| chandankumar | sean-k-mooney: Hello, can you approve last 2025.1 backport https://review.opendev.org/c/openstack/cyborg/+/1009168, when get time ,thank you! | 11:48 |
|---|---|---|
| sean-k-mooney | done | 11:49 |
| chandankumar | thank you! | 11:49 |
| opendevreview | Merged openstack/cyborg stable/2025.1: device_profile: reject groups with no resources: key at POST https://review.opendev.org/c/openstack/cyborg/+/1009168 | 12:32 |
| chandankumar | Cyborg IRC meeting starts in less than 1 hr, Please add your agenda here: https://etherpad.opendev.org/p/openstack-cyborg-irc-meeting#L47 | 13:07 |
| chandankumar | #startmeeting cyborg | 14:01 |
| opendevmeet | Meeting started Tue Oct 27 14:01:22 2026 UTC and is due to finish in 60 minutes. The chair is chandankumar. Information about MeetBot at http://wiki.debian.org/MeetBot. | 14:01 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 14:01 |
| opendevmeet | The meeting name has been set to 'cyborg' | 14:01 |
| chandankumar | who all are around for todays' meeting? | 14:01 |
| chandankumar | #courtesy ping: sean-k-mooney, amoralej bogdando[m] jgilaber_ melwitt | 14:02 |
| *** jgilaber_ is now known as jgilaber | 14:02 | |
| jgilaber | o/ | 14:02 |
| chandankumar | in the mean time, feel free to add the agenda here: | 14:02 |
| chandankumar | #link https://etherpad.opendev.org/p/openstack-cyborg-irc-meeting#L47 | 14:02 |
| sean-k-mooney | o/ | 14:03 |
| melwitt | o/ | 14:04 |
| chandankumar | let me start with the first topic | 14:04 |
| chandankumar | #topic cutting new release for stable branches | 14:04 |
| chandankumar | I have added this topic | 14:04 |
| chandankumar | In the last meeting we were talking about 1010931: [cyborg] Transition 2025.1 Epoxy to Unmaintained | https://review.opendev.org/c/openstack/releases/+/1010931 | 14:05 |
| jgilaber | I saw that we merged the last backport today | 14:05 |
| jgilaber | thanks for the reviews there | 14:05 |
| chandankumar | moving 2025.1 epoxy to un maintained | 14:05 |
| jgilaber | I was preparing a patch to make a final release | 14:05 |
| chandankumar | yes, we wanted to do a final release | 14:05 |
| sean-k-mooney | jgilaber: ack can you take care of updating it and doing the release | 14:05 |
| chandankumar | Since we have merged backports in all of the stable branches, Do we want to cut a release of all stable branches? | 14:06 |
| chandankumar | except 2026.2 it is still new | 14:06 |
| sean-k-mooney | well i guesss we coudl do call but there may be other backports open | 14:06 |
| sean-k-mooney | we shoudl do 2025.1 | 14:06 |
| sean-k-mooney | the rest i do not object too but i was specificly looking at the minimus set of patche requried for merging it on 2025.1 | 14:07 |
| chandankumar | https://review.opendev.org/q/project:openstack/cyborg+branch:stable/2025.2 all merged | 14:07 |
| jgilaber | I was planning to do only 2025.1 now | 14:07 |
| jgilaber | and wait until milestone 1 to do the rest | 14:07 |
| chandankumar | https://review.opendev.org/q/project:openstack/cyborg+branch:stable/2026.1 - all merged | 14:07 |
| sean-k-mooney | ok m1 is fine | 14:07 |
| sean-k-mooney | if there is noting open we coudl do them now | 14:07 |
| jgilaber | sure I can do them now as well | 14:08 |
| chandankumar | milestone 1 is Nov 16 - Nov 20 based on https://releases.openstack.org/indri/schedule.html | 14:08 |
| chandankumar | thanks jgilaber! | 14:08 |
| chandankumar | #action jgilaber will do a release for stable branches. | 14:09 |
| chandankumar | that's it from myside on topic | 14:09 |
| chandankumar | Any question on this one before moving to next topic? | 14:09 |
| chandankumar | Moving to next one | 14:11 |
| chandankumar | #topic core team update | 14:11 |
| chandankumar | sean-k-mooney has added this topic | 14:11 |
| sean-k-mooney | ya, so there has been no responce to my core team update mail | 14:11 |
| sean-k-mooney | so after this meeting if there is no concern im going to clean up the core group | 14:12 |
| chandankumar | #link core membership email: https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/MGYIDZTFD7CSWRGEU2YFYTAEMWCVJ23J/ | 14:12 |
| sean-k-mooney | i plan to update teh maintainer guide with the ptg feedback and push that later today | 14:12 |
| sean-k-mooney | my aim is to get that into shap before the next meeting an merge it in the next week | 14:12 |
