| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2022-12-01 | |||
| 14:13:19 | sean-k-mooney | tobias-urdin: before that you could only live migrate if the same cpus were free on the dest and that was kind of a hack | |
| 14:13:26 | sean-k-mooney | so before train really only cold migration | |
| 14:13:34 | sean-k-mooney | after train live migration should work properly | |
| 14:14:11 | sean-k-mooney | tobias-urdin: the same is true for hugepages. it was done in the same feautre | |
| 14:14:20 | sean-k-mooney | tobias-urdin: there is still one unfixed bug | |
| 14:14:52 | sean-k-mooney | live migrating between hosts with different vcpu_pin_sets or different cpu_shared_sets | |
| 14:15:04 | sean-k-mooney | for non numa instnace is still technically broken | |
| 14:15:15 | opendevreview | Merged openstack/nova stable/yoga: Record SRIOV PF MAC in the binding profile https://review.opendev.org/c/openstack/nova/+/855023 | |
| 14:15:21 | opendevreview | Merged openstack/nova stable/yoga: Remove double mocking https://review.opendev.org/c/openstack/nova/+/855024 | |
| 14:15:24 | tobias-urdin | ack, so if nodes are identical in terms of config (pin sets) it shouldn't be a problem | |
| 14:15:26 | opendevreview | Merged openstack/nova stable/yoga: Remove double mocking... again https://review.opendev.org/c/openstack/nova/+/855025 | |
| 14:15:32 | sean-k-mooney | we do not update the cpus for floating instnaces until you hard reboot the vm | |
| 14:15:32 | opendevreview | Merged openstack/nova stable/yoga: Add compute restart capability for libvirt func tests https://review.opendev.org/c/openstack/nova/+/855026 | |
| 14:15:42 | opendevreview | Merged openstack/nova stable/victoria: [compute] always set instance.host in post_livemigration https://review.opendev.org/c/openstack/nova/+/863903 | |
| 14:16:20 | sean-k-mooney | tobias-urdin: for pinend vms the pin sets can be differnt and it shoudl not be a problem | |
| 14:16:56 | sean-k-mooney | in trian+ | |
| 14:18:19 | tobias-urdin | ack, not sure I understand the bug tho | |
| 14:19:37 | sean-k-mooney | the but is if you are not using cpu pinning, and you have set cpu_share_set or vcpu_pin_set | |
| 14:20:01 | sean-k-mooney | if that is not the same on all hosts floating vms will float over the wrong cores after migrating | |
| 14:20:27 | sean-k-mooney | the numa aware migration feature only updates the cpus if the vm has a numa toplogy | |
| 14:20:43 | sean-k-mooney | if its a normal vm with no numa info it does not update the cpus | |
| 14:21:14 | sean-k-mooney | and if you have eiterh vcpu_pin_set defiend or cpu_shared_set then the floating vms will be soft pinned to float over those cores | |
| 14:21:40 | tobias-urdin | ah I see, thanks, that shouldn't be an issue on our part then. thanks as always! | |
| 14:21:48 | sean-k-mooney | so this bug has been there since like icehouse wehn vcpu_pin_set was added as a config option | |
| 14:22:33 | sean-k-mooney | we have just never fixed it as it need an rpc change to pass back the requried info | |
| 14:45:27 | tobias-urdin | ack | |
| 15:32:36 | atmark | hello, looking at resume_state_on_host_boot() in https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L4150-L4172, I like to add OR condition in https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L4164 such that it also ignores if the VM has property `NoRestart=True' e.g. `if state in ignored_states or instance_properties = | |
| 15:32:38 | atmark | 'NoRestart=True':`. Appreciate any pointers on how can I accomplish this. | |
| 15:54:51 | bauzas | sean-k-mooney: gibi found a good issue in your stable/xena series for VDPA https://review.opendev.org/c/openstack/nova/+/864932 | |
| 16:00:03 | bauzas | sean-k-mooney: just update the commit msg and we're all cool | |
| 16:01:30 | sean-k-mooney | bauzas: yes i think i have seen that | |
| 16:01:36 | sean-k-mooney | and i think its adressed in the train version | |
