Tuesday, 2026-06-09

opendevreviewchandan kumar proposed openstack/cyborg-tempest-plugin master: Add scenario test for PCI device discovery and profile creation  https://review.opendev.org/c/openstack/cyborg-tempest-plugin/+/99221007:10
opendevreviewchandan kumar proposed openstack/cyborg master: Add NVMeVirt emulated NVMe device support for devstack  https://review.opendev.org/c/openstack/cyborg/+/97708307:16
opendevreviewchandan kumar proposed openstack/cyborg-tempest-plugin master: Add scenario test for PCI device discovery and profile creation  https://review.opendev.org/c/openstack/cyborg-tempest-plugin/+/99221009:13
chandankumarCyborg team meeting starts in another 53 mins, Feel free to add your agenda here: https://etherpad.opendev.org/p/openstack-cyborg-irc-meeting#L4213:06
sean-k-mooneychandankumar: thansk for taking a look at my cyborg tempest fixes by the way13:10
sean-k-mooneyi have one minor bug to fix on the cybrog side but i think im more or less feature complete on the srbac work13:11
sean-k-mooneyobvioulys we still need to review an approve the spec but i have a poc of it workign now with test more or less end to end13:11
chandankumarI have not looked at the srbac tempest tests, only focusing on ruff patches13:35
sean-k-mooneyyep i got annored by the fact i had to manually fix pep813:42
sean-k-mooneyalso i ment ot do that a few months ago13:42
sean-k-mooneybut i the cve work took precendce13:43
sean-k-mooneyi still need to follwo ups wiht the client and some of the watcher repos but its not very hard to do13:43
chandankumarHello all, Team meeting time14:00
rlandyo/14:00
chandankumar#startmeeting cyborg14:00
opendevmeetMeeting started Tue Jun  9 14:00:59 2026 UTC and is due to finish in 60 minutes.  The chair is chandankumar. 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'14:00
chandankumarwho all are aroud?14:01
chandankumaro/14:01
chandankumar#courtsey ping sean-k-mooney dviroel amoralej bogdando[m] jgilaber_14:01
sean-k-mooneyo/14:01
rlandyo/14:01
jgilaber_o/14:01
*** jgilaber_ is now known as jgilaber14:01
chandankumarHere is the link to etherpad https://etherpad.opendev.org/p/openstack-cyborg-irc-meeting#L42, feel free to add your topic here.14:01
chandankumarSince We donot have any topic to discuss I will directly jump to review topic14:04
chandankumar#topic Reviews14:04
chandankumar#link Remove glance image verification: https://review.opendev.org/q/topic:%22remove-glance-sig%2214:05
chandankumarIn last meeting, we discussed about dropping glance image verification code and respective parameters14:05
chandankumarI broke the patches into 3 patches14:05
chandankumarhttps://review.opendev.org/c/openstack/cyborg/+/991027: Remove broken image signature verification14:06
chandankumarhttps://review.opendev.org/c/openstack/cyborg/+/991598: Use GenericDriver.update() as the FPGA programming interface14:06
chandankumarhttps://review.opendev.org/c/openstack/cyborg/+/991579: Fix FakeDriver to support FPGA programming in CI14:06
chandankumarand tested here: https://review.opendev.org/c/openstack/cyborg-tempest-plugin/+/991081: Add scenario test for FPGA programming with FakeDriver14:06
sean-k-mooneythe midle patch is still kidn fo incorrect14:07
sean-k-mooneythe fake driver shoudl not inherit form the fpga one14:07
sean-k-mooneyso really the thrid patch should be the secodn one14:08
chandankumarI pasted it in wrong order14:08
sean-k-mooneythe order you pasted is correct14:08
jgilabercan't we just skip the second patch in the chain?14:08
sean-k-mooneythe order you have them in gerrit is not14:08
sean-k-mooneyjgilaber: right14:08
sean-k-mooneywe shoudl not need https://review.opendev.org/c/openstack/cyborg/+/991579/214:09
chandankumarhttps://review.opendev.org/c/openstack/cyborg/+/99159814:09
chandankumarI will do that14:09
sean-k-mooneythe main think we need to make it work is https://review.opendev.org/c/openstack/cyborg/+/991598/2/cyborg/agent/manager.py14:09
sean-k-mooneyand oblviy uypdatign the fpga dirver to use update internally14:10
sean-k-mooneyim sure there are other places in teh manger we need to fix14:10
sean-k-mooneybut evenutlly we shoudl nto call any driver spcific method14:10
