Monday, 2026-09-07

opendevreviewKamil Sambor proposed openstack/nova master: Replace StaticallyDelayingWrapper with futurist DelayingExecutor  https://review.opendev.org/c/openstack/nova/+/99741007:12
opendevreviewHyungJoon Kim proposed openstack/nova master: Add regression test for bug #2166652  https://review.opendev.org/c/openstack/nova/+/100438808:17
gibistephenfin: Uggla: do we need to bump the min version of os-vif in nova to 5.2.1? https://github.com/openstack/nova/blob/master/requirements.txt#L5408:48
stephenfingibi: Unless we need some particular feature that was added recently, I don't think we have to. The packaging issues was only present in 5.2.0 which won't get dragged in by default10:06
stephenfin3.1.0 is nearly 4 years old https://pypi.org/project/os-vif/3.1.0/ There's a high possibility it doesn't actually work, but we could likely say the same about a lot of our minimums?10:07
stephenfinSo maybe we could bump that, or maybe we should just go and bump to the latest major version (X.0.0) for all our deps? idk10:08
opendevreviewHyungJoon Kim proposed openstack/nova master: Sort claimed PCI devices by host PCI address  https://review.opendev.org/c/openstack/nova/+/100441111:10
opendevreviewMerged openstack/os-vif master: Fix typos  https://review.opendev.org/c/openstack/os-vif/+/100432611:33
Ugglagouthamr, dansmith, melwitt well done for the virtiofs cold migration! 👍12:55
opendevreviewEmma Foley proposed openstack/placement master: [pyproject] Update setuptools config  https://review.opendev.org/c/openstack/placement/+/100441813:17
sean-k-mooneyUggla: an fyi we may need to fix one packaging issue in placement before we cut rc1 ^13:18
sean-k-mooneyor do an RC2 if we have already done that13:18
sean-k-mooneyemma basiclaly is portign the same fix that stephenfin did for nova https://github.com/openstack/nova/commit/3881260b1cb179cd0f8bb2a37282325a5f5fb24a13:18
sean-k-mooneyi also fixed this in cybrog watcher as well13:22
sean-k-mooneywe shoudl proably check all the pyproject.tomls for this13:22
sean-k-mooneylooks like os-vif might be impacted as well https://github.com/openstack/os-vif/blob/master/pyproject.toml#L40-L4613:23
sean-k-mooneystephenfin: ^ we shoudl fix that too ya?13:23
stephenfinideally yes13:24
stephenfinThough as I noted in the nova commit message you linked, setuptools currently provides a fallback. It just spews a lot of warnings, and there's a good likelihood the PyPA folks will remove it at some point (they love deleting code more than I do)13:25
sean-k-mooneyit was apprently breaking something downstream when we were buidlign placement 13:25
sean-k-mooneybut for the libs that less of a problem i think13:25
sean-k-mooneyso i think we cna fix it in os-vif and the other repos but its less impactful there13:26
sean-k-mooneyos-vif is also a bit weired in that we have 3 seperate packages13:26
sean-k-mooneythe fined shoudl still work fine13:26
sean-k-mooneybut its a littel diffent then placment or nova13:26
gibistephenfin: OK, so we don't need 5.2.1 os-vif as minimum just need to ignore 5.2.0 which was yanked I guess so it won't get pulled in13:27
stephenfinyeah, there's no reason for 5.2.0 to get pulled in13:28
stephenfinwe could yank if we wanted but 🤷13:28
gibicool then I think we don't need to risk RC1 to change the minimums now13:28
gibieven though I would like to have better minimums13:28
sean-k-mooneyi dont think we added any hard dep on the new version 13:28
gibiI just never had the time13:29
sean-k-mooneyit fixes some bugs13:29
gibias of the placement pyproject.toml change, it fails if placement is used as a git submodule 13:29
sean-k-mooneybut i dont think we are depending on a new fucntionaltiy although neutorn will be enabling a new feature by default13:29
gibiprobably to wide matching happens13:29
stephenfinsean-k-mooney: My comment was less about e.g. 5.1.x -> 5.2.x than it was about 3.1.x -> 5.2.x13:29
sean-k-mooneyright i dont hink we striclly need to go form 3.1.x to 5.x13:30
sean-k-mooneywe shoudl eventully do that13:30
