| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-03-10 | |||
| 14:29:51 | dansmith | quality warning has been in place since May 2019 | |
| 14:30:24 | dansmith | last actual change was Feb 2019, but from canonical which may or may not indicate actual usage, just a failed test or something | |
| 14:34:48 | lyarwood | stephenfin: https://review.opendev.org/#/c/706900/ - could you hit this again when you have time? | |
| 14:35:47 | sean-k-mooney | dansmith: ya well we could always move it out of tree, into a repo in the x namespace and if people care they can use it or step up to support it for xen | |
| 14:36:01 | dansmith | -2 on that plan :) | |
| 14:36:13 | sean-k-mooney | i know you hate out of tree drivers | |
| 14:36:13 | dansmith | it's in the git tree if they want to find it | |
| 14:36:36 | sean-k-mooney | well i was wondering should we be removing it form in tree | |
| 14:37:08 | sean-k-mooney | i assume that was what you were evenutally proposing for vmware if we start the clock and no one steps up | |
| 14:38:33 | mordred | people still use xen? | |
| 14:39:36 | dansmith | mordred: no, that's the point :) | |
| 14:39:56 | dansmith | sean-k-mooney: removing it from the tree does't remove it from the git history and previous releases | |
| 14:40:03 | dansmith | sean-k-mooney: and it's likely not working now anyway | |
| 14:40:37 | mordred | yeah - I agree with dansmith - it's always there - if new people want to work on it, they can always start with a revert patch reverting the removal | |
| 14:41:00 | mordred | doesn't need an x/ repo | |
| 14:41:57 | sean-k-mooney | dansmith: oh that is what you ment by it being in git | |
| 14:42:06 | sean-k-mooney | dansmith: ok that makes more sense | |
| 14:42:52 | sean-k-mooney | mordred: i think one of the issue with the xenserver driver is that it relise on python 2 | |
| 14:43:34 | sean-k-mooney | mordred: libvirt/xen does not have that limitaion as far as i am aware but the xen direct driver definetly had issues with python 3 elast year | |
| 14:44:57 | sean-k-mooney | lyarwood: the privsep discussion has come up a bunch of times. be we did not want to change how we use privsep until we had finished the rootwrap removal | |
| 14:47:17 | sean-k-mooney | lyarwood: the hacking check i was refing too was the one that require you to do "import nova.privsep.x" instead of from nova.privsep import x | |
| 14:47:35 | sean-k-mooney | so that when you use a privaldaged function it must be fully spelled out | |
| 14:48:43 | lyarwood | sean-k-mooney: kk wasn't aware of the previous discussion then or I've conveniently forgtten them :) | |
| 14:50:14 | sean-k-mooney | lyarwood: nova currently only has 1 privsep context which basically give all privladged calls full root permissions | |
| 14:50:17 | sean-k-mooney | https://github.com/openstack/nova/blob/master/nova/privsep/__init__.py | |
| 14:50:38 | sean-k-mooney | so the way we currently use it provides very little securtiy benifit | |
| 14:50:39 | openstackgerrit | Dan Smith proposed openstack/nova master: Deprecate the vmwareapi driver https://review.opendev.org/712101 | |
| 14:52:22 | lyarwood | sean-k-mooney: I can follow up and tighten this up under https://review.opendev.org/#/q/topic:bug/1861071_followups but for the intital bugfix series that I want to backport I'd like to avoid as much churn as possible tbh | |
| 14:52:51 | sean-k-mooney | lyarwood: what we should be doing is creating multiple context with a limited set of permission and only granting the permissions that are needed. | |
| 14:53:30 | sean-k-mooney | ya that is fair. i think we should be considring this a potential blueprint/work item for victoria at this point | |
| 14:54:02 | sean-k-mooney | i would prefer not to propagate more examples of the patteren but i agree wew should not mix this with the bug fix | |
| 14:56:11 | lyarwood | sean-k-mooney: kk thanks, I'll try to take a swing at that later this week and tbh that work will bleed into the V release anyway at this point. | |
