| *** mhen_ is now known as mhen | 02:12 | |
| *** elodilles_pto is now known as elodilles | 08:55 | |
| opendevreview | Lucas HENRY proposed openstack/nova master: Accept ws and wss as forwarded console origin protocols https://review.opendev.org/c/openstack/nova/+/1014034 | 09:28 |
|---|---|---|
| gibi | folks bauzas suggested to start collecting ideas of what edge cases we want to test exploratorily during the cclm work. So I created https://etherpad.opendev.org/p/nova-cclm and did a braindump into it | 09:43 |
| gibi | cc dansmith | 09:43 |
| bauzas | gibi: thanks for doing it | 09:54 |
| opendevreview | Merged openstack/nova master: Add Glance preferred stores support https://review.opendev.org/c/openstack/nova/+/1011470 | 10:15 |
| Uggla | gibi, dansmith would you like to have a upstream meeting dedicated topic regarding cclm. It would be cool to have a weekly status, similar on what we did for eventlet removal. | 10:23 |
| gibi | I'm supportive to mention something about cclm on each IRC meeting | 10:25 |
| *** sambork_ is now known as sambork | 10:31 | |
| tkajinam | I might be asking a dumb question, but what cclm stands for ? cross-cluster live migration ? | 10:58 |
| tkajinam | or maybe cross-cell live migration. | 10:58 |
| gibi | cross cell live migration | 11:05 |
| gibi | as in https://review.opendev.org/c/openstack/nova-specs/+/1008081/ | 11:05 |
| gibi | it just too long of an expression to always type it out | 11:06 |
| gibi | I think a week or so ago downstream the cclm accronym was born | 11:06 |
| opendevreview | Stephen Finucane proposed openstack/placement master: Remove use of oslotest in fixtures https://review.opendev.org/c/openstack/placement/+/1014310 | 12:16 |
| opendevreview | Stephen Finucane proposed openstack/placement master: Remove unused dependency https://review.opendev.org/c/openstack/placement/+/1014311 | 12:16 |
| opendevreview | Stephen Finucane proposed openstack/placement master: Add placement.fixtures package https://review.opendev.org/c/openstack/placement/+/1014312 | 12:16 |
| opendevreview | Stephen Finucane proposed openstack/placement master: Add ruff https://review.opendev.org/c/openstack/placement/+/1014313 | 12:16 |
| opendevreview | Stephen Finucane proposed openstack/placement master: Enable pyupgrade rules https://review.opendev.org/c/openstack/placement/+/1014314 | 12:16 |
| opendevreview | Stephen Finucane proposed openstack/placement master: Remove use of oslotest https://review.opendev.org/c/openstack/placement/+/1014315 | 12:16 |
| opendevreview | Stephen Finucane proposed openstack/nova master: Use same env for all functional jobs https://review.opendev.org/c/openstack/nova/+/1014329 | 13:30 |
| stephenfin | gibi: bauzas: sean-k-mooney: I believe that's needed to fix the placement gate ☝️ | 13:30 |
| opendevreview | Stephen Finucane proposed openstack/placement master: DNM: Test nova gate fix https://review.opendev.org/c/openstack/placement/+/1014330 | 13:31 |
| opendevreview | TAKAYUKI OKUDA proposed openstack/nova-specs master: Add virtio-vsock support to the libvirt driver https://review.opendev.org/c/openstack/nova-specs/+/1014331 | 13:31 |
| opendevreview | Kamil Sambor proposed openstack/nova master: Fix test discovery in placement functional job https://review.opendev.org/c/openstack/nova/+/1014096 | 13:31 |
| stephenfin | ah, I see sambork is already on the case | 13:32 |
| stephenfin | sambork: I've taken a slightly different tack to ensure the two envs are kept in sync. wdyt? https://review.opendev.org/c/openstack/nova/+/1014329 | 13:33 |
| opendevreview | TAKAYUKI OKUDA proposed openstack/nova-specs master: Add virtio-vsock support to the libvirt driver https://review.opendev.org/c/openstack/nova-specs/+/1014331 | 13:33 |
| sambork | @stephenfin, I honestly don't mind a different approach if it's working, especially since it looks more general | 13:39 |