| 16:01:44 | sean-k-mooney | but ill need to look at it again | |
| 16:01:58 | sean-k-mooney | i know i have to respine it for something and redo the train oen as a result | |
| 18:14:36 | opendevreview | sean mooney proposed openstack/nova stable/xena: refactor: remove duplicated logic https://review.opendev.org/c/openstack/nova/+/864932 | |
| 18:14:37 | opendevreview | sean mooney proposed openstack/nova stable/xena: Record SRIOV PF MAC in the binding profile https://review.opendev.org/c/openstack/nova/+/864933 | |
| 18:14:38 | opendevreview | sean mooney proposed openstack/nova stable/xena: Remove double mocking https://review.opendev.org/c/openstack/nova/+/864934 | |
| 18:14:39 | opendevreview | sean mooney proposed openstack/nova stable/xena: Remove double mocking... again https://review.opendev.org/c/openstack/nova/+/864935 | |
| 18:14:40 | opendevreview | sean mooney proposed openstack/nova stable/xena: Add compute restart capability for libvirt func tests https://review.opendev.org/c/openstack/nova/+/864936 | |
| 18:14:41 | opendevreview | sean mooney proposed openstack/nova stable/xena: enable blocked VDPA move operations https://review.opendev.org/c/openstack/nova/+/864937 | |
| 18:15:53 | sean-k-mooney | bauzas: once ^ are merged ill redo the wallaby ones to update teh cherry-picked lines | |
| 18:16:15 | sean-k-mooney | i just updated the commit messages in the first two patches the rest are the same just rebased | |
| 19:25:46 | opendevreview | Merged openstack/nova stable/yoga: enable blocked VDPA move operations https://review.opendev.org/c/openstack/nova/+/855035 | |
| #openstack-nova - 2022-12-02 | |||
| 09:30:09 | whoami-rajat | hey #openstack-nova , nova-ceph-multistore is failing on wallaby gate, can anyone take a look at this patch fixing it? https://review.opendev.org/c/openstack/nova/+/866326 | |
| 09:51:50 | bauzas | whoami-rajat: haha, I saw the problem | |
| 09:52:19 | bauzas | whoami-rajat: thanks for your change, but I have a question | |
| 09:53:17 | bauzas | why couldn't we have a new os-brick version ? | |
| 09:53:46 | bauzas | because Wallaby is on EM ? | |
| 10:07:11 | whoami-rajat | bauzas, yes, the only way (I know of) is to take os-brick from source, I talked to the release team and not possible to release another os-brick for wallaby | |
| 10:10:22 | bauzas | whoami-rajat: which release model do you have for os-brick ? | |
| 10:10:32 | bauzas | lemme check actually | |
| 10:16:03 | bauzas | https://releases.openstack.org/teams/cinder.html#os-brick hah, you're not using the independent model | |
| 10:16:23 | bauzas | ok, so I understand then | |
| 10:16:41 | bauzas | we can't release a new version in the wallaby branch indeed as it's EM | |
| 10:27:20 | bauzas | but given it's on EM... | |
| 10:30:04 | bauzas | gibi: sean-k-mooney elodilles: if you could quickly review https://review.opendev.org/c/openstack/nova/+/866326 that'd be loved | |
| 10:31:43 | sean-k-mooney | oh the os brick change | |
| 10:31:46 | sean-k-mooney | sure | |
| 10:33:18 | sean-k-mooney | done | |
| 10:37:49 | bauzas | thanks | |
| 11:01:50 | whoami-rajat | bauzas, nope, we use cycle-with-intermediary https://github.com/openstack/releases/blob/4f29f584f85a7b26b69f6b5f8e47af4026dea686/deliverables/antelope/os-brick.yaml#L4 | |
| 11:01:57 | whoami-rajat | thanks bauzas and sean-k-mooney for the quick reviews! | |
| 11:04:12 | sean-k-mooney | its the same as os-vif | |
| 11:04:36 | sean-k-mooney | cycle-with-intermediary is the default and most useful release model we have | |
| 11:05:37 | whoami-rajat | yep, works good for client/non-client libraries | |
| 11:29:29 | opendevreview | Jorge San Emeterio proposed openstack/nova-specs master: Review usage of oslo-privsep library on Nova https://review.opendev.org/c/openstack/nova-specs/+/865432 | |
| 12:14:23 | elodilles | bauzas: sean-k-mooney: i think that patch is not needed: https://review.opendev.org/c/openstack/nova/+/866326 | |
