| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2022-04-05 | |||
| 09:58:33 | Uggla | this field is not used by the api. | |
| 10:02:29 | sean-k-mooney | yes | |
| 10:02:42 | sean-k-mooney | so a volume can have multiple attachments when its moving | |
| 10:03:19 | sean-k-mooney | althoguh this is a manilla share | |
| 10:03:28 | sean-k-mooney | so not entirly sure | |
| 10:03:35 | sean-k-mooney | i woudl ahve to read the spec again | |
| 10:05:49 | sean-k-mooney | Uggla so attachment id is the interanl id filed in the share_mappings table https://specs.openstack.org/openstack/nova-specs/specs/yoga/approved/libvirt-virtiofs-attach-manila-shares.html#data-model-impact | |
| 10:07:37 | sean-k-mooney | i think the intent is to allow the share to be assoated with teh vim but with differe exprot locations during move operations | |
| 10:08:03 | sean-k-mooney | but im not certin | |
| 10:08:23 | Uggla | sean-k-mooney, each line of the table is unique by the couple instance_uuid, share_id | |
| 10:08:50 | sean-k-mooney | i think attachment id was ment to be in the unique constraint | |
| 10:09:03 | sean-k-mooney | lyarwood: ^ if you are around | |
| 10:09:18 | sean-k-mooney | lyarwood: do you remember why we have attachment id | |
| 10:11:26 | sean-k-mooney | if the id filed is just intended as a primary key it should be an integer not a uuid so the fact its listed as a uuid is what is making me belive it for move operations | |
| 10:11:52 | Uggla | sean-k-mooney, I think I can add it easily, it is just to understand the purpose of it. | |
| 10:11:53 | sean-k-mooney | so that we can create a new attachmet with a different export location for the destination. | |
| 10:12:02 | sean-k-mooney | yep | |
| 10:13:11 | Uggla | let's see id lyarwood can answer. If not I'll try to ping him. | |
| 10:13:59 | sean-k-mooney | have you already implemented move operations | |
| 10:14:37 | sean-k-mooney | i dont think you will need it before then | |
| 10:14:52 | sean-k-mooney | the other suecases is allowign the share to be mounted to the vm multiple times | |
| 10:15:46 | lyarwood | yeah I was just typing that | |
| 10:15:54 | lyarwood | I think this was more to do with the multi attach case of a share | |
| 10:16:07 | Uggla | sean-k-mooney, with this first version --> Support for move operations once a share is attached will also not be covered by this spec | |
| 10:16:44 | Uggla | sean-k-mooney, ny requests to shelve, evacuate, resize, cold migrate or live migrate an instance with a share attached will be rejected for the time being | |
| 10:16:47 | lyarwood | that and debugging issues, I've always found it easier to use the underlying bdm id when tracing volume attachment issues | |
| 10:18:21 | Uggla | lyarwood, ok make sens. | |
| 10:18:58 | Uggla | lyarwood, thx | |
| 10:25:51 | opendevreview | Stephen Finucane proposed openstack/nova master: hacking: Prevent use of six https://review.opendev.org/c/openstack/nova/+/836588 | |
| 10:25:59 | stephenfin | sean-k-mooney: bauzas: ^ | |
| 10:33:06 | sean-k-mooney | stephenfin: care to add a pre-commit check? i guess that should be picked up by the flake8 check | |
| 10:33:17 | sean-k-mooney | so i guess it already is there | |
| 10:33:38 | sean-k-mooney | have you tested that locally to confirm one way or another? | |
| 10:33:51 | sean-k-mooney | otherwise ya that looks like a good idea to me | |
| 10:50:10 | gibi | Uggla: left a bunch of comments / questions on https://review.opendev.org/c/openstack/nova-specs/+/833669 (manila shares spec). Let me know if something of them are unclear | |
| 11:14:46 | opendevreview | Stephen Finucane proposed openstack/nova master: Sync rootwrap.conf from oslo.rootwrap https://review.opendev.org/c/openstack/nova/+/823229 | |
| 11:19:23 | sean-k-mooney | stephenfin: actully i need to add that to the cinder ptg | |
| 11:19:37 | sean-k-mooney | stephenfin:we only have rootwrap because of os-brick | |
| 11:19:41 | sean-k-mooney | nova does not use it directly | |
| 11:19:48 | sean-k-mooney | so i would like to move ti to os-brick | |
| 11:20:13 | sean-k-mooney | that or we could finsih porting os-brick to privsep so that it does not need it either | |
| 11:22:09 | stephenfin | makes sense | |
| 11:25:02 | sean-k-mooney | i pinged the cinder folks to see if i can get a slot to chat to them about it in there ptg adgenda | |
| 11:25:18 | sean-k-mooney | it looks pretty full so maybe we can do it in ours if they can have people join | |
| 11:57:37 | sean-k-mooney | stephenfin: ping me when you have time to review my comments on the hacking change https://review.opendev.org/c/openstack/nova/+/836588 | |
| 11:57:53 | sean-k-mooney | ill be afk for the next 30 mins or so but happy to look at that again when im back | |
| 11:59:28 | opendevreview | Stephen Finucane proposed openstack/nova master: hacking: Prevent use of six https://review.opendev.org/c/openstack/nova/+/836588 | |
| 12:03:23 | opendevreview | Stephen Finucane proposed openstack/nova master: pre-commit: Sync hacking version, bump plugins https://review.opendev.org/c/openstack/nova/+/836639 | |
| 12:03:42 | stephenfin | sean-k-mooney: Thanks for the review. Fixed that issue | |
| 12:03:47 | stephenfin | (good spot) | |
| 12:15:41 | Uggla | gibi, thx I'm will have a look at your comments. | |