| 14:58:58 | sean-k-mooney | lyarwood: ya im debating if we should defer it to a dedicated effort | |
| 14:59:03 | sean-k-mooney | like the nova-net removal | |
| 15:01:21 | lyarwood | sean-k-mooney: ah understood | |
| 15:03:21 | mriedem | vmware ci hasn't reported reliably in forever | |
| 15:03:58 | dansmith | mriedem: yeah I was going to point out that it doesn't report on this patch as evidence, assuming it doesn't | |
| 15:04:07 | dansmith | I haven't seen it report on vmware-touching patches in a while | |
| 15:04:54 | mriedem | heh, i haven't seen vmware touching patches in awhile | |
| 15:05:33 | dansmith | well, some things that touch all the drivers should have gotten a report | |
| 15:06:02 | mriedem | the vmware ci just hasn't worked in a couple of years | |
| 15:06:27 | mriedem | i think they got the band together long enough to make it work for the live migration patch because i said i wouldn't +2 it otherwise (and it needed to be multinode and show it passing live migrate) | |
| 15:06:30 | mriedem | but after that... | |
| 15:06:54 | mriedem | the xen driver was deprecated in train so it's probably too early to drop that in ussuri yeah? V at the earliest i'd think. | |
| 15:07:05 | mriedem | since anyone using xen isn't likely upgrading much these days anyway... | |
| 15:09:39 | openstackgerrit | Dan Smith proposed openstack/nova master: Deprecate the vmwareapi driver https://review.opendev.org/712101 | |
| 15:09:45 | mriedem | i'm happy to give a tenured +2 on ^ though :) | |
| 15:10:26 | dansmith | mriedem: yeah I commented on the xenapi driver a little earlier | |
| 15:10:42 | dansmith | seems like we should probably wait until V for that, just to be squeaky clean, even though I doubt anyone cares anymore | |
| 15:10:43 | mriedem | pep8 pass on that line you had to wrap? | |
| 15:10:56 | mriedem | someone will care 1 year after it's already deleted :/ | |
| 15:11:05 | mriedem | "care" | |
| 15:12:22 | mriedem | dansmith: i think you've got a pep8 error | |
| 15:12:24 | mriedem | line too long | |
| 15:12:47 | dansmith | mriedem: do I? | |
| 15:12:51 | mriedem | it was 79 before | |
| 15:12:57 | dansmith | fast8: commands succeeded | |
| 15:13:03 | mriedem | oh, well i'll +2 then | |
| 15:55:23 | openstackgerrit | Stephen Finucane proposed openstack/nova stable/train: Fix os-keypairs pagination links https://review.opendev.org/711895 | |
| 15:56:52 | openstackgerrit | Stephen Finucane proposed openstack/nova stable/stein: Fix os-keypairs pagination links https://review.opendev.org/711896 | |
| 15:58:13 | openstackgerrit | Stephen Finucane proposed openstack/nova stable/rocky: Fix os-keypairs pagination links https://review.opendev.org/711898 | |
| 16:06:32 | openstackgerrit | Sylvain Bauza proposed openstack/nova master: Pass allocations to virt drivers when resizing https://review.opendev.org/589085 | |
| 16:06:33 | openstackgerrit | Sylvain Bauza proposed openstack/nova master: Pass allocations to virt drivers when reverting resize https://review.opendev.org/712118 | |
| 16:41:49 | stephenfin | lyarwood: dead easy stable patch for you here https://review.opendev.org/711895 | |
| 16:48:25 | openstackgerrit | Stephen Finucane proposed openstack/nova master: Add transform_image_metadata request filter https://review.opendev.org/665775 | |
| 16:58:51 | lyarwood | stephenfin: are there many more of these? | |
| 16:59:00 | stephenfin | I think that's the only other one | |
| 16:59:12 | lyarwood | kk | |
| 16:59:20 | stephenfin | it's the only one I could spot anyway | |
| 17:33:51 | openstackgerrit | Monty Taylor proposed openstack/nova master: Run sdk functional tests on nova changes https://review.opendev.org/712092 | |
| 17:45:25 | openstackgerrit | John Garbutt proposed openstack/nova master: Add unified limits configuration https://review.opendev.org/712137 | |
| 17:45:25 | openstackgerrit | John Garbutt proposed openstack/nova master: WIP: Enforce unified limits using oslo.limit https://review.opendev.org/615180 | |