sean-k-mooneyas next cycle neuton are removign supprot for disabling a feature that will require the new version in nova13:30
sean-k-mooneytoday you can configre neuton to work with the older os-vif on the nova side13:31
sean-k-mooneythey are makign the new feature the default this cycle13:31
sean-k-mooneyso it woudl be good to reaise our min sooner rather then later13:31
opendevreviewStephen Finucane proposed openstack/placement master: tox: Use constraints option  https://review.opendev.org/c/openstack/placement/+/99689313:49
opendevreviewStephen Finucane proposed openstack/placement master: Replace license classifier  https://review.opendev.org/c/openstack/placement/+/100442213:49
opendevreviewStephen Finucane proposed openstack/placement master: Migrate requirements to pyproject.toml  https://review.opendev.org/c/openstack/placement/+/100442313:49
opendevreviewStephen Finucane proposed openstack/placement master: [pyproject] Update setuptools config  https://review.opendev.org/c/openstack/placement/+/100441813:50
opendevreviewStephen Finucane proposed openstack/placement master: Replace license classifier  https://review.opendev.org/c/openstack/placement/+/100442213:50
opendevreviewStephen Finucane proposed openstack/placement master: tox: Use constraints option  https://review.opendev.org/c/openstack/placement/+/99689313:50
opendevreviewStephen Finucane proposed openstack/placement master: Migrate requirements to pyproject.toml  https://review.opendev.org/c/openstack/placement/+/100442313:50
sean-k-mooneystephenfin: i assume we will do most of ^ early next cycle13:55
stephenfinthe license classifer might be worth doing now, since that's another thing PyPA might pull at a moments notice, but we could backport it later also. The rest can wait13:56
sean-k-mooneythat the one i think we shoudl hold13:56
sean-k-mooneywell not hold13:56
sean-k-mooneybut it was nto suprpoted on older distos13:56
sean-k-mooneyright we had to bump to ubuntu 24.0413:56
sean-k-mooneyto have the string form supproted in our package tools13:56
sean-k-mooneyi dotn knwo if cents 10 stream ships a new enough setuptools?13:56
sean-k-mooneywe just swaped to usign it for pypi in our source to image builds so that hsoudl be fine13:57
sean-k-mooneyi just dont knwo about rdo13:57
sean-k-mooneynot that that really activly maintianed right now13:57
sean-k-mooneyso ya it may jsut be worth checkign the versoin requrieemnt for that before we proced with it13:58
sean-k-mooneythat said we have moved a numebr of things to the string already 13:58
sean-k-mooneyso that ship may have already sailed13:58
stephenfinsean-k-mooney: not quite. The issue was it requires setuptools >= 77.0.0 https://setuptools.pypa.io/en/latest/userguide/license_migration.html13:59
stephenfinand that only supports Python >= 3.913:59
sean-k-mooneyright but it was capped in rdo, i knwo we had to go fix that btu i dont recall if it was merged13:59
sean-k-mooneyi think it was but not entirly sure13:59
stephenfinthe change you're thinking of was that we had to move off Ubuntu 20.04 since that ships with Python 3.8 by default13:59
stephenfinmove the packaging job off that, more specifically14:00
sean-k-mooneyah it was https://review.rdoproject.org/r/c/deps/python-setuptools/+/5884414:00
sean-k-mooneystephenfin: for reason rdo was still using 65.5.114:00
sean-k-mooneybut we fixed that in may so all good14:01
stephenfincool14:01
stephenfinI mean, if you're talking strictly about packaging rdo could always carry a revert patch too14:01
sean-k-mooneystephenfin: no i was aware of that as well (moving off 20.04) but that was not the only place usign old things to build14:01
sean-k-mooneyin general i woudl not advise they ever do that14:02
sean-k-mooneybut yes14:02
sean-k-mooneyrdo shoudl not carrry patches to the code14:02
stephenfinThe distro setuptools version no longer matters for installs with pip since builds are isolated by default, so it would only be packaging (or the aforementioned "my Python is too old") scenarios that would break14:02
stephenfinagree, but the escape hatch is there14:03