sean-k-mooneyonly the methods on the base driver14:10
sean-k-mooneyanyway the ci is green14:11
jgilaberthe last patch looks ok from a quick glance14:11
sean-k-mooneyand you updated tempest as well14:11
sean-k-mooneyso if you remove the midel patch i think we are good directionally14:11
sean-k-mooneyand can pick this up in gerrit14:11
jgilaber+1, since update is already in the base driver, this looks correct14:12
chandankumarhttps://github.com/openstack/cyborg/blob/master/cyborg/accelerator/drivers/modules/generic.py#L8214:12
sean-k-mooneyif you read the docsting/comment for it14:12
sean-k-mooneyits pretty clear that it was alwasy inteded for progrming the device14:12
chandankumarone more thing https://github.com/openstack/cyborg/blob/master/cyborg/accelerator/drivers/modules/generic.py#L82 and https://github.com/openstack/cyborg/blob/master/cyborg/accelerator/drivers/driver.py also14:12
chandankumarDo we want to keep both of the files? or keep one 14:13
sean-k-mooneyyes so https://github.com/openstack/cyborg/blob/master/cyborg/accelerator/drivers/driver.py14:13
sean-k-mooneydefiens the interface14:13
chandankumaryes14:13
sean-k-mooneywe may combine them into one file14:13
sean-k-mooneybut lets do that eperatly14:13
sean-k-mooney*seperatly14:13
chandankumaryes that would be much better14:14
chandankumarOk I will propose a patch after the meeting14:14
sean-k-mooneywell we need to decied on the direction14:14
sean-k-mooneymy linclition is to delet this entirly14:15
sean-k-mooneybut i want to look a the history and usage14:15
sean-k-mooneybefore we make a code change for it14:15
sean-k-mooneyid ont think https://github.com/openstack/cyborg/blob/master/cyborg/accelerator/drivers/modules/generic.py#L8214:15
sean-k-mooneyis currently used14:15
sean-k-mooneyso we will liekly delete the enfire https://github.com/openstack/cyborg/tree/master/cyborg/accelerator/drivers/modules directory14:16
sean-k-mooneybut as i said we need to investigate if this is infact dead code or not14:16
chandankumarhttps://review.opendev.org/c/openstack/cyborg/+/473186 - added the generic driver14:16
jgilaberit looks like not all drivers inherit from the base one e.g https://github.com/openstack/cyborg/blob/233f2a5b7e396c24bf15ae22f8cf64a480666f5c/cyborg/accelerator/drivers/gpu/base.py#L3014:17
sean-k-mooneychandankumar: yes but that generic dirver was lated remvoed14:17
sean-k-mooneyjgilaber: ya that also a problem kind of14:18
sean-k-mooneyit will  work but id dont want to chagne thie picemeal14:18
chandankumarI will look into the history14:19
chandankumarI can see two specs also https://github.com/openstack/cyborg-specs/blob/981405a8938ed786feaaf63ea1955aa7174a367b/specs/pike/implemented/cyborg-driver-proposal.rst#L8 and https://github.com/openstack/cyborg-specs/blob/981405a8938ed786feaaf63ea1955aa7174a367b/specs/train/implemented/cyborg-accelerator-driver.rst#L8 around generic driver14:19
sean-k-mooneychandankumar: ok 14:19
sean-k-mooneyso i have been hining at the fact that this might need a spec or mor thought then a quick bugfix14:20
sean-k-mooneyi.e. tha that you shoudl proably not focus on this right now14:20
sean-k-mooneylet me know what you find14:20
chandankumaryes correct14:20
chandankumarsure, I will dig into the history and open a bug around that14:20
chandankumarwe can take it from there14:20
sean-k-mooneythis was somithing i was plannign to look into closer to the end of the cycle ebfor ethe next ptg14:20
sean-k-mooneyi dont really expect use to make large changes to this ebfore then14:21
sean-k-mooneybasiclly i want to look at propsoign a new generic driver framework for next chcel14:22
sean-k-mooneyand prot the exitign fucntionaltiy as part fo that work14:22
chandankumaryup, I will keep it to finding, you can take it from there14:22
sean-k-mooneycool so i think we can move onto jgilaber spec14:23
chandankumaryes14:23
chandankumar#link https://review.opendev.org/c/openstack/cyborg-specs/+/982276: Add generic mdev driver spec for 2026.214:23
chandankumarjgilaber: I have added your spec since you made changes to it14:23
