| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-08-20 | |||
| 13:38:54 | openstackgerrit | Vlad Gusev proposed openstack/nova stable/stein: libvirt: Use 'writeback' QEMU cache mode when 'none' is not viable https://review.opendev.org/747163 | |
| 13:44:36 | openstackgerrit | Mohammed Naser proposed openstack/nova master: db: fix database migrations when name includes dash https://review.opendev.org/747164 | |
| 13:45:29 | openstackgerrit | Mohammed Naser proposed openstack/nova master: db: fix database migrations when name includes dash https://review.opendev.org/747164 | |
| 13:45:30 | mnaser | ^ very trivial, honestly, i don't know how we'd test it | |
| 14:00:19 | stephenfin | alex_xu: I think there are a whole class of bugs here to do with races between setting the migration status to foo and dropping a move claim | |
| 14:00:26 | stephenfin | Thanks for the rabbit hole \o/ :P | |
| 14:06:02 | kashyap | lyarwood: Or anyone who recalls stable rules (my mind is ill right now, can't recall), does the above 'writeback' backport (https://review.opendev.org/#/c/747163/) look okay? | |
| 14:13:10 | lyarwood | kashyap: we would retain the existing cachemodes during a LM right? | |
| 14:13:34 | kashyap | lyarwood: Correct, until the instance is cold-rebooted on the destination. | |
| 14:13:40 | lyarwood | kashyap: if it doesn't impact running instances during LM then I'd be okay with this assuming it's now doing the correct thing | |
| 14:14:28 | lyarwood | kashyap: I'll review it in detail later | |
| 14:14:33 | kashyap | Yeah, it does the right thing now. I've spent several days extensively documenting it | |
| 14:14:48 | kashyap | lyarwood: No rush at all for it; just noticed only when Vlad (above) posted it | |
| 14:32:41 | artom | sean-k-mooney, whenever you're around, wanna talk about the multiple vif bindings thing? | |
| 14:33:03 | sean-k-mooney | artom: am sure want to do it on video chat or here | |
| 14:33:08 | sean-k-mooney | i want to grab coffee first | |
| 14:33:11 | sean-k-mooney | in either case | |
| 14:33:24 | sean-k-mooney | but im more or less free so say in 10 mins | |
| 14:33:40 | sean-k-mooney | our quater to the hour | |
| 14:34:03 | artom | sean-k-mooney, so :45? OK | |
| 14:34:17 | sean-k-mooney | yes | |
| 14:34:23 | sean-k-mooney | that is what i ment by quater to the hour | |
| 14:34:40 | sean-k-mooney | i debated if :45 would translate well | |
| 14:42:04 | openstackgerrit | Vlad Gusev proposed openstack/nova stable/ussuri: replace the "hide_hypervisor_id" to "hw:hide_hypervisor_id" https://review.opendev.org/747189 | |
| 14:42:52 | sean-k-mooney | i guess that is valid to backport | |
| 14:43:38 | sean-k-mooney | its techncially adding a new extra spec which we would allow downtream but its just an alias for an existing one that is broken/problematic | |
| 14:43:53 | sean-k-mooney | so its fixing a bug without really providing new functionality | |
| 14:45:33 | artom | sean-k-mooney, err, give me 15 minutes? Brb | |
| 14:50:49 | sean-k-mooney | ping me when it suits you | |
| 15:02:11 | sean-k-mooney | lyarwood: when through your libvirt version change series more less teh look correct but the cloud archive is not being used currently so you need your devstack change for them to pass | |
| 15:02:34 | sean-k-mooney | so stick a depends-on in https://review.opendev.org/#/c/746981/1 and rebase the reset | |
| 15:03:43 | lyarwood | sean-k-mooney: yup going to work through that again later | |
| 15:04:27 | sean-k-mooney | most of it looks sane and your deleteing lots of code whcih si always nice to see | |
| 15:05:12 | sean-k-mooney | we might want to send a note to thrid part ci maintainer on the mailing list too | |
| 15:05:25 | sean-k-mooney | since they will need to make sure they have the new min version avaiable too | |
