remove MIN_{LIBVIRT,QEMU}_PMEM_SUPPORT https://review.opendev.org/746986 | |
| 14:21:04 | openstackgerrit | Lee Yarwood proposed openstack/nova master: fup: Stop returning a retry function from detach_device_with_retry https://review.opendev.org/756084 | |
| 14:50:56 | openstackgerrit | Lee Yarwood proposed openstack/nova master: zuul: Merge nova-evacuate into nova-multinode-live-migration https://review.opendev.org/755309 | |
| 15:04:13 | openstackgerrit | Lee Yarwood proposed openstack/nova master: DNM: Test Ib9ed7069cef5b73033351f7a78a3fb566753970d on Focal https://review.opendev.org/756097 | |
| 15:11:43 | openstackgerrit | Stephen Finucane proposed openstack/nova master: objects: Add 'InstanceNUMACell.min_threads' https://review.opendev.org/756070 | |
| 15:11:44 | openstackgerrit | Stephen Finucane proposed openstack/nova master: tests: Add functional test for #1898272 https://review.opendev.org/756100 | |
| 15:11:44 | openstackgerrit | Stephen Finucane proposed openstack/nova master: hardware: Check inventory of shared CPUs for 'mixed' policy https://review.opendev.org/756101 | |
| 15:14:18 | stephenfin | artom, sean-k-mooney: There's the fix for that bug I pointed out Friday. Just as easy as I thought, thankfully :) | |
| 15:14:49 | openstackgerrit | Lee Yarwood proposed openstack/nova master: libvirt: Log exception when unable to import rbd or rados https://review.opendev.org/746873 | |
| 15:15:14 | stephenfin | sean-k-mooney also ^ | |
| 15:22:22 | sean-k-mooney | stephenfin: cool open ill review after meeting | |
| 15:27:36 | openstackgerrit | Lee Yarwood proposed openstack/nova master: libvirt: Log exception when unable to import rbd or rados https://review.opendev.org/746873 | |
| 15:34:57 | stephenfin | gibi: Easy +W here https://review.opendev.org/#/c/746873/ if you've time | |
| 15:35:53 | gibi | stephenfin: done | |
| 15:35:58 | stephenfin | ta | |
| 15:36:54 | lyarwood | stephenfin: https://review.opendev.org/#/c/755928/ - likewise, an easy one if you have time | |
| 15:37:33 | stephenfin | I thought we'd done that already? | |
| 15:37:58 | stephenfin | Ah, no, started but incomplete https://review.opendev.org/#/c/752073/ | |
| 15:38:17 | lyarwood | ah my bad, missed the conflict | |
| 15:38:32 | lyarwood | came up over the weekend when we tried to switch the py38 jobs over to centos | |
| 15:39:46 | stephenfin | All good | |
| 15:39:59 | lyarwood | bt-- | |
| 15:40:06 | stephenfin | Federico has another related patch here that would be good to get in IMO https://review.opendev.org/#/c/750652/ | |
| 15:41:12 | lyarwood | stephenfin: ack'd | |
| 15:53:55 | openstackgerrit | Stephen Finucane proposed openstack/nova master: functional: Add test for #1898272 https://review.opendev.org/756100 | |
| 15:53:55 | openstackgerrit | Stephen Finucane proposed openstack/nova master: hardware: Check inventory of shared CPUs for 'mixed' policy https://review.opendev.org/756101 | |
| 15:53:56 | openstackgerrit | Stephen Finucane proposed openstack/nova master: objects: Add 'InstanceNUMACell.min_threads' https://review.opendev.org/756070 | |
| 16:06:46 | openstackgerrit | Stephen Finucane proposed openstack/nova master: hardware: Check inventory of shared CPUs for 'mixed' policy https://review.opendev.org/756101 | |
| 16:06:47 | openstackgerrit | Stephen Finucane proposed openstack/nova master: objects: Add 'InstanceNUMACell.min_threads' https://review.opendev.org/756070 | |
| 16:10:28 | openstackgerrit | Merged openstack/os-vif master: Refactor code of linux_net to more cleaner and increase performace https://review.opendev.org/746673 | |