chandankumarAnything you want to highlight from last revision14:24
jgilaberthanks for the reviews on the spec14:24
jgilaberthere is one thing worth pointing out I think14:24
jgilaberwhile going through sean-k-mooney comments there is one that made me doubt, let me find the link14:25
jgilaber#link https://review.opendev.org/c/openstack/cyborg-specs/+/982276/comment/b5f2b851_3cfdee76/14:25
jgilaber#undo14:25
jgilabersorry, I though I pasted the wrong link14:25
sean-k-mooneythe deplicate device coment14:26
jgilaberyes14:26
sean-k-mooneyya so in the medium term the conducrot shoud nto eb talkign to palcement at all14:26
sean-k-mooneythe cybrog compute agents should14:26
jgilaberit looks to me that currently in cyborg we cannot currently have a deployable with more than one resource class14:26
jgilaberwhich we might want to do to use multiple mdev classes from one device14:26
sean-k-mooneywell that proably ok14:27
sean-k-mooneywhat we need to model is as follows14:27
sean-k-mooneywe need 1 placement resouce provider per PF14:27
sean-k-mooneyand we need an inventory of mdev benetat it14:28
sean-k-mooneyso each mdev will be a deployable right but there wotn be a 1:1 mapping between deployabels and resouce providers14:28
sean-k-mooneyi need to look into that in more deail myself to see how this is currently working14:29
sean-k-mooneyone thing that would help is some ascii diagrams in the spec14:29
sean-k-mooneyto visually show how this works14:29
sean-k-mooneywe have 2 related concepts in cybrogs api devices and deployables14:30
jgilabergood point I'll add some diagrams, it was confusing for me too14:30
jgilaberbut if I understood correctly, in cyborg the resrouce provider is created from a DriverDeployable14:30
jgilabernot from a DriverDevice14:31
sean-k-mooneyso the only way we woudl supprot more then one mdev type per phsyica device14:32
sean-k-mooneyis if we were addign the VFs14:32
sean-k-mooneywith that said we coudl aslo suffix the pci adress in the rp name14:33
sean-k-mooneywith the type14:33
sean-k-mooneyso i think we coudl workaroudn that limiation14:33
sean-k-mooneybut we may want to think about the modelign more14:33
sean-k-mooneyin nova we started with 1 type per device14:33
sean-k-mooneyso that woudl eb a reasonable limiation fro v114:33
sean-k-mooneythe nvidia driver didnt supprotr multipel asfar as im aware either14:33
jgilaberthat is my understanding yes, the simplest workaround would be one DriverDeployable for each mdev type14:34
sean-k-mooneycan you write up some of the options in the spec14:34
sean-k-mooneyand we can discuss the tradeoffs14:34
jgilaberthe nvidia driver just picks the first vgpu type it finds and uses that to create the rp14:34
sean-k-mooneyi see14:34
jgilabersure, I already have some of this but I'll expand with the diagrams14:34
sean-k-mooneythat less then useful14:35
sean-k-mooneythat comemnt was it picking the first one14:36
sean-k-mooneyjgilaber: did you have any other questions you wanted to raise14:36
jgilaberno, thanks, the rest of the feedback was clear and I think I addressed it all14:37
chandankumarthank you jgilaber 14:37
chandankumarmoving to last review14:37
chandankumar#link SRBAC: https://review.opendev.org/q/topic:%22bp/consistent-and-secure-rbac%2214:37
chandankumar#link srbac spec https://review.opendev.org/c/openstack/cyborg-specs/+/99193214:38
chandankumarsean-k-mooney: do you want to take it14:38
sean-k-mooneysure14:38
sean-k-mooneyso this is one of 2 specs im planing to work on this cycle14:38
sean-k-mooneypart of the work was already compelted as part fo the CVE fixes14:38
sean-k-mooneythat focused on geting a secure baselien for api policy14:39
sean-k-mooneythe spec covers the 3 phases of the srbac comunity goal14:39
sean-k-mooneyit intnet for this cycle ist to supprot all standard roels 14:39
sean-k-mooneyadmin, manager, member, reader and service14:39
sean-k-mooneyi have mostly implemtned the spec already14:39
sean-k-mooneyincluding tmpest tests14:40
sean-k-mooneybut i woudl liek to ask for folks to review the spec14:40
sean-k-mooneywith attention pated to the polices/personas mapped to each api endpoint14:40