| 15:07:09 | artom | sean-k-mooney, ok, back, ready when you are | |
| 15:08:51 | sean-k-mooney | ah finishing an email but more or less one sec | |
| 15:11:02 | sean-k-mooney | ok sent | |
| 15:11:41 | sean-k-mooney | what to chat on google meet | |
| 15:11:51 | sean-k-mooney | or here | |
| 15:12:42 | artom | sean-k-mooney, meet | |
| 15:13:19 | artom | meet.google.com/zzq-favq-pxj | |
| 15:50:58 | lyarwood | Meeting in #openstack-meeting-3 in ~10mins | |
| 15:51:04 | lyarwood | I'll keep it short again this week :) | |
| 16:00:02 | lyarwood | Starting the meeting shortly | |
| 18:53:35 | artom | sean-k-mooney, you'll probably also have to fiddle with the migrate_data.vifs stuff in nova/conductor/tasks/live_migrate.py for your eventual fix, btw | |
| 19:07:45 | artom | Also, writing this commit message is a major PITA | |
| #openstack-nova - 2020-08-21 | |||
| 00:30:34 | openstackgerrit | sean mooney proposed openstack/nova master: [WIP] Set migrate_data.vifs only when using multiple port bindings https://review.opendev.org/742180 | |
| 02:45:02 | openstackgerrit | Merged openstack/nova master: Pass the actual target in baremetal nodes policy https://review.opendev.org/741717 | |
| 02:45:13 | openstackgerrit | Merged openstack/nova master: Add scope and new default roles in extensions policies https://review.opendev.org/743046 | |
| 02:45:24 | openstackgerrit | Merged openstack/nova master: Add new default roles in networks policies https://review.opendev.org/742566 | |
| 02:45:31 | openstackgerrit | Merged openstack/nova master: Pass the actual target in networks policy https://review.opendev.org/742568 | |
| 02:45:38 | openstackgerrit | Merged openstack/nova master: libvirt: Do not reference VIR_ERR_DEVICE_MISSING when libvirt is < v4.1.0 https://review.opendev.org/746210 | |
| 08:23:51 | openstackgerrit | Lee Yarwood proposed openstack/nova stable/ussuri: libvirt: Do not reference VIR_ERR_DEVICE_MISSING when libvirt is < v4.1.0 https://review.opendev.org/747357 | |
| 08:27:11 | openstackgerrit | Lee Yarwood proposed openstack/nova stable/train: libvirt: Do not reference VIR_ERR_DEVICE_MISSING when libvirt is < v4.1.0 https://review.opendev.org/747358 | |
| 08:49:59 | alex_xu | stephenfin: yea, guess, the db status and rt isn't update under same lock | |
| 08:50:07 | alex_xu | s/guess/guess so/ | |
| 08:50:47 | stephenfin | alex_xu: The ResourceTracker is so broken :'( | |
| 08:51:02 | stephenfin | Spent all day merely trying to figure out how broken it is, heh | |
| 08:51:14 | openstackgerrit | Lee Yarwood proposed openstack/nova stable/stein: libvirt: Do not reference VIR_ERR_DEVICE_MISSING when libvirt is < v4.1.0 https://review.opendev.org/747360 | |
| 08:51:17 | alex_xu | hah | |
| 08:52:44 | openstackgerrit | Lee Yarwood proposed openstack/nova stable/rocky: libvirt: Do not reference VIR_ERR_DEVICE_MISSING when libvirt is < v4.1.0 https://review.opendev.org/747361 | |
| 08:55:17 | openstackgerrit | Lee Yarwood proposed openstack/nova stable/queens: libvirt: Do not reference VIR_ERR_DEVICE_MISSING when libvirt is < v4.1.0 https://review.opendev.org/747362 | |
| 08:56:22 | lyarwood | elod_off: https://review.opendev.org/#/q/topic:bug/1887946+status:open - When you're back this series will need some stable reviews :) | |
| 10:06:37 | openstackgerrit | Merged openstack/nova master: functional: Move single-use function to its caller https://review.opendev.org/745904 | |
| 10:06:43 | openstackgerrit | Merged openstack/nova master: functional: Drop '_api' suffix from placement fixture https://review.opendev.org/745905 | |
| 11:04:00 | lyarwood | https://github.com/openstack/nova/blob/835440e3f9dace4074511036473ebdae22f72c25/nova/context.py#L268-L279 - does anyone know if there's a way to get an admin context with a populated service catalog? | |