| gmaan | gibi: thanks for starting etherpad, a few of the shutdown scenario can be covered by running existing graceful_shutdown_jobs on v2 API | 13:40 |
| stephenfin | sambork: if you're happy with that approach and want to rework yours, I can +2 :) | 13:40 |
| stephenfin | unless gibi wants to treat your +1 on mine as +2 and avoid the unnecessary dance 👀 | 13:41 |
| sambork | yeah, I thought that additional dance was unnecessary so gibi if you don't mind :) | 13:45 |
| gmaan | stephenfin: sambork I think setting OS_NOVA_DISABLE_EVENTLET_PATCHING explicitly is good as we want to remove it from base functional env | 13:47 |
| gibi | looking | 13:48 |
| gmaan | we discussed somwhere this cycle that we should run default unit/functional jobs in threading moe | 13:48 |
| stephenfin | gmaan: we also want to remove it from that env though? | 13:48 |
| stephenfin | we'll get that for free | 13:48 |
| gmaan | stephenfin: that env you mean in functional-without-sample-db-tests ? or functional? | 13:48 |
| stephenfin | in functional-without-sample-db-tests | 13:48 |
| stephenfin | iiuc it should be identical to functional but without sample DB tests (per the name)? | 13:49 |
| stephenfin | so if functional switches to threading, so should functional-without-sample-db-tests | 13:49 |
| gmaan | not sure if that will work as placement if faiing. so ok to keep functional-without-sample-db-tests=false in that env and rest everything run in default mode what service has | 13:49 |
| gmaan | stephenfin: make sense, but what is other option to fix the placement? | 13:49 |
| gibi | it is not placement specific. nova has tox target that is broken, it just happened that only placement related jobs calling that tox target | 13:50 |
| gmaan | i have not explored much on that but my point is not to reply on [functional]OS_NOVA_DISABLE_EVENTLET_PATCHING as that will be removed soon | 13:50 |
| stephenfin | gmaan: what gibi said | 13:51 |
| gmaan | yeah, i mean placement jobs | 13:51 |
| gibi | but when we remove OS_NOVA_DISABLE_EVENTLET_PATCHING we want to remove it from all jobs (except maybe one kept back on eventlet) | 13:51 |
| gibi | (for the next 4 monts) | 13:51 |
| stephenfin | right | 13:51 |
| stephenfin | gmaan: in case you need to be caught up, we're really just bikeshedding on whether to inherit `setenv` from functional or explicitly duplicate it | 13:52 |
| gibi | and 4 months from now we will rip it out with incredible joy | 13:53 |
| stephenfin | 🥳 | 13:53 |
| stephenfin | the inherit fix is here https://review.opendev.org/c/openstack/nova/+/1014329 I have a DNM patch for placement here https://review.opendev.org/c/openstack/placement/+/1014330 | 13:53 |
| stephenfin | gmaan: if you're happy to +2 now, I can stick the +W on it later once the DNM job has passed | 13:54 |
| gibi | just to confuse everybody I +2d both stephenfin's and sambork's patch now :) | 13:54 |
| gmaan | i see, it is about switching the mode so removal of OS_NOVA_DISABLE_EVENTLET_PATCHING from functional env and this env will work. | 13:55 |
| gmaan | I am good on either one. will check dnm result and +w thanks stephenfin gibi sambork | 13:55 |
| stephenfin | while I have your attention (🙃) I spotted this on https://review.opendev.org/c/openstack/placement/+/1014310 which I want so we can finally push and merge the placement changes in OSC | 13:56 |
| gibi | stephenfin: can I assume that it is a reasonable re-impelmentation of the oslotest logger fixture? | 13:58 |
| stephenfin | yeah, copy-paste basically | 13:58 |
| gmaan | stephenfin: ack, will check today | 13:58 |
| gmaan | i will +w stephenfin as sambork is also +1 there, hope it is ok for sambork ? | 13:59 |
| gibi | +2d | 13:59 |
| sambork | gmaan sure, that was the plan I think :D | 13:59 |
| gmaan | cool | 13:59 |
| stephenfin | OSC doesn't use oslotest (SDK and OSC both avoid oslo where possible because they're heavy) and I don't want to add it just for this, but PlacementFixture is very cool | 14:00 |