sean-k-mooneybasiclly ask does it make sense for this persona to eb able to call this api14:40
sean-k-mooneyi have 1 small bug in the cybrog patches which isll fix today/tomorrow14:41
sean-k-mooneyits technical debt form the cve fixes14:41
sean-k-mooneythat i want to think about a little more before fixing14:41
sean-k-mooneythe simple fix is trivali but im thinkign fo cleaning up a little more tech debt while im there14:41
sean-k-mooneyany questions on the effort or in general?14:41
jgilabernot really, I've added the spec to my review list, I hope to get to it by the end of the week14:42
chandankumarthank you for all the patches and spec, will go over the spec and have feedback there14:42
sean-k-mooneyin the spec i detail the timelien as well. this cycle suprpot the new polices, 2027.1 make the new polices the default and 2027.2 delete the old code14:42
sean-k-mooneyoh before i forget14:43
sean-k-mooneyafter the patches on the tempest plugin i also did the ruff and pre-commit work for that repo14:43
sean-k-mooneyi coudl have doen that first but i had alrady written the patches :)14:44
sean-k-mooneyso if folk want me to do that first i can mvoe it to the front14:44
sean-k-mooneyone goal i have is to compelte that for all the cybrog and watcher repos14:45
jgilaberI think it's fine, we don't have a lot of open patches so it won't be a problem to rebase14:45
chandankumaranything is fine14:45
sean-k-mooneyok ill leave it where it is for now then14:45
sean-k-mooneythats all i had on that14:45
chandankumarthank you sean-k-mooney!14:46
chandankumarSince we donot have any bugs moving to open discussion14:46
chandankumar#topic Open Discussion14:46
chandankumarI see an item there 14:46
chandankumar#topic driver support level14:46
chandankumarI think sean-k-mooney has added it14:47
sean-k-mooneyya more an fyi form me14:47
sean-k-mooneyso im planning to start on the patch to add the supprot levels to our docs14:47
sean-k-mooneyeither later this week or next14:47
sean-k-mooneybut ill try and prepare that for review before the next meeting14:47
sean-k-mooneythat was basiclly it14:47
chandankumarthank you!14:48
chandankumarMoving to last one14:48
chandankumar#topic volunteer to chair next meeting14:48
chandankumarAnyone wants to chair next meeting?14:48
rlandyI can do it 14:48
rlandyhaven't done one of these in a while14:49
sean-k-mooneyworks for me, welcome back14:49
chandankumarthank you rlandy !14:49
rlandyno problem  14:49
chandankumartime to close the meeting14:49
chandankumarthank you all for attending!14:49
chandankumar#endmeeting14:49
opendevmeetMeeting ended Tue Jun  9 14:49:49 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)14:49
opendevmeetMinutes:        https://meetings.opendev.org/meetings/cyborg/2026/cyborg.2026-06-09-14.00.html14:49
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/cyborg/2026/cyborg.2026-06-09-14.00.txt14:49
opendevmeetLog:            https://meetings.opendev.org/meetings/cyborg/2026/cyborg.2026-06-09-14.00.log.html14:49
jgilaberthanks chandankumar 14:49
sean-k-mooneyjgilaber: to your point ealier https://github.com/openstack/cyborg/blob/master/cyborg/accelerator/drivers/pci/base.py#L21 14:51
sean-k-mooneymost of the dirver dont inherit form base14:51
sean-k-mooneyone of of the openquetion i have is do we want ot use a base clss or a protocol to defien the internfce14:52
sean-k-mooneyeither works and a common base class has its advangates as woudl a mixin for the common functionatliy14:52
sean-k-mooneyfor the new medev  dirver you shoudl inherit form https://github.com/openstack/cyborg/blob/master/cyborg/accelerator/drivers/driver.py#L1714:53
sean-k-mooneybut in genreal i think i want to spend some time thinking about restucture the repo a bit next cycel to normalise the code base a bit14:55
sean-k-mooneyfor exampel we shoudl not ahve any config stuff https://github.com/openstack/cyborg/blob/master/cyborg/accelerator/configuration.py outside fo https://github.com/openstack/cyborg/tree/master/cyborg/conf14:55
sean-k-mooneyand there shoudl not be any db code 14:56
sean-k-mooneyhttps://github.com/openstack/cyborg/blob/master/cyborg/accelerator/accelerator.py14:56
