| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-07-01 | |||
| 19:19:48 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/stein: Add functional recreate test for regression bug 1825537 https://review.opendev.org/666959 | |
| 19:19:48 | openstack | bug 1825537 in OpenStack Compute (nova) stein "finish_resize failures incorrectly revert allocations" [Medium,In progress] https://launchpad.net/bugs/1825537 - Assigned to Matt Riedemann (mriedem) | |
| 19:19:48 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/stein: Drop source node allocations if finish_resize fails https://review.opendev.org/667155 | |
| 19:29:28 | mriedem | sean-k-mooney: left an idea in https://review.opendev.org/#/c/667177/ on how to avoid the driver interface change for backports | |
| 19:29:48 | mriedem | sean-k-mooney: heh you thought of the same, "maybe there is a way we can use the migration_context which we have to lookup the migration object?" | |
| 19:31:50 | sean-k-mooney | mriedem: hehe ya i looked at that very breifly in v1 https://review.opendev.org/#/c/667177/1/nova/virt/libvirt/driver.py@9138 and didnt foolow up as i didnt thinkwe initally needed too until artom looked at the logs. | |
| 19:33:39 | sean-k-mooney | the migration context does not have the migration object in it directly right | |
| 19:34:01 | mriedem | no but it has the id primary key you can use to get it | |
| 19:34:26 | sean-k-mooney | this https://github.com/openstack/nova/blob/master/nova/objects/migration_context.py#L45 | |
| 19:34:31 | sean-k-mooney | ok | |
| 19:34:59 | sean-k-mooney | that was where i got lost and was not sure where to go from next to get the migration object | |
| 19:36:11 | sean-k-mooney | so we can just do nova.objects.migration.get_by_id https://github.com/openstack/nova/blob/master/nova/objects/migration.py#L137 | |
| 19:37:28 | sean-k-mooney | ya that simpler then i thought it would be. are we ok with the extra db lookup or is the api chage preferable? | |
| 19:44:38 | openstackgerrit | sean mooney proposed openstack/nova-specs master: add libvirt pqos spec https://review.opendev.org/662264 | |
| 19:46:04 | sean-k-mooney | when a spec is under review should the blueprint definition be in Review, New or Pending Approval? or do we just not use that at and just go from new to approved when the spec is approved | |
| 19:51:22 | sean-k-mooney | mriedem: oh just noiced you already said that in https://review.opendev.org/#/c/667177/5/nova/virt/driver.py@741 | |
| 20:01:58 | mriedem | sean-k-mooney: "or do we just not use that at and just go from new to approved when the spec is approved" mostly that | |
| 20:02:11 | mriedem | definition isn't really used (i didn't use it really anyway) | |
| 20:02:21 | mriedem | if there was a spec i'd mark the definition as review or pending approval | |
| 20:18:01 | efried | mriedem, sean-k-mooney: If I cared more, I would enforce that "Direction: Approved" means "we want to do this" and "Design: Approved" means "and we know how we're gonna do it" (i.e. spec approved). | |
| 20:20:50 | sean-k-mooney | efried: i set https://blueprints.launchpad.net/nova/+spec/libvirt-pqos since the spec is under review | |
| 20:21:19 | sean-k-mooney | the direct is still needs approval and the implemetation technically should not started but i left it at teh default of unknown | |
| 20:21:34 | efried | sfine | |
| 20:23:43 | efried | mriedem: favor... | |
| 20:23:47 | openstackgerrit | Eric Fried proposed openstack/nova master: Clean up TestDriverBlockDevice helper Of DOOOOOM https://review.opendev.org/668556 | |
| 20:26:26 | efried | mriedem: if you would please review (and hopefully merge) https://review.opendev.org/#/c/566153/46 so nobody ever has to do it again, and it doesn't have to keep being rebased... | |
