Tuesday, 2026-03-03

« Monday, 2026-03-02 Index Wednesday, 2026-03-04 »
opendevreviewBogdan Dobrelya proposed openstack/cyborg master: Filter out SR-IOV VF devices from NVIDIA GPU driver  https://review.opendev.org/c/openstack/cyborg/+/97872510:56
opendevreviewchandan kumar proposed openstack/cyborg master: Add NVMeVirt emulated NVMe device support for devstack  https://review.opendev.org/c/openstack/cyborg/+/97708312:12
opendevreviewsean mooney proposed openstack/cyborg master: Migrate project metadata to pyproject.toml  https://review.opendev.org/c/openstack/cyborg/+/97788612:40
opendevreviewsean mooney proposed openstack/cyborg master: Add ruff config and fix pre-commit hook failures  https://review.opendev.org/c/openstack/cyborg/+/97788812:40
opendevreviewsean mooney proposed openstack/cyborg master: Fix E402 in Alembic migrations and template  https://review.opendev.org/c/openstack/cyborg/+/97788912:40
opendevreviewsean mooney proposed openstack/cyborg master: Fix S311 and S603 ruff security violations  https://review.opendev.org/c/openstack/cyborg/+/97789012:40
opendevreviewsean mooney proposed openstack/cyborg master: Migrate to ruff-format + ruff lint; drop hacking plugin  https://review.opendev.org/c/openstack/cyborg/+/97789112:40
opendevreviewchandan kumar proposed openstack/cyborg master: Add NVMeVirt emulated NVMe device support for devstack  https://review.opendev.org/c/openstack/cyborg/+/97708312:43
jgilaberHi! The IRC meeting will start in ~20 minutes, feel free to add topics to the agenda https://etherpad.opendev.org/p/openstack-cyborg-irc-meeting#L4413:37
opendevreviewchandan kumar proposed openstack/cyborg master: Add NVMeVirt emulated NVMe device support for devstack  https://review.opendev.org/c/openstack/cyborg/+/97708313:50
jgilaber#startmeeting cyborg - March 3rd14:00
opendevmeetMeeting started Tue Mar  3 14:00:45 2026 UTC and is due to finish in 60 minutes.  The chair is jgilaber. Information about MeetBot at http://wiki.debian.org/MeetBot.14:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.14:00
opendevmeetThe meeting name has been set to 'cyborg___march_3rd'14:00
sean-k-mooneyo/14:00
jgilaberHi everyone, who is around?14:00
bogdando[m]o/14:01
jgilabercourtesy ping: amoralej rlandy chandankumar 14:01
jgilaberwe'll start shortly feel free to add any topics to the agenda14:01
jgilaber#link https://etherpad.opendev.org/p/openstack-cyborg-irc-meeting#L4414:01
amoralejo/14:01
chandankumaro/14:02
jgilaberfor now there are not topics, only a few bugs to triage. Does anyone have any last minute topic or do we skip directly to the bugs14:03
jgilaberah I see chandankumar is adding a topic now14:03
sean-k-mooneyi didnt prepare them this week14:03
chandankumarjgilaber: yes, that would be quick14:04
jgilaber#topic updates on nvmevirt devstack plugin14:04
jgilaber#link https://review.opendev.org/c/openstack/cyborg/+/97708314:04
jgilabergo ahead chandankumar 14:04
chandankumarI want to give an update on using https://github.com/snu-csl/nvmevirt to emulate nvme device in CI for nvme cyborg testing in devstack14:05
chandankumarRight now,  By just passing cpus flag while loading the kernel module, it lists the nvme device14:05
chandankumarwe do not need to pass isolcpus in grub config14:05
sean-k-mooneyack14:06
sean-k-mooneyso if we do that14:06
chandankumarWith this, it will help us to test in devstack having vms with 4 vgpus14:06
chandankumar*vcpus14:06
sean-k-mooneyhave you noticed any negitive impact if we run a vm on the emulated device14:06
sean-k-mooneywe do not really care about performace as long as it works well enough for the basic testing we woudl be doing14:07
sean-k-mooneyso if its stable then that woudl be the mvp14:07
chandankumarnot yet. I just wanted to make sure, cyborg recognize the device14:07
sean-k-mooneyack14:07
chandankumarand then we can test the workload on how it behaves14:07
chandankumarcurrent nvmevirt does not support any nvme cleanup functionlity14:08
chandankumar*currently14:08
chandankumarthat's the update from my side. thank you!14:08
sean-k-mooneyi will try and find time to test this but probly next week. if there is no objection i would prefer to wait until after RC1 and ideally until after we have 1 tempest test usign this before merging it14:08
sean-k-mooneyi dont know if i asked this already in teh review14:09
sean-k-mooneybut can you provide a sample local.conf for this 14:09