sean-k-mooneyoutside of https://github.com/openstack/cyborg/tree/master/cyborg/db14:56
sean-k-mooneybut i suspect at elast some fo this code is deadcode14:56
sean-k-mooneyso the dirver not ineritign form the base driver is just one of the peices fo tehcnial debt we need to clean up in the repo layout14:57
jgilaberI'll double check but I think I have it already in my spec, but yes either way it would work14:57
jgilaberit was unexpected to have that base class but not use it14:57
sean-k-mooneyya... so there si a lot of have completed code migration still in the repo14:58
sean-k-mooneyout side fo the work you and chandankumar are doing implanning to spend the rest fo the cycle doign these cleanups before the next ptg14:58
sean-k-mooneyso that we are in a better places to looks at larger architecurall isseus over the next cycle ot two14:59
sean-k-mooneyonce the image verifcation cleanup is done im plannign to work on https://bugs.launchpad.net/openstack-cyborg/+bug/2143943 assuming you or chandan doen pick it up by then15:00
sean-k-mooneyits on my list to start after im happy with the srback work adn have the inital poc of the pci mangaged mode changes so my eta for that is about 2-3 weeks15:02
jgilaberI'll be busy for a bit with the nova mtty and the the rest of the of the nova mdev work15:06
sean-k-mooneyya no worreis this was more an fyi15:06
sean-k-mooneyof what i plan to work on next15:06
sean-k-mooneyin think i cna have the 2 "feature" i plan to delver for the cycel done by the end of june or early july, then im plannign to swap to buring down technial debt adn reviewing ye're sutff for the rest of the cycle15:07
sean-k-mooneyjgilaber: speaking of you lates mtty run is currenly runnign temspt15:10
sean-k-mooneyits goign to fail on some unrelated tests15:10
jgilaberoh cool15:10
sean-k-mooneybut we should have results form it in the next hour or so15:10
jgilaberyeah it's going to fail for sure since the whitebox tests have check for valid pci address15:10
sean-k-mooneyits acutlly failing on unrelated live mgiration tests15:11
jgilaberI was testing the zuul config and see if I had the configuration right and then I push a patch to make that check accept mdev devices15:11
sean-k-mooneyso ill looint why ocne we have actul logs15:11
jgilaberack thanks!15:11
sean-k-mooneymy guess is the whitebox plugin may have confirued something tha tbrok live migrtion or some other unrelated issues but ill need to look at the nova-comptue logs to figure that out rather then just the tempest log stream15:13
sean-k-mooneyi know you didnt set the cpu config options in the josb def for whitebox but i dont recall what the devstack plugisn defaults are15:14
jgilaberyes I saw your comment I probably missed a few settings, I tried to add the minimal set to get the test to run and then iteratively add the rest15:15
jgilaberto make sure that I understood what each does15:15
sean-k-mooneyya once we have a full run we can look at the rendered config files and then see what we need to tweak15:15
sean-k-mooneyi dont thin that will be hard15:15
sean-k-mooneythe ohter option we have is to write new tests in tempest iteslf15:16
sean-k-mooneythat done need whitebox15:16
opendevreviewJoan Gilabert proposed openstack/cyborg-specs master: Add generic mdev driver spec for 2026.2  https://review.opendev.org/c/openstack/cyborg-specs/+/98227615:30
sean-k-mooneyjgilaber: so the live migtion fialed because the qemu process crashed17:06
jgilabersean-k-mooney, do you think it is related to the patch?18:01
sean-k-mooneyno18:02
sean-k-mooneyi think you jsut got unlucky18:02
jgilaberack, thanks18:02
jgilaberI pushed a new version because the tempest regex was wrong18:03
jgilaberand I saw no reference to the whitebox18:03
sean-k-mooneythat also failing i think but ill take a look when its done18:06
sean-k-mooneywe may want to disabel the other jobs while we are debuging this18:07
sean-k-mooneyjust so it does not us as much gate resouces and or take as long18:08
sean-k-mooneybut we can do that tomorrow18:08
jgilabergood point, I'll do that tomorrow18:14
blanson[m]\\\=-0987654321KL;'19:48

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