| * stephenfin wishes we could have a NovaFixture etc... | 14:00 | |
| sean-k-mooney | ^ same | 14:07 |
| sean-k-mooney | i kind of wish we had nova-lib | 14:07 |
| sean-k-mooney | and could put all the scheduler filters and weighers there too | 14:08 |
| sean-k-mooney | so taht all resuabel part of nova woudl live in nova-lib (including a placement style fixture) | 14:08 |
| sean-k-mooney | althoguh unlike neutron-lib | 14:09 |
| gibi | making PlacementFixture play nice with our DB fixture after the threading change was nasty. So I cannot even describe how scarry it sound to have a NovaFixture to maintain | 14:09 |
| sean-k-mooney | i proably would not split it into a seprate repo | 14:09 |
| sean-k-mooney | i would jsut move it into a lib dir and have 2 packages | 14:09 |
| sean-k-mooney | gibi: ack i understand | 14:09 |
| sean-k-mooney | gibi: we have been building a nova-simulator in watcher | 14:09 |
| sean-k-mooney | because we needed a equivlent to the placement-fixture for our functional tests | 14:10 |
| stephenfin | oh, it's not at all realistic (in Python at least), but one can dream | 14:10 |
| sean-k-mooney | :) | 14:10 |
| sean-k-mooney | anyway back to the orginal topic | 14:10 |
| sean-k-mooney | im not agisnt removeing the oslo-test dep in the palcement fixture | 14:11 |
| sean-k-mooney | i didnt quite sell the watcher team on using the placement fixture yet | 14:11 |
| sean-k-mooney | we have a very minimal simulator for out tests | 14:11 |
| sean-k-mooney | https://github.com/openstack/watcher/blob/master/watcher/tests/local_fixtures/placement_api_emulator.py | 14:12 |
| sean-k-mooney | https://github.com/openstack/watcher/blob/master/watcher/tests/local_fixtures/nova_api_emulator.py | 14:12 |
| sean-k-mooney | those can both be used with gabbi tests normal fucntional test and as deamons that you can have client talk to via http requests | 14:13 |
| stephenfin | gmaan: gibi: Any concerns with me backporting https://review.opendev.org/c/openstack/placement/+/1014310 once it merges (so I can consume it in OSC via a patch release)? | 14:14 |
| gmaan | I am fine backporting | 14:14 |
| *** gmaan is now known as gmaan_afk | 14:20 | |
| gibi | that is test only so I'm OK | 14:29 |
| dansmith | gibi: in case it's not clear, I'm kinda hoping for a quick skim of the non-test code here: https://review.opendev.org/c/openstack/nova/+/1013978/2 and an impression of whether or not you do think we should squash instance allocations into that | 14:34 |
| dansmith | if so I'll do it and actually look at the tests | 14:35 |
| dansmith | and if you know but just haven't had time, no worries, I just wasn't sure I was explicit enough | 14:35 |
| dansmith | no particular rush there | 14:35 |
| gibi | reading it now... | 14:41 |
| gibi | dansmith: responded there. bottom line those maintenance writes are scarry so I'm OK to just go with your first patch now | 14:49 |
| gibi | add +1 to the first patch (asking for the report.py change still) as well | 14:52 |
| gibi | s/add/added/ | 14:53 |
| dansmith | gibi: yeah that max() version thing was.... yeah :) tbh, I'm not sure why we don't coalesce everything to a current microversion for placement stuff because it feels like something we could track a lot closer and don't need to support wildly different placement versions | 14:53 |
| dansmith | like the list of versions at the top of report.py surprised me that it has grown that large | 14:53 |
| dansmith | it's us using our stable api to avoid changing stuff on the nova side when we add a new feature, but it leads to some debt for sure | 15:02 |
| dansmith | kinda like db migration compaction I guess | 15:02 |
| gibi | yeah I think it served a purpose during active placement development but it should be ripped out now and replaced with a single version across the board | 15:04 |