| 12:14:50 | elodilles | i mean the above fixed already the issue ^^^ | |
| 12:15:03 | elodilles | so this is not needed and does not change anything: https://review.opendev.org/c/openstack/nova/+/866326 | |
| 12:15:19 | elodilles | by default the 'required project' is checked out with the matching branch | |
| 12:17:02 | elodilles | the override-branch is needed only if we run the job from a different branch, which we don't do | |
| 12:21:20 | opendevreview | Merged openstack/nova stable/wallaby: [stable-only] Use os-brick from source in wallaby https://review.opendev.org/c/openstack/nova/+/866326 | |
| 12:21:26 | opendevreview | Merged openstack/nova stable/wallaby: Add nova-ovs-hybrid-plug job https://review.opendev.org/c/openstack/nova/+/828418 | |
| 12:21:47 | sean-k-mooney | elodilles: oh sorry your right | |
| 12:21:50 | sean-k-mooney | its not | |
| 12:22:03 | sean-k-mooney | elodilles: i miss read it i tought it was adding os-brick | |
| 12:22:28 | sean-k-mooney | not adding the branchi missed that because of the comment | |
| 12:22:51 | sean-k-mooney | actully its also using the explit fully quallifed version which we shoudl revert | |
| 12:23:15 | sean-k-mooney | opendev.org/openstack/os-brick make it harder for third party cis | |
| 12:23:20 | opendevreview | Alexey Stupnikov proposed openstack/nova master: Don't ignore InstanceNotFound exception by libvirt https://review.opendev.org/c/openstack/nova/+/863665 | |
| 12:24:31 | opendevreview | Merged openstack/nova stable/wallaby: Revert "Revert resize: wait for events according to hybrid plug" https://review.opendev.org/c/openstack/nova/+/828419 | |
| 12:25:31 | opendevreview | sean mooney proposed openstack/nova stable/wallaby: Revert "[stable-only] Use os-brick from source in wallaby" https://review.opendev.org/c/openstack/nova/+/866454 | |
| 12:25:46 | sean-k-mooney | elodilles: ^ bauzas ^ | |
| 12:26:32 | sean-k-mooney | elodilles: the other disadvantage of the patch we just merged is i think override-branch: takes precidence over depneds on but not 100% sure about that | |
| 12:29:36 | opendevreview | Alexey Stupnikov proposed openstack/nova master: Don't ignore InstanceNotFound exception by libvirt https://review.opendev.org/c/openstack/nova/+/863665 | |
| 12:32:26 | elodilles | sean-k-mooney: thanks, +2'd | |
| 12:33:31 | sean-k-mooney | whoami-rajat: ^ just so you are aware we are reverting that patch since we already had an equivlanet merged | |
| 12:35:13 | opendevreview | Alexey Stupnikov proposed openstack/nova master: Don't ignore InstanceNotFound exception by libvirt https://review.opendev.org/c/openstack/nova/+/863665 | |
| 12:54:38 | whoami-rajat | sean-k-mooney, not sure what's the reason of revert since the gate on devstack is still broken with existing changes -- as mentioned in my commit message | |
| 12:55:05 | whoami-rajat | which my patch intended to fix | |
| 12:57:38 | opendevreview | Alexey Stupnikov proposed openstack/nova master: Log some InstanceNotFound exceptions from libvirt https://review.opendev.org/c/openstack/nova/+/863665 | |
| 13:01:14 | elodilles | whoami-rajat: are you sure? | |
| 13:01:43 | whoami-rajat | elodilles, well, it was broken when i proposed the patch, checking again | |
| 13:01:59 | elodilles | whoami-rajat: our fix landed on Nov 22nd, ~ 7 pm | |
| 13:02:21 | sean-k-mooney | whoami-rajat: the two shoudl mostly be the same exactp you are forcing the branch to be stable/wallaby | |
| 13:02:41 | sean-k-mooney | which is the default that will be used by zuul on the stable brnach | |
| 13:03:11 | sean-k-mooney | the only other change is you are explcitly referncing the connection that provides the repo | |
| 13:04:31 | whoami-rajat | ah i see now | |
| 13:05:12 | whoami-rajat | the gate report on this is 11:49 (my time) 22nd november | |
| 13:05:35 | whoami-rajat | and the patch you've referenced merged 11:48 22nd november https://review.opendev.org/c/openstack/nova/+/865134 | |