| 20:26:46 | efried | I've proposed a fup (see above) to address the issues I noted there. | |
| 20:27:17 | efried | (and that one is much smaller) | |
| 20:27:49 | openstackgerrit | Merged openstack/nova stable/queens: Stop logging traceback when skipping quiesce https://review.opendev.org/660328 | |
| 20:28:12 | mriedem | jesus i really don't want to review a nearly 600 LOC mox->mock conversion patch | |
| 20:28:48 | mriedem | as you know i've never really been on board with the push to de-mox-ify everything as a goal | |
| 20:29:03 | mriedem | enforcing that new tests are in mock sure, but this is.... | |
| 20:33:56 | openstackgerrit | Merged openstack/nova stable/queens: xenapi/agent: Change openssl error handling https://review.opendev.org/656308 | |
| 20:34:03 | openstackgerrit | Merged openstack/nova stable/queens: Fix {min|max}_version in ironic Adapter setup https://review.opendev.org/656523 | |
| 20:34:10 | openstackgerrit | Merged openstack/nova stable/queens: Fix retry of instance_update_and_get_original https://review.opendev.org/659320 | |
| 20:34:16 | openstackgerrit | Merged openstack/nova stable/queens: DRY up test_rollback_live_migration_set_migration_status https://review.opendev.org/662470 | |
| 20:34:23 | openstackgerrit | Merged openstack/nova stable/queens: Include all network devices in nova diagnostics https://review.opendev.org/666152 | |
| 20:34:32 | openstackgerrit | Merged openstack/nova stable/queens: Add regression recreate test for bug 1830747 https://review.opendev.org/662774 | |
| 20:34:32 | openstack | bug 1830747 in OpenStack Compute (nova) queens "Error 500 trying to migrate an instance after wrong request_spec" [High,In progress] https://launchpad.net/bugs/1830747 - Assigned to Matt Riedemann (mriedem) | |
| 20:35:00 | efried | aspiers: SEV series in merge conflict after the first patch :( | |
| 20:49:16 | spatel | sean-k-mooney: yt | |
| 20:51:24 | spatel | getting this issue - https://bugzilla.redhat.com/show_bug.cgi?id=1176020 | |
| 20:51:25 | openstack | bugzilla.redhat.com bug 1176020 in libvirt "libvirt should do a right check for numa cpus set" [Medium,Closed: errata] - Assigned to mprivozn | |
| 21:03:14 | openstackgerrit | Eric Fried proposed openstack/nova master: Rename CinderFixtureNewAttachFlow to CinderFixture https://review.opendev.org/668561 | |
| 21:07:11 | mriedem | i see what you're doing | |
| 21:10:59 | mriedem | bbiab | |
| 21:19:47 | openstackgerrit | Merged openstack/nova stable/queens: Workaround missing RequestSpec.instance_group.uuid https://review.opendev.org/663110 | |
| 21:22:20 | efried | o/ jaypipes, how goes? | |
| 21:30:25 | sean-k-mooney | spatel: you are getting that because you are livemigrating instance with a numa toplogy | |
| 21:30:45 | sean-k-mooney | spatel: when you do that we do not recalulate the guest xml currently | |
| 21:31:13 | sean-k-mooney | as a result if you live migrate form a host with more cpus to one with less you can have an xml on the dest that reference cpus that are not present | |
| 21:31:35 | spatel | sean-k-mooney: now i am stuck and not sure how to get out | |
| 21:31:55 | sean-k-mooney | spatel: cold migration should work | |
| 21:32:08 | spatel | i tried all option but getting same error | |
| 21:32:27 | spatel | now thinking to re-build it... | |
| 21:32:57 | spatel | when you said - you are getting that because you are livemigrating instance with a numa toplogy | |
| 21:33:11 | spatel | what does that means? | |
| 21:33:28 | sean-k-mooney | i assume you are getting that error becasue you tried to do a live migration | |
| 21:33:37 | spatel | if i have machine and i said live migrate, where numa comes in picture | |