stephenfinspeaking of escape hatches, looks like setuptools' docs aren't happy https://setuptools.pypa.io/en/latest/userguide/license_migration.html14:03
sean-k-mooneyyep, although i belive we have to disabel build isolation in some case for downsteam reasons14:03
stephenfinI could fix it but they ignore my issues so 🤷14:03
sean-k-mooneyspecifcly to use a fips complient python-cyrotography14:04
stephenfinnot bitter...14:04
stephenfinsean-k-mooney: you need a fips-complint cryptopgraphy during build??14:04
stephenfin*compliant14:04
sean-k-mooneyyes and in the final artifact14:05
sean-k-mooneyanyway htat out of scope fo upstream14:05
sean-k-mooneybut downstream python3-cyrptograph must come form the rpms for complance reasons as it need to be built with a specific fips certifed version fo openssl ectra14:06
opendevreviewribaudr proposed openstack/nova master: This adds an alias for Hibiscus  https://review.opendev.org/c/openstack/nova/+/100442414:09
opendevreviewribaudr proposed openstack/nova master: Add service version for Hibiscus  https://review.opendev.org/c/openstack/nova/+/100443014:24
opendevreviewribaudr proposed openstack/nova-specs master: Move Hibiscus implemented specs  https://review.opendev.org/c/openstack/nova-specs/+/100444215:29
UgglaReminder: upstream meeting in ~20mn15:42
gibiUggla: I added a topic to the open discussion agenda15:45
Ugglagibi, ok!15:46
* bauzas was just one click away from upgrading his DSM server until he remembered that his IRC bouncer is on that server :)16:00
Uggla#startmeeting nova16:00
opendevmeetMeeting started Mon Sep  7 16:00:47 2026 UTC and is due to finish in 60 minutes.  The chair is Uggla. Information about MeetBot at http://wiki.debian.org/MeetBot.16:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.16:00
opendevmeetThe meeting name has been set to 'nova'16:00
UgglaHello everyone16:00
fwieselo/16:00
bauzaso/16:00
samborko/16:01
tkajinamo/16:02
UgglaLet's start16:03
Uggla#topic Bugs (stuck/critical) 16:03
Uggla#info No Critical bug16:03
Uggla#topic Gate status16:04
Uggla#link https://bugs.launchpad.net/nova/+bugs?field.tag=gate-failure Nova gate bugs 16:04
Uggla#link https://etherpad.opendev.org/p/nova-ci-failures-minimal16:04
Uggla#link https://zuul.openstack.org/builds?project=openstack%2Fnova&project=openstack%2Fplacement&branch=stable%2F*&branch=master&pipeline=periodic-weekly&skip=0 Nova&Placement periodic jobs status16:04
Uggla#info Please look at the gate failures and file a bug report with the gate-failure tag.16:04
Uggla#info Please try to provide a meaningful comment when you recheck16:04
gibio/16:04
UgglaAnything new about the gate? 16:05
gibiI had to put the functional test stabiliziton on hold for now due to internal priorities. But we still have some unstability in the functional threading job16:05
gibiand I have in flight patches to work on 16:05
gibibut it need to wait a bit16:05
UgglaYep I know about one on my side.16:06
gibifeel free to ping me if you see a functional instability and I will help judging it if it needs a new tracker or if we already have one 16:06
Ugglaare we good for that topic ?16:07
Uggla#topic Release Planning 16:07
Uggla#link https://releases.openstack.org/hibiscus/schedule.html16:07
Uggla#info Nova deadlines are set in the above schedule16:08
Uggla         #info RC1 target week.16:08
* Uggla working on pre RC1 patches. (in progress).16:08
UgglaI should propose the highlight patch shortly.16:09
Uggla#topic Review priorities 16:09
Uggla#link New file for Hibiscus https://etherpad.opendev.org/p/nova-2026.2-status16:09
Uggla#info Starting: https://etherpad.opendev.org/p/nova-2026.2-status#L16 interesting bugs to review.16:09
bauzasack16:09
bauzasdon't hesitate to ping me, I have a microtime I can use for upstream reviews :)16:10
Ugglabauzas sure. Thanks!16:10
Uggla#topic Stable Branches16:10
bauzas(more time than the time it takes for the light in between Moon and Earth, but less than a Earth round)16:10
UgglaI know elodilles is on a business trip so I don't know if he can make it.16:11
Ugglaseems not, so lets move on or defer it later.16:12