| 11:05:23 | lyarwood | ^ at the moment we are using this during self.driver.resume_state_on_host_boot and failing when attempting to hard reboot instances with encrypted volumes as the key_manager service isn't listed in the service catalog | |
| 11:06:07 | lyarwood | appreciate that even if it was the admin context wouldn't be able to fetch user secrets without changing policy but I'd still like to fix this if possible | |
| 11:49:32 | artom | sean-k-mooney, well my VPN just died, but https://github.com/openstack/nova/blob/master/nova/virt/libvirt/migration.py#L96 | |
| 11:50:07 | artom | And https://github.com/openstack/nova/blob/master/nova/virt/libvirt/migration.py#L144 | |
| 12:40:01 | sean-k-mooney | artom that only does it for numa migration | |
| 12:40:11 | sean-k-mooney | artom: it does not print them for all migrations | |
| 12:40:32 | sean-k-mooney | because of https://github.com/openstack/nova/blob/835440e3f9dace4074511036473ebdae22f72c25/nova/virt/libvirt/migration.py#L66 | |
| 12:40:55 | sean-k-mooney | we should move the loging to the start and end of get_updated_guest_xml | |
| 12:41:06 | sean-k-mooney | so it does it for all migrations | |
| 12:43:38 | artom | sean-k-mooney, ah, in that sense, yeah | |
| 12:44:01 | openstack | Launchpad bug 1888395 in OpenStack Compute (nova) "shared live migration of a vm with a vif is broken in train" [High,In progress] - Assigned to sean mooney (sean-k-mooney) | |
| 12:44:01 | sean-k-mooney | dansmith: by the way i figured out why https://bugs.launchpad.net/nova/+bug/1888395 happens im working on a repoducer now but the fix is basicaly https://review.opendev.org/#/c/742180/ | |
| 12:46:39 | sean-k-mooney | stephenfin: is the only functional test with live migration and port your new one or do you know if gibi added some for bandwidth based schduling | |
| 12:46:53 | sean-k-mooney | i think he did but just going to look for them now | |
| 12:47:58 | sean-k-mooney | artom: you siad your numa live migation test dont use interface right | |
| 12:48:05 | artom | sean-k-mooney, nope | |
| 12:48:07 | sean-k-mooney | ah yes server = self._create_server(flavor_id=flavor, host=host, | |
| 12:48:10 | sean-k-mooney | networks='none') | |
| 12:48:27 | artom | sean-k-mooney, I'm wondering if it wouldn't be easier to reproduce that migrate_data.vifs bug with a DNM CI patch | |
| 12:48:43 | artom | Is there a way we can deploy Neutron in CI without the multiple port bindings extension? | |
| 12:49:29 | sean-k-mooney | no but i can hack the nova neutron client to always return false | |
| 12:50:05 | sean-k-mooney | that may or may not tirgger it i guess | |
| 12:50:45 | sean-k-mooney | i might however ill try that quickly i would prefer to have a regression test however | |
| 12:51:02 | artom | sean-k-mooney, yeah, long-term something merged in-tree would be better | |
| 12:51:05 | artom | Or a real CI job for it | |
| 12:51:18 | sean-k-mooney | im not standing up a contrial ci :P | |
| 12:52:06 | sean-k-mooney | live migration is really uncloudy we should maybe just remove it instead. it would be a lot less work and headaces then a contrail ci :) | |
| 12:53:24 | artom | lulz | |
| 12:53:43 | sean-k-mooney | it would "fix" all the live migration bugs :) | |
| 12:58:10 | sean-k-mooney | so there is a live migration test with the old microvertion that does not support it for bandwith | |
| 12:58:13 | sean-k-mooney | https://github.com/openstack/nova/blob/master/nova/tests/functional/test_servers.py#L5993-L6029 | |
| 12:58:41 | sean-k-mooney | but not the new one so ya i think i would need stephenfin patches to cover the gaps in the neutron fixture | |
| 12:59:38 | sean-k-mooney | oh there are more test for move operatoins later | |