sean-k-mooneyas part of the patch14:09
chandankumarcurrent cyborg tempest plugin uses https://github.com/openstack/cyborg-tempest-plugin/blob/master/cyborg_tempest_plugin/tests/scenario/test_accelerator_basic_ops.py#L56 uses fake driver14:10
sean-k-mooneycorrect14:10
sean-k-mooneywe need a new test14:10
chandankumarI think this one needs some changes so that it should use the correct pci device14:10
sean-k-mooneyno14:10
chandankumaryes new test would be better14:11
sean-k-mooneythat one shoudl continue to use the fake driver14:11
sean-k-mooneybut we need to add new tests that actully work with real drivers14:11
chandankumarmaking sure the pci driver get listed with in the vm14:11
chandankumars/driver/device14:11
sean-k-mooneyi dont know if we want to discuss the approch now but yes sshing into the vms and listing the devices would be one option14:11
sean-k-mooneywe do somethign similar for cidner volume tests14:12
sean-k-mooneyso there is tempest supprot for that already14:12
sean-k-mooneyis there anyting else to raise on this topic today?14:13
chandankumarI will look on the cinder volume tests and will pass an update in future meeting14:13
chandankumarjgilaber: you can move to the next topic14:14
jgilaberack thanks chandankumar 14:14
jgilaber#topic RC114:14
jgilaberI think that is from sean-k-mooney 14:14
sean-k-mooneyah yes14:15
sean-k-mooneyso the actual deadlien is thrusday week but i plan to tag RC1 on firday assuming there are no critical bugs14:15
sean-k-mooneyonce rc1 is tagged the stabel branch will be created and master will be 2026.2 at that point14:16
sean-k-mooneydoes that sound reasonabel to folks?14:16
jgilabersounds good, I'll review the linters series later today14:16
sean-k-mooneythanks that proably a good segway to the reviews topic14:17
jgilaberyep14:17
sean-k-mooneythe first being the linters serise14:17
jgilaber#topic reviews14:17
jgilaber#link https://review.opendev.org/q/topic:%22linters%22+is:open+project:openstack/cyborg14:17
sean-k-mooneyso the last patch in that serisee complete the move to ruff for formating14:17
sean-k-mooneythe earleir patchs are enabling sets of linter and fixign the issues found14:18
sean-k-mooneyso the last patch is just whitespace formating change and makeign sure its enfoced by the pep8 job14:18
sean-k-mooneythe eailer patchs fix the lint issues found14:18
sean-k-mooneyslighlty off topic but we agreed that i will do the saem in watcher after RC114:19
sean-k-mooneybut i expect to update python-cyborg client and the tempest plugin after rc1 as well 14:19
sean-k-mooneythe lint changes for the other repos will not be part of the 2026.1 release as a result which is ok14:20
jgilaberany question/comment on these patches?14:20
sean-k-mooneyif there are any concerns please let me knwo 14:20
jgilaberthanks sean-k-mooney, moving to the next review14:21
jgilaber#link https://review.opendev.org/c/openstack/cyborg/+/97872514:21
jgilaberthis is a fix for https://launchpad.net/bugs/198738014:21
bogdando[m]it needs more work for migration paths14:22
sean-k-mooneyyep bogdando[m]  started working on ^14:22
sean-k-mooneybogdando[m]: it might actully function as is but we need to confirm what happens in those edgcases14:22
sean-k-mooneyand proably add some testing14:22
sean-k-mooneydirectionally the patch looks correct14:23
sean-k-mooneybogdando[m]: do you have access to test this on real hardware with this issue14:23
sean-k-mooneyi.e. a sriov capable nvida gpu14:24
bogdando[m]I am not sure there would be real migration paths, given that no functional deployments will be possible for VFs by nvidia driver until kernels support vfio variant driver?..14:24
sean-k-mooneyso that not the limitation exactly14:24
sean-k-mooneyit was possibel to passthough the vf it just would not function correctly in the guest14:25
sean-k-mooneymy conder is not for the workload14:25
sean-k-mooneyit for the cybrog service14:25
sean-k-mooneywe need to ensure that if ther is an allcoation in place that is handeled gracefully14:25
bogdando[m]we cannot filter out VFs in such a case?14:26
sean-k-mooneythat can be refusing the remove the resocue provider and deployable  until the allcoation is deleted with a loud error message14:26
sean-k-mooneybogdando[m]: we can but not the ones that are currently assigned14:26
bogdando[m]ack14:26
sean-k-mooneyso nova has a protection for this for the pci tracker14:26
sean-k-mooneyif you remove a device form the devspec14:27