gibiI think it is pretty late for Elod at the moment16:12
Ugglayep probably.16:12
Uggla#topic vmwareapi 3rd-party CI efforts Highlights16:12
fwieselHi, no updates from my side.16:13
Ugglafwiesel something you'd like to share?16:13
Ugglaok thanks. Welcome back!16:13
fwieselThanks!16:13
Uggla#topic Kamil's news about eventlet removal16:13
Ugglasambork, gibi something to report on that front ?16:13
samborkNot much since last week,  key fix: restored eager thread executor creation in service managers rather than lazily in thread_pool_factory.py (#1001572) .Meanwhile, patch 1001369 (native threading test fix) waiting for update. Separately, patch 997410 (review) was updated to use futurist.DelayingExecutor, though it will merge in next release16:14
samborkeom16:14
gibinothing to add form my side. I will continue stabilizing the functional tests when I get some time16:15
Ugglasambork, gibi I'll notify you about the HL patch as soon as ready, so we could iterate to report the meaningful changes.16:15
Ugglaskipping sdk for neutron topic as Layos is not available.16:16
Uggla#topic Confidential computing, status of SNP and TDX features16:16
UgglaI guess nothing new here. 16:16
tkajinamyeah16:16
gibiyeah16:16
antiahttps://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/1003958 16:17
gibioooh16:17
tkajinamoh nice16:17
gibiniiice16:17
antiaI took the chance to submit those 16:17
gibiantia: thanks I will check it 16:17
Ugglacool antia!16:17
antiagibi: Thanks!16:17
antianothing else from me16:18
Ugglawould you guys want to maintain this topic. I think we can remove it now?16:18
gibiUggla: I'm OK to drop it16:19
gibithanks everyone :)16:19
Ugglaor course we can speak about it if we need in open discussion16:19
tkajinamyeah I think we can remove it from the regular agenda.16:20
Ugglaand yes thanks for this great collaboration around those topics.16:20
Uggla#topic Bug scrubbing 16:20
Uggla#info down to 67 (-1).16:20
Uggla#link https://etherpad.opendev.org/p/nova-bug-triage-roster16:21
Uggla#link https://truc.uggla.fr/ to follow the trend.16:21
Uggla#info Next meeting (this week): [public] Upstream bug triage. Wednesday, Septembre 09th · 15:30 – 16:00 UTC. Video call link: meet.google.com/zjr-rxus-hzj16:21
Uggla#topic Open discussion 16:21
Ugglaon topic from gibi16:22
Uggla(gibi): I need technical guidance what to do with https://bugs.launchpad.net/nova/+bug/2166474 (Uggla: I will describe the case on the meeting)16:22
* Uggla giving the mic to gibi16:23
gibio/16:24
gibiso we have that bug, it describe a nasty race where nova allows two evacuation action to overlap on the same VM at the same time16:24
gibithen today I realized that we started allowing that couple of years ago16:24
gibiby allowing evacuation from a non empty task_state16:24
gibito support things like16:25
gibi1. compute with a VM dies16:25
gibi2. owner try to restart the VM, VM goes to powering_off task_state that never progresses16:25
gibi3. operator realizes that the compute is dead an wants to evacuate the VM16:25
gibito support this nova allows triggering evacuation with task_state not empty16:26
gibibut task_state is our lock to avoid two operation to overlap on the same VM16:26
gibiobviously I can develop a patch that allow evacuation to start only from non rebuilding* task_state16:27
gibibut that only solves the currently proven overlap16:27
gibiI'm a bit hesitant about all the other task states what should be allowed and what should not 16:28
gibiso ideas folks? 16:28
gibi(here or offline in the bug)16:28
bauzasI see16:28
bauzastook a bit of time to load the context16:28
bauzaswe relied on task_state for knowing whether we were having an user instance action16:28
bauzasbut given it was an issue if we want to evacuate, we removed the conditional16:29
gibiyeah16:29
bauzasbut now we dunno whether a current task state is because of an user action or because an evacuation is starting, right?16:29
gibiright, basically we don't know if a non empty task_state at the start of evacuation means an action that cannot progress because it is against the dead compute and therefore allowed to evacuate, or it measn an action that is progressing already on another compute for this VM like a previous evac16:30