| 12:16:01 | gibi | ack | |
| 12:29:13 | sean-k-mooney | stephenfin: ack ill test them both locally again and then review | |
| 12:30:14 | bauzas | stephenfin: ack too | |
| 12:30:46 | bauzas | stephenfin: given it's only for new changes (thanks to hacking), it looks good to me | |
| 12:36:25 | bauzas | oh waity | |
| 12:37:30 | opendevreview | Merged openstack/nova master: Fix unit tests when they are run with OS_DEBUG=True https://review.opendev.org/c/openstack/nova/+/833115 | |
| 12:39:57 | bauzas | stephenfin: sean-k-mooney: https://review.opendev.org/c/openstack/nova/+/836588/2/nova/hacking/checks.py#1046 | |
| 12:39:58 | sean-k-mooney | bauzas: we shoudl not currently have any usage of six in the codebases | |
| 12:40:52 | sean-k-mooney | bauzas: https://codesearch.opendev.org/?q=six&i=nope&literal=nope&files=&excludeFiles=&repos=openstack%2Fnova | |
| 12:41:00 | bauzas | that's what I was verifying | |
| 12:42:30 | sean-k-mooney | bauzas: tox -e pep8 check all files and i verified that with pre-commit run --all-files also | |
| 12:42:45 | sean-k-mooney | so i know that the only files that it flagged on where the two places i added it for testing | |
| 12:43:25 | sean-k-mooney | so on master at least we are all good. replied in the revew with the codesearch link too | |
| 12:49:49 | bauzas | sean-k-mooney: sent to the gate | |
| 13:07:47 | opendevreview | Merged openstack/nova master: Remove the code that never reach https://review.opendev.org/c/openstack/nova/+/809572 | |
| 13:07:56 | opendevreview | Merged openstack/nova master: Fix eventlet.tpool import https://review.opendev.org/c/openstack/nova/+/830383 | |
| 13:08:04 | opendevreview | Merged openstack/nova master: Imported Translations from Zanata https://review.opendev.org/c/openstack/nova/+/836113 | |
| 13:18:33 | opendevreview | Alexey Stupnikov proposed openstack/nova stable/xena: stable/xena: fix openstack-tox-lower-constraints https://review.opendev.org/c/openstack/nova/+/836650 | |
| 13:19:34 | tobias-urdin | now when running_vms key is removed from /os-hypervisors/detail after microversion 2.87 what is the best way (without looping every single servers) to get total amount of instance and running vms using api? | |
| 13:24:03 | opendevreview | Alexey Stupnikov proposed openstack/nova stable/xena: stable/xena: fix openstack-tox-lower-constraints https://review.opendev.org/c/openstack/nova/+/836650 | |
| 13:33:53 | opendevreview | Maxim Korezkij proposed openstack/nova master: Handle volume attachments https://review.opendev.org/c/openstack/nova/+/833234 | |
| 13:40:02 | bauzas | reminder : we start our PTG in 20 mins \o/ | |
| 13:45:07 | artom | Precarious Tricky Game? | |
| 13:46:44 | bauzas | muh | |
| 13:47:14 | tobias-urdin | hm I guess I could hit os-instance_usage_audit_log but that isn't reliable | |
| 14:00:01 | sean-k-mooney | normally you would get the totals form placment | |
| 14:00:38 | sean-k-mooney | and the runnign instance i guess via a openstack server list --all-tenant and filter that for running | |
| 14:00:51 | sean-k-mooney | it depends on what you actually are trying to do | |
| 14:00:54 | sean-k-mooney | tobias-urdin: ^ | |
| 14:01:19 | sean-k-mooney | https://www.openstack.org/ptg/rooms/newton | |
| 14:01:23 | sean-k-mooney | nova room ^ | |
| 14:03:00 | bauzas | what sean-k-mooney said | |
| 14:03:03 | bauzas | opening the PTG | |
| 14:03:58 | stephenfin | I'll be along at :30. Got a conflict | |
| 14:08:04 | opendevreview | Maxim Korezkij proposed openstack/nova master: Handle volume attachments https://review.opendev.org/c/openstack/nova/+/833234 | |
| 14:10:19 | tobias-urdin | sean-k-mooney: we get allocations from placement (vcpu, memory, disk, gpu etc) instead of the old hypervisor statistics API but cannot find a way to get number of instances only resource usage | |
| 14:11:49 | sean-k-mooney | you would count allocations of type instance | |
| 14:12:33 | tobias-urdin | oh there is a consumer_type on allocations, i see, thanks :) | |
| 14:12:36 | sean-k-mooney | although honestly openstack server list --all-projects is proably the best way to do this | |
| 14:12:52 | sean-k-mooney | if you are an admin at least | |
| 14:13:11 | sean-k-mooney | placement wont actully count shelved instnaces | |
| 14:13:24 | sean-k-mooney | since they are not on a host | |
| 14:13:57 | tobias-urdin | it's just to get an overview the easiest way (an less intensive way), isn't actually used for anything else since we have resource usage for planning | |
| 14:14:40 | sean-k-mooney | so one of the reasons we remvoed this is it was cached info that can get out of sync if the resouce tracker had a problem | |
| 14:15:08 | sean-k-mooney | e.g. any that broke the update_available_resouces preiodic task ment the data in the hypervior api was out of date | |
| 14:15:39 | sean-k-mooney | server list on the other hand will always be correct | |
| 14:15:57 | sean-k-mooney | if you do no hit the server detail api then i think it also only hits the api db | |
| 14:16:00 | sean-k-mooney | so its fast | |
| 14:25:51 | tobias-urdin | for now i'll just lazy guess with len(allocations) when hitting /resource_provider/<uuid>/allocations since the returned allocations doesn't include consumer_type like /allocations does | |