| 17:06:33 | jiriprox | @gibi @sean-k-mooney Hi, thank you for your support. Finally we've found issue outside of nova images. New images and issue in lab environment come together :-( | |
| 17:16:23 | openstackgerrit | Stephen Finucane proposed openstack/os-vif master: tox: Rename 'UPPER_CONSTRAINTS_FILE' -> 'TOX_CONSTRAINTS_FILE' https://review.opendev.org/756135 | |
| 17:50:00 | rm_work | Looking at upgrading Nova/Placement from Rocky to Stein (which I understand requires switching to using placement service, which is either shipped WITH nova still in Stein, OR packaged separately which will be required in Train) | |
| 17:50:30 | rm_work | I've run through the documents about upgrading -- but I'm wondering if anyone remembers any other tips/tricks or gotchas that might not have made it into the docs | |
| 17:50:36 | sean-k-mooney | jiriprox: well at least you found it in the end | |
| 17:50:43 | rm_work | Figured it is worth asking before I charge ahead :D | |
| 17:51:20 | sean-k-mooney | rm_work: dependign on your installer it should be fully automated for you | |
| 17:51:37 | rm_work | no real installer to speak of | |
| 17:51:52 | sean-k-mooney | the change to intree vs out of tree did not really alter the service much it was disinged to run independtly form the start | |
| 17:52:54 | rm_work | right, I figured that part | |
| 17:52:54 | sean-k-mooney | the main thing to remember is not the db is managed with placement manage instead of nova-manage | |
| 17:52:59 | rm_work | so I'm planning to do that move separately | |
| 17:53:15 | sean-k-mooney | rm_work: placment should be updated before nova generally | |
| 17:53:24 | rm_work | hmm | |
| 17:53:45 | rm_work | I guess that's reasonable since nova isn't using it yet -- so can do placement to stein -> nova to stein | |
| 17:53:52 | sean-k-mooney | they use microversion so new nova can work with an old placment but only if you do not use any feature provide by the new placment | |
| 17:54:07 | rm_work | also good to know, we probably do not | |
| 17:54:25 | sean-k-mooney | rm_work: well nova is useing placment in rocky | |
| 17:54:39 | rm_work | just writes stuff to it informationally? | |
| 17:54:48 | rm_work | the scheduler doesn't use it, does it? | |
| 17:55:07 | sean-k-mooney | i belive we start usigning it optionally in pike | |
| 17:55:12 | rm_work | ok | |
| 17:55:13 | sean-k-mooney | in rocky i think it was required | |
| 17:55:15 | rm_work | hmm | |
| 17:58:00 | sean-k-mooney | queens has some requirement https://github.com/openstack/nova/blob/stable/rocky/releasenotes/notes/queens-compute-requires-placement-1.14-for-nested-rps-8abb49df061b167e.yaml | |
| 18:02:32 | rm_work | ok, so it'd generally be safe to upgrade placement as much as we want first, even up to train victoria or something? :D | |
| 18:03:17 | sean-k-mooney | am you should be able to technically run the master verion of placment with any older verion of nova since nova will always set the required api microverison | |
| 18:03:38 | sean-k-mooney | so it will tell placment via the microvioson to act like the older version until nova is upgraded | |
| 18:04:20 | sean-k-mooney | so if you plan to upgrade the service indepently across multiple verios then it should be ok to move placment directly to victoria | |
| 18:04:31 | sean-k-mooney | that said we only test with the same version upstream | |
| 18:04:48 | rm_work | k | |
| 18:04:51 | sean-k-mooney | placment has functional test to assert the old microversion behavior with new versions | |
| 18:05:09 | sean-k-mooney | so its functional test shoudl maintain the correct behavior we just dont have tempest integration tests | |