bauzashmmm OK16:31
gibibut also a previous evac can be against the dead compute16:31
bauzas(thinking)16:31
gibiso in theory you might want to re-try evacuation as the previous evac is not progressing as that compute died during evac16:31
bauzasI guess we can't look at the VM state thn ?16:31
gibinot fully mapped the states, but I think you want to be able to evac from ACTIVE, STOPPED, ERROR16:32
bauzasand we don't have a Migration object when evacuating ?16:32
bauzasbecause we could look at it16:32
gibiwe have but we can have multiple unfinished migration object with type evac16:32
gibihaving a migration object does not mean the action of that migration can progress16:33
bauzashmmmm16:33
gibimaybe the dest or the source of the migration is dead and you want to evac :)16:33
gibiso a perfect solution is impossible I think16:33
gibiwe need to compromise some requirements16:33
tkajinamthe most safe approach might be add new api microversion to require task_state=None and require operators to do reset-state16:34
tkajinamthough I know that's not something operators may like16:34
gibiI will never ask an operator in our do to do a reset-state16:34
gibithat is like asking for a bigger problem because you had a small one :)16:34
gibis/in our do/ in our doc/16:35
tkajinamhmm ok. I agree16:35
gibiI mean I spent so many time in customer cases where the deployer used reset-state and created a big mess16:36
gibimaybe, a smarter reset-state...16:36
tkajinamI know :-)16:36
gibibut that is probably the same problem of deciding what action can be reset safely16:36
gibithe same task_state can be ignored in evac precondition as well16:36
bauzasare we already checking in the API evacuation API whether an evacuation is already run ?16:37
gibiwe checked in the past by looking at the task_state, but we removed that check effectively when we started allowing to evacuate from task_state=powering_off to support the above use cae16:38
gibicase16:38
bauzashttps://github.com/openstack/nova/blob/master/nova/compute/api.py#L574716:39
gibibtw task_state=power_off can happen *during* an evac, as we started evacuating to STOPed state in microvesrion 2.9516:40
bauzashere, we ask for a task state to be None, right? 16:40
gibibauzas: that task_state=None means do not check task_state, the code that expects an empty task_state is expected_task_state=(None,)16:40
gibior [None]16:41
bauzassorry, yeah16:41
bauzasso we could say that we expect no task state16:41
gibino we cannot as that breaks the evac from task_state=powering_off case 16:41
bauzasbut if we have an exception, we could look at the task state16:41
gibiwe support since 202216:41
bauzasI mean, if we get a task state here, we can look whether this is due to an already existing rebuild task16:42
gibithe problem is basically what task_state is allowed and what does not16:42
bauzasif so, we can say "mmmh, nope"16:42
gibiI don't have a list of allowable task_states16:43
gibiI don't even have a list of definietly rejectable task_states16:43
gibias rebuild* task_state could mean an ongoing evac, or a failed evac that you want to retry16:43
gibiwhere failed, is failed do the dest compute is dead16:43
gibis/do the/ do to/16:44
gibis/do the/ due to/16:44
bauzasoh right the problem is not only when concurrent evacuation requests, this is also about ensuring we can ask for evac again if it fails16:44
bauzasbut if the evac failed, then the instance is not in ACTIVE vm state, right ?16:45
bauzasI have to look at the code16:45
bauzasI don't think we move the instance back to ACTIVE if we fail on evac16:46
bauzas(compared to migrations for example)16:46
gibiI guess I can propose PTG topic :)16:46
gibior write a book about distributes systems :)16:46
tkajinamif evacuation "fails" normally it may be easy though if compute node dies in the middle then evacuation may get stuck16:46
tkajinamcan we check state of the dest compute node to determine whether the current evac can progress ?16:47