bogdando[m]wrt HW to test, I could use hardware, but no way to install cyborg there yet14:27
sean-k-mooneywe will not remove it form the db until it is freeded by a vm that was assigned to it14:27
sean-k-mooneybogdando[m]: ack that is fine for now14:27
sean-k-mooneywhile we will eventually want to do more hadware based testing we can simulate the test env for the short term in unit tests14:28
sean-k-mooneyi assume the lspci data you used in the test came form a real system14:28
sean-k-mooneythats why i was asking if you had access14:29
bogdando[m]it looks real, yes14:29
sean-k-mooneyi didnt really have any other questions to ask 14:31
bogdando[m]simulating test env with minimal services to have placement db would be great14:31
sean-k-mooneyadding fucntional tests like nova has with the palcment fixture14:32
sean-k-mooneyis definetly soemting we shoudl build14:32
sean-k-mooneyjust not in that bug fix14:32
bogdando[m]right14:32
sean-k-mooneythe placement-fixture form placement provide a placement api backed by sqlite14:32
sean-k-mooneywhich you can use in your tests witouth actully deploying placmenet seperatly14:33
sean-k-mooneythat allows you to test the end to end behavior14:33
bogdando[m]so we can build the migration path like you noted w/o func tests yet in that patch, right?14:33
sean-k-mooneywhat i think we need to do is test how cyborg handels removal of placemetn rps and deployables in general14:34
sean-k-mooneyso we can test that with the pci driver for example14:34
sean-k-mooneythen based on that we can determin how to proceed with the nvidia driver14:34
sean-k-mooneythe management of the placment inventories and deployables shoudl not be driver dependnet in general but we will need to review how that works today14:35
sean-k-mooneyif we find that will break14:36
sean-k-mooneythen we may put the fix behind a config option for now14:36
sean-k-mooneyso that you can move forward with the simple fix which will for for 99% of deployments14:36
sean-k-mooneyand we can defer the upgrade question to a sperate patch14:36
bogdando[m]makes sense14:37
bogdando[m]I will fence that fix with a config knob then14:38
sean-k-mooneyack14:39
sean-k-mooneywe can discuss it more in the review14:39
jgilaberack thanks, any other comment on this review?14:40
jgilaberlet's move to bug triage then14:41
jgilaber#topic Bugs14:41
jgilaber#link https://bugs.launchpad.net/openstack-cyborg/+bug/194051114:41
sean-k-mooneyso this i need to build more context on14:42
sean-k-mooneythe api docs site is unfortunetlly down right now for me14:42
sean-k-mooneyopenstack accelerator arq create command14:43
sean-k-mooneyisue used to create the Accleerator resouce request, this is normally only done by nova14:43
sean-k-mooneyso humans shoudl not really call this directly14:43
sean-k-mooneyhttps://github.com/openstack/cyborg/blob/master/api-ref/source/arqs.inc14:44
jgilaberlooking at the api the command is not correct, but the client does have a "--group-id" option14:45
sean-k-mooneythe request looks somethign like this https://github.com/openstack/cyborg/blob/master/doc/api_samples/accelerator_requests/accelerator_requests-post-curl.json14:45
sean-k-mooneyand this is the responce https://github.com/openstack/cyborg/blob/master/doc/api_samples/accelerator_requests/accelerator_requests-create-resp.json14:46
sean-k-mooneyso the comemnt is that https://github.com/openstack/cyborg/blob/master/doc/api_samples/accelerator_requests/accelerator_requests-create-resp.json#L1414:46
sean-k-mooneyis always 014:46
sean-k-mooneywe likely shoudl create a repoducer14:48
sean-k-mooneybut i thinkw e need to better understand teh semantics of the api14:48
sean-k-mooneyby reviewing the related spec14:48
sean-k-mooneyi think for now we shoudl move on14:49
jgilaberagreed, not sure if it's a real bug or not, do we mark as incomplete or we leave it as is?14:49
sean-k-mooneylets mark it incomplete for now sure14:50
sean-k-mooneyi added the api tag14:50
jgilaberack thanks14:50
sean-k-mooneyit not clear if its a clinet or api bug14:50
sean-k-mooneyor neither14:50
jgilaberlet's move to the second one14:50
jgilaber#link https://bugs.launchpad.net/openstack-cyborg/+bug/194102514:50
sean-k-mooneyso this is a feature reqeust14:51
sean-k-mooneyits also in the wrogn repo14:51
sean-k-mooneyit shoudl be in the python-cyborg client repo14:51
sean-k-mooneyits valid i think to supprot this in the future14:51