| 17:45:26 | openstackgerrit | John Garbutt proposed openstack/nova master: Add logic to enforce local api and db limits https://review.opendev.org/712139 | |
| 17:45:26 | openstackgerrit | John Garbutt proposed openstack/nova master: Add tests for the UnifiedLimits quota driver https://review.opendev.org/712138 | |
| 17:45:27 | openstackgerrit | John Garbutt proposed openstack/nova master: Make unified limits APIs return reserved of 0 https://review.opendev.org/712141 | |
| 17:45:27 | openstackgerrit | John Garbutt proposed openstack/nova master: WIP: Add tests for API using unified limits driver https://review.opendev.org/712140 | |
| 17:45:28 | openstackgerrit | John Garbutt proposed openstack/nova master: WIP: Update quota_class APIs for db and api limits https://review.opendev.org/712143 | |
| 17:45:28 | openstackgerrit | John Garbutt proposed openstack/nova master: WIP: Enforce api and db limits https://review.opendev.org/712142 | |
| 19:54:16 | donnyd | sean-k-mooney: can you kick off a multi-numa job if you have anything that can be - I was hoping to test to make sure it works | |
| 19:59:39 | sean-k-mooney | donnyd: just rechecked https://review.opendev.org/#/c/708977/ | |
| 20:07:21 | sean-k-mooney | donnyd: seams to be working fine http://zuul.openstack.org/stream/a34b85c571ed47db8c4f1a4b2f1a0d82?logfile=console.log | |
| 20:50:24 | donnyd | Woot woot | |
| 21:11:06 | openstackgerrit | Monty Taylor proposed openstack/nova master: Run sdk functional tests on nova changes https://review.opendev.org/712092 | |
| 21:54:42 | openstackgerrit | Merged openstack/nova stable/stein: Fix hypervisors paginted collection_name. https://review.opendev.org/711669 | |
| #openstack-nova - 2020-03-11 | |||
| 00:28:23 | openstackgerrit | Merged openstack/nova master: Don't overwrite greenthread-local context in host manager https://review.opendev.org/582332 | |
| 00:29:12 | openstack | Launchpad bug 1863190 in OpenStack Compute (nova) "Server group anti-affinity no longer works" [Undecided,New] | |
| 00:29:12 | melwitt | rm_work: if you're around, do you recall what version you tried your test with re: your comment on this bug? https://bugs.launchpad.net/nova/+bug/1863190 | |
| 01:22:03 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/queens: Add retry_on_deadlock to migration_update DB API https://review.opendev.org/712212 | |
| 02:43:25 | openstackgerrit | melanie witt proposed openstack/nova master: Add config option for neutron client retries https://review.opendev.org/712226 | |
| 05:07:26 | openstackgerrit | Merged openstack/nova stable/queens: libvirt: Ignore DiskNotFound during update_available_resource https://review.opendev.org/711279 | |
| 05:36:21 | rm_work | melwitt: so that was on rocky -- but we figured out that we had build retries set to 0 so it would never retry, thus breaking that logic | |
| 05:37:34 | rm_work | And with soft antiaffinity, obviously it would just "work" and so given certain packing schemes, we're almost guaranteed to have them land on the same host, which seems kinda dumb | |
| 05:38:27 | rm_work | The conclusion we came to is that relying on the retry behavior to fix schedule races did not seem like a good approach | |
| 05:38:59 | rm_work | And that there should be better more up front ways of handling that kind of locking | |
| 05:39:33 | rm_work | Octavia does similar locking using DB logic | |
| 06:19:35 | rm_work | melwitt: johnsom could detail some of that more probably | |
| 10:49:39 | openstackgerrit | Stephen Finucane proposed openstack/nova master: vif: Remove dead code https://review.opendev.org/703146 | |
| 10:50:48 | stephenfin | bauzas: Any chance you could blast these through this morning? https://review.opendev.org/#/c/703973/ https://review.opendev.org/#/c/703974/ | |
| 10:51:11 | stephenfin | Final two patches for nova-net removal | |
| 10:52:36 | stephenfin | bauzas: and if you're doing that, there's another dead easy one here https://review.opendev.org/#/c/686997/ | |