opendevreviewAshish Gupta proposed openstack/nova master: Fix init_host crash migration test under native threading  https://review.opendev.org/c/openstack/nova/+/100136916:47
gibibauzas: evac is possible from ERROR state :)16:48
bauzasdoh16:48
gibitkajinam: the dest compute is not well defined unfortunately16:48
bauzasgibi: if you write a book about distributed systems, you wouldn't be the first :)16:48
gibithere is instance.host that evacuate sets pretty late. There is the migration object dest host but we can have multiple unfinished evacuation migration object with different dests16:49
bauzasI guess this sounds a PTG discussion indeed16:49
tkajinamyeah16:49
gibibauzas: mine will solve it!16:49
gibi:)16:49
bauzashah then I can review16:49
bauzasprovided you're not proposing a state machine :)16:50
gibiyou will get a dedicate pre-release copy of it for sure16:50
gibi"a" state machine that is short sighted :)16:50
bauzasif someone says "etcd will solve that", I swear I will find you16:50
gibiOK I think I will propose a ptg topic16:50
Ugglaetcd will solve that :)16:51
bauzasI'll find you16:51
bauzasand I know your home16:51
gibithere are "better" (mean newer) trials than etcd :)16:51
Uggla;)16:51
gibianyhow I think I'm done here. mic drop16:51
bauzasgibi: so, are you asking for reviews about your WIP or do we punt that to PTG?16:51
gibibauzas: feel free to read the commit message of my WIP and my last comment on the bug16:52
bauzasit will require a substantial amount of context loading, but that sounds a nasty bug tho16:52
gibibut we won't have solution on RC1 week16:52
bauzasgibi: OK'll try16:52
UgglaTo my mind that sounds complex enough to discuss to the PTG.16:52
gibiso we can continue after H release16:52
gibiprobably on the PTG (depending of downstream push)16:52
Ugglaok16:53
bauzasmy concern is that a PTG discussion will probably require some prerequisite effort for context loading16:53
gibiI will create more context if this will end up being a PTG topic16:53
bauzascool16:53
gibiI cannot help loading it though16:53
Ugglaanything else to discuss in the last 5mn, if not I think we are done.16:54
gibiI don't yet have a remote exploit in your head16:54
bauzaswell, I think I know how we manage our VM and task states, but I'm more concerned by the quirks we have in code (all the nasty assumptions we have as code conditionals)16:54
gibi(sssh do not tell them)16:54
bauzasI could write a book about all nova API preconditions16:55
bauzas(that one would probably be new compared to some random book about distributed systems :D )16:55
Ugglaok so closing now.16:56
UgglaThanks for joining this meeting. Have a nice day/evening.16:56
Uggla#endmeeting16:57
opendevmeetMeeting ended Mon Sep  7 16:57:02 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)16:57
opendevmeetMinutes:        https://meetings.opendev.org/meetings/nova/2026/nova.2026-09-07-16.00.html16:57
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/nova/2026/nova.2026-09-07-16.00.txt16:57
opendevmeetLog:            https://meetings.opendev.org/meetings/nova/2026/nova.2026-09-07-16.00.log.html16:57
bauzasinteresting talks, thanks16:57
gibithanks folks16:58
gibiI'm dropping for today16:58
tkajinamthanks16:59
Ugglagibi, bauzas, gmaan, sambork, dansmith, melwitt, gouthamr, sean-k-mooney  https://review.opendev.org/c/openstack/releases/+/1004465 if you'd like to update/review the highlight17:10
sean-k-mooneywe could note the startup time improvment but that about it form me i think the prelude looks reasonabel as is17:16
*** mhen_ is now known as mhen18:07
*** nicolairuckel_ is now known as nicolairuckel18:09
opendevreviewMerged openstack/placement master: [pyproject] Update setuptools config  https://review.opendev.org/c/openstack/placement/+/100441819:21
opendevreviewMerged openstack/nova master: Parallelize per-node resource updates  https://review.opendev.org/c/openstack/nova/+/98067921:05
gouthamrUggla: Ty, looks great - except the name of the release :)23:59

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!