| Uggla | gmaan_afk, fyi reviewed 1007750: Reproduce part of bug/2166474 | https://review.opendev.org/c/openstack/nova/+/1007750 | 15:12 |
| *** gmaan_afk is now known as gmaan | 16:12 | |
| polyedre | gmaan: when you have time, could you take another look at https://review.opendev.org/c/openstack/nova/+/1009164 (bug 2169804)? I addressed your comments and dropped the executor shutdown move: the manager graceful shutdown now only starts once the main RPC server has drained. | 16:12 |
| polyedre | It sits on a small new change, https://review.opendev.org/c/openstack/nova/+/1010011, which stops the libvirt event threads so the default executor can shut down on compute (without it, nova-graceful-shutdown hangs until the timeout). Thanks! | 16:12 |
| gmaan | Uggla: thanks | 16:13 |
| gmaan | polyedre: yes, its was in my open tab but got distracted to other things. I will review it today | 16:14 |
| opendevreview | Chris Breu proposed openstack/nova master: libvirt: reserve device slots for new Cinder data volumes https://review.opendev.org/c/openstack/nova/+/1011008 | 16:28 |
| opendevreview | Chris Breu proposed openstack/nova master: libvirt: reserve device slots for new Cinder data volumes https://review.opendev.org/c/openstack/nova/+/1011008 | 16:28 |
| gibi | dansmith: (cc bauzas) I reproduced the cclm abort migration issue cleanly now. https://paste.openstack.org/show/bpg4J0Az6Na5xqNJmcbc/ | 16:45 |
| gibi | it seems that all OK with the domain itself | 16:46 |
| opendevreview | Merged openstack/nova master: Use same env for all functional jobs https://review.opendev.org/c/openstack/nova/+/1014329 | 16:46 |
| bauzas | ack | 16:46 |
| gibi | but nova for some reason puts both the migration and the instance to ERROR state | 16:46 |
| gibi | wrong link above | 16:47 |
| gibi | this is the correct one https://paste.openstack.org/show/bbELonjjRlOeYUCwx0Oz/ | 16:47 |
| dansmith | cool and not surprised there's still some stuff to work on there.. I've done nothing to address force or abort.. I expect the _actual_ abort part to work as it does today, just that the cleanup routines (which is most of what is changed) doesn't account for it | 16:49 |
| gibi | yeah that make sense | 16:52 |
| opendevreview | Merged openstack/nova master: Reproduce part of bug/2166474 https://review.opendev.org/c/openstack/nova/+/1007750 | 17:08 |
| gibi | dansmith: I also noticed that the new os-migrateLiveV2 action is accepted regardless of the microversion requested | 17:17 |
| gibi | is this intentional? | 17:17 |
| gmaan | it seems it is missing the @wsgi.api_version() decorator https://review.opendev.org/c/openstack/nova/+/1008839/2/nova/api/openstack/compute/migrate_server.py#184 | 17:26 |
| opendevreview | Takashi Kajinami proposed openstack/nova master: Kill remaining implementation for MIDONET https://review.opendev.org/c/openstack/nova/+/1014369 | 17:27 |
| gmaan | @wsgi.api_version("2.105") | 17:27 |
| dansmith | yeah just poc convenience for testing :) | 17:29 |
| gibi | ack :) | 17:29 |
| opendevreview | Takashi Kajinami proposed openstack/nova master: Drop support for midonet VIF type https://review.opendev.org/c/openstack/nova/+/1014369 | 17:33 |
| opendevreview | Takashi Kajinami proposed openstack/nova master: Remove support for IO Visor VIF Type https://review.opendev.org/c/openstack/nova/+/1014371 | 17:33 |
| opendevreview | Takashi Kajinami proposed openstack/nova master: Drop support for IO Visor VIF Type https://review.opendev.org/c/openstack/nova/+/1014371 | 17:33 |
| tkajinam | I wonder if we need formal deprecation for these "dead" code or can directly remove these | 17:34 |
| opendevreview | Takashi Kajinami proposed openstack/nova master: Remove Agilio OVS VIF Type https://review.opendev.org/c/openstack/nova/+/1014372 | 17:38 |