sean-k-mooneyi have fixed that14:52
sean-k-mooneywhat shoudl we triage this ass14:52
sean-k-mooney*as14:52
sean-k-mooneywishlist since it a feature reqeust ro medium becuase its a partiy gap14:53
sean-k-mooneybetween the cli and api14:53
jgilaberI'm leaning towards wishlist for now14:53
jgilaberbut no strong opinion either way14:54
sean-k-mooneyworks for me14:54
jgilaberack, let's move to the third one14:55
jgilaber#link https://bugs.launchpad.net/openstack-cyborg/+bug/195488814:55
jgilaberthis is a doc request14:55
sean-k-mooneyso there im conflcited14:56
sean-k-mooneyonly nova and zun should do that directly14:56
sean-k-mooneyas bidnign and unbinding the acclerator via ARQs shoudl reqruei the service role14:56
sean-k-mooneywith that said if you used cybrog in standaloen deployment14:57
sean-k-mooneywithout either then it would not be a terible thing to document14:57
sean-k-mooneyim just not sure we want to supprot that usage mode14:57
sean-k-mooneyor at least not right now14:57
sean-k-mooneyso this feels low to wishlist if not invalid 14:58
sean-k-mooneyim all for adding documeition fo how the bidnign flow shoudl work14:58
sean-k-mooneyin the contibutor docs14:58
sean-k-mooneybut i dont think we shoudl be documenting a manual end user/admin procedure14:58
jgilaberI'm wondering if the intention was to document how to use nova with cyborg to bind accelerators?14:58
jgilaberI'm not sure how well that is documented right now14:59
sean-k-mooneyit could be interepeted as a requce to docuemtn the function as a serivce flow14:59
sean-k-mooneyi think we set this to incomplete and come back to it next week14:59
sean-k-mooneywe need to basiclly refine the scope of it14:59
jgilabersounds good15:00
jgilabersince we're at time, we can leave the last two bugs for the next meeting, any comment for the open discussion?15:00
jgilaberif not we can move to the final topic15:01
jgilaber#topic Volunteers to chair next meeting15:01
jgilaberany volunteer?15:01
chandankumarI can chair15:01
jgilaberthanks chandankumar!15:01
jgilaberthat is all for today, thanks all!15:01
opendevreviewchandan kumar proposed openstack/cyborg master: Add NVMeVirt emulated NVMe device support for devstack  https://review.opendev.org/c/openstack/cyborg/+/97708315:01
jgilaber#endmeeting15:01
opendevmeetMeeting ended Tue Mar  3 15:01:47 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:01
opendevmeetMinutes:        https://meetings.opendev.org/meetings/cyborg___march_3rd/2026/cyborg___march_3rd.2026-03-03-14.00.html15:01
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/cyborg___march_3rd/2026/cyborg___march_3rd.2026-03-03-14.00.txt15:01
opendevmeetLog:            https://meetings.opendev.org/meetings/cyborg___march_3rd/2026/cyborg___march_3rd.2026-03-03-14.00.log.html15:01
opendevreviewchandan kumar proposed openstack/cyborg master: Add NVMeVirt emulated NVMe device support for devstack  https://review.opendev.org/c/openstack/cyborg/+/97708315:03
sean-k-mooneyi tried to capture what we discussed on the last bug here https://bugs.launchpad.net/openstack-cyborg/+bug/1954888/comments/115:03
opendevreviewsean mooney proposed openstack/cyborg master: adopt isort  https://review.opendev.org/c/openstack/cyborg/+/97876815:11
sean-k-mooney^ is the last patch in the linter serise by the way. there may be other work we do in the future but i do not expect to make other changes in the near term15:11
opendevreviewBogdan Dobrelya proposed openstack/cyborg master: Filter out SR-IOV VF devices from NVIDIA GPU driver  https://review.opendev.org/c/openstack/cyborg/+/97872515:22
opendevreviewBogdan Dobrelya proposed openstack/cyborg master: Filter out SR-IOV VF devices from NVIDIA GPU driver  https://review.opendev.org/c/openstack/cyborg/+/97872516:01
opendevreviewBogdan Dobrelya proposed openstack/cyborg master: Filter out SR-IOV VF devices from NVIDIA GPU driver  https://review.opendev.org/c/openstack/cyborg/+/97872516:23
opendevreviewBogdan Dobrelya proposed openstack/cyborg master: Filter out SR-IOV VF devices from NVIDIA GPU driver  https://review.opendev.org/c/openstack/cyborg/+/97872516:33
opendevreviewBogdan Dobrelya proposed openstack/cyborg master: Filter out SR-IOV VF devices from NVIDIA GPU driver  https://review.opendev.org/c/openstack/cyborg/+/97872516:47
« Monday, 2026-03-02 Index Wednesday, 2026-03-04 »

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