| 21:33:39 | sean-k-mooney | if that is not the case how did you hit this issue | |
| 21:34:04 | sean-k-mooney | if your instance use hugepage or cpu pinning it will have a numa toplogy | |
| 21:34:19 | spatel | someone from my team trying to do and it endup in this situation, i have to talk to him to get full picture | |
| 21:35:29 | spatel | oh.. yes that instance had hugepage/cpu pinning | |
| 21:35:57 | spatel | you are saying we can't do live migration if instance has hugepage/cpu pinning? | |
| 21:36:01 | spatel | sean-k-mooney: ^ | |
| 21:36:18 | sean-k-mooney | technically it can work but its not properly supported by nova | |
| 21:36:31 | sean-k-mooney | we dont currently update teh xml when migrating | |
| 21:36:37 | sean-k-mooney | for live migrations | |
| 21:37:12 | sean-k-mooney | spatel: its a knwon issue that is being adressed by https://review.opendev.org/#/q/topic:bp/numa-aware-live-migration+(status:open+OR+status:merged) | |
| 21:38:36 | spatel | so it not merge yet? | |
| 21:38:41 | sean-k-mooney | cold migrations are supported but until ^ is implemented nova will not update teh libvirt xml and you can get that error amoung others | |
| 21:38:45 | sean-k-mooney | no its not | |
| 21:39:07 | sean-k-mooney | for stien we specifcialy disable live migration for instnace with numa toplogies | |
| 21:39:17 | sean-k-mooney | i think we started backporting that too | |
| 21:39:57 | spatel | oh! | |
| 21:40:04 | spatel | this is a big issue then.. | |
| 21:40:38 | spatel | how to migrate instance if my compute node is dead and all vm running on shared storage? | |
| 21:41:05 | sean-k-mooney | if the compute node is dead and the vms are using shared storage you should use evacuate | |
| 21:41:28 | spatel | hmm.. | |
| 21:41:30 | sean-k-mooney | it will rebuild the instance on another host | |
| 21:41:42 | sean-k-mooney | but you should only use evacuate if the instance is stopped | |
| 21:41:56 | spatel | what if i want to migrate specific vm ? any heck there? | |
| 21:42:28 | spatel | sean-k-mooney: so you are saying i can shutdown instance and do evaluate ? | |
| 21:42:30 | sean-k-mooney | evacuate works for a specifc vm | |
| 21:43:03 | spatel | This is very important information, i was totally not aware of it.. | |
| 21:43:55 | spatel | I need to make big banner now to tell everyone don't do live migration period.. | |
| 21:44:13 | sean-k-mooney | if you have not already read it the go to text on evacuate is http://www.danplanet.com/blog/2016/03/03/evacuate-in-nova-one-command-to-confuse-us-all/ | |
| 21:44:23 | sean-k-mooney | what release of openstack are you running | |
| 21:44:46 | spatel | my old cloud running on queens and new cloud i am build on stein | |
| 21:45:46 | sean-k-mooney | looks we have the config option on rocky https://github.com/openstack/nova/blob/stable/rocky/nova/conf/workarounds.py#L223 | |
| 21:46:58 | sean-k-mooney | ah we havnt merged https://review.opendev.org/#/c/629597/ | |
| 21:48:31 | spatel | hmm is it config option which i can turn on to protect from accident ? | |
| 21:48:47 | sean-k-mooney | spatel: live migration with hugepage works most of the time. cpu pinning is much less likely to work correctly | |
| 21:49:17 | sean-k-mooney | with that patch by default we will not allow instace with a numa toplogy to be live migrated | |
| 21:49:24 | sean-k-mooney | so its safe by default | |
| 21:49:26 | jaypipes | efried: ahoy! :) going well, ty! | |
| 21:49:39 | efried | good to hear | |
| 21:50:15 | spatel | sean-k-mooney: hmm we used both pinning/hugepage :) | |
| 21:51:10 | sean-k-mooney | :) yes. this si something we have wanted to fix for several relese. it will get done this cycle one way or another. | |
| 21:51:19 | sean-k-mooney | unfortuanetly its not backportable | |