| opendevreview | Merged openstack/nova master: Fix evacuate race conditions with periodic task and task_state https://review.opendev.org/c/openstack/nova/+/999978 | 18:18 |
| sean-k-mooney | tkajinam: i have a patch up to do that already | 18:28 |
| sean-k-mooney | tkajinam: well sort of i starte reviewign this last year https://review.opendev.org/c/openstack/nova/+/957278 | 18:29 |
| sean-k-mooney | tkajinam: i went thoug all the one we supprot and start to add command about htere usage with the intent to remvoe/deprecate them in 2026.1 | 18:30 |
| sean-k-mooney | the pleit being if there was any recent release it woudl be deprecated and if not remvoed | 18:30 |
| sean-k-mooney | tkajinam: in the case of midonet and agilio ovs i think they can jsut be remvoed at this point | 18:31 |
| tkajinam | ah that's nice | 18:31 |
| sean-k-mooney | i think the saem is ture for iovisor | 18:31 |
| sean-k-mooney | so for those three i woudl be onbaord i suggest addign it to next weeks irc meeting | 18:32 |
| tkajinam | yeah that's exactly what I thought (adding it to next weeks irc meeting) | 18:32 |
| tkajinam | but tried to put some examples to gerrit before the discussion | 18:32 |
| sean-k-mooney | we could also move some of the logic out into os-vif as a followup | 18:33 |
| sean-k-mooney | i.e. supprot fo vif_type=tap | 18:33 |
| tkajinam | yeah I agree | 18:33 |
| sean-k-mooney | that may allow all or almost all the legacy code paths to go away | 18:34 |
| tkajinam | oh I didn't know neutron plugin for tungsten fabric is still alive. I've been checking https://github.com/tungstenfabric/tf-neutron-plugin but it has been dead for some time. | 18:34 |
| tkajinam | sean-k-mooney, yeah | 18:34 |
| sean-k-mooney | im happy to review that cleanup if other are ok with it since i ahve wanted to do it for years so feel free to add me too those reviews | 18:34 |
| tkajinam | I started digging into it after I found "MIDONET" there but later noticed there are some amount of items I can drop. | 18:35 |
| sean-k-mooney | tkajinam: it moved like twice since it last had any testing | 18:35 |
| tkajinam | sean-k-mooney, nice, and thanks. | 18:35 |
| tkajinam | ahhh ok I finally found it https://github.com/OpenSDN-io/tf-neutron-plugin | 18:38 |
| tkajinam | I hate these reblanding tings :-( | 18:40 |
| tkajinam | things | 18:40 |
| sean-k-mooney | oh sis ir move yet again | 18:42 |
| sean-k-mooney | *did it move | 18:43 |
| tkajinam | yeah | 18:51 |
| tkajinam | https://superuser.openinfra.org/articles/opensdn-24-1-the-next-evolution-in-open-source-sdn/ | 18:51 |
| fungi | Uggla: not urgent, but where are you collecting topics for the weekly meeting agenda, or how do i get a topic added for next week at least? thanks in advance! | 19:58 |
| frickler | fungi: see the /topic ;) https://etherpad.opendev.org/p/nova-meetings | 20:01 |
| fungi | d'oh! i should have thought to look there. thanks frickler! | 20:06 |
| opendevreview | Merged openstack/nova master: Remove the compute RPC 5.x server endpoint https://review.opendev.org/c/openstack/nova/+/1012320 | 20:12 |
| opendevreview | Merged openstack/placement master: Remove use of oslotest in fixtures https://review.opendev.org/c/openstack/placement/+/1014310 | 20:12 |
| opendevreview | Merged openstack/placement master: Remove unused dependency https://review.opendev.org/c/openstack/placement/+/1014311 | 20:12 |
| opendevreview | Dan Smith proposed openstack/nova master: Add consumer_type for migration placeholder allocations https://review.opendev.org/c/openstack/nova/+/1012083 | 20:20 |
| opendevreview | Dan Smith proposed openstack/nova master: Add consumer_type for migration placeholder allocations https://review.opendev.org/c/openstack/nova/+/1012083 | 20:42 |
| opendevreview | Dan Smith proposed openstack/nova master: WIP: Set consumer types universally for instances https://review.opendev.org/c/openstack/nova/+/1013978 | 20:42 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!