| sean-k-mooney | then next week ill start creating the project config patch to create the gerrit groups | 14:13 |
| sean-k-mooney | thats all i really had unless there are questions | 14:13 |
| chandankumar | #link 1008481: Document Cyborg maintainer roles and conventions | https://review.opendev.org/c/openstack/cyborg/+/1008481 | 14:13 |
| jgilaber | sounds good to me | 14:14 |
| chandankumar | +1 to above all | 14:15 |
| chandankumar | since no questions moving to next topic | 14:15 |
| chandankumar | #topic Reviews | 14:15 |
| chandankumar | first one | 14:16 |
| chandankumar | #link stable/2025.1 backports https://review.opendev.org/q/project:openstack/cyborg+branch:stable/2025.1+status:open | 14:16 |
| chandankumar | we already talked about 2025.1 backport, it is already merged so no action needed here | 14:16 |
| chandankumar | I have just added fpga series for review | 14:16 |
| chandankumar | #link https://review.opendev.org/q/topic:%22fpga_program%22 | 14:17 |
| chandankumar | Thank you sean-k-mooney, jgilaber, melwitt for the review | 14:17 |
| chandankumar | I have rebased 998004: Select the FPGA programming driver by control path | https://review.opendev.org/c/openstack/cyborg/+/998004 on top of 999938: agent: separate driver loading from driver initialisation | https://review.opendev.org/c/openstack/cyborg/+/999938 | 14:17 |
| chandankumar | I have also made fpga tempest tests generic to run under guard | 14:18 |
| sean-k-mooney | ack, ill try and review these this week. i think we were getting pretty close | 14:18 |
| sean-k-mooney | we have a few chagnes that are kind of stepping on each others toes | 14:19 |
| chandankumar | I have also dropped Dropped init_host() startup frm 998004 and we can add it in https://review.opendev.org/c/openstack/cyborg/+/1002256 | 14:19 |
| sean-k-mooney | so i woudl like to make some incremental progres | 14:19 |
| chandankumar | sure! | 14:19 |
| chandankumar | that's it from my side on this review | 14:20 |
| chandankumar | Any other reviews anyone wants to highlight here? | 14:20 |
| chandankumar | ok moving to next one | 14:21 |
| chandankumar | #topic Bugs | 14:21 |
| chandankumar | We donot have any new bugs reported this week on https://bugs.launchpad.net/openstack-cyborg/ and https://bugs.launchpad.net/python-cyborgclient | 14:22 |
| chandankumar | moving forward. | 14:23 |
| chandankumar | We are almost come to the end of the meeting | 14:23 |
| chandankumar | #topic Open Discussion | 14:23 |
| jgilaber | I just pushed the patch for the releases | 14:23 |
| jgilaber | https://review.opendev.org/c/openstack/releases/+/1012500 | 14:23 |
| sean-k-mooney | +1 | 14:23 |
| jgilaber | I proposed all by hibiscus as bugfixes | 14:23 |
| jgilaber | but for hibiscus I proposed a minor version for the removal of the api console script | 14:24 |
| chandankumar | Anyone wants to bring anyone thing. | 14:24 |
| jgilaber | which had some upgrade impact | 14:24 |
| jgilaber | any thoughts on that? | 14:24 |
| sean-k-mooney | it didnt work | 14:26 |
| sean-k-mooney | it was broken 2026.1 | 14:26 |
| sean-k-mooney | with the removal of eventlet i had indented to drop it there as well | 14:27 |
| jgilaber | I thought it worked, hence the upgrade impact flag on the patch | 14:29 |
| jgilaber | should we downgrade to bugfix then? | 14:29 |
| sean-k-mooney | nothign in the code path monkey patched | 14:29 |
| sean-k-mooney | so it would not hace actuly used eventlet adn the websever woudl not have workded properly | 14:30 |
| sean-k-mooney | there is no implemation of the webserver in oslo.service in threaded mode | 14:30 |
| sean-k-mooney | so this was just dead code removal | 14:30 |
| sean-k-mooney | https://github.com/openstack/cyborg/blob/master/cyborg/cmd/__init__.py#L27 | 14:31 |
| sean-k-mooney | we hardcode the treaded backedn since 2026.1 | 14:31 |
| sean-k-mooney | i just forgot to remove those fiels when doing that | 14:32 |
| sean-k-mooney | so ya we can consdier the relase a bugfix | 14:32 |
| jgilaber | ack, I'll change that no problem | 14:32 |
| sean-k-mooney | for what it sworht we also had no testing of that | 14:33 |
| sean-k-mooney | devstgack used the wsgi script and uwsgi | 14:34 |
| sean-k-mooney | not cyborg-api | 14:34 |
| sean-k-mooney | so it may or may not have worked before | 14:34 |
| *** rlandy_ is now known as rlandy | 14:34 | |
| sean-k-mooney | it probaly did but it was a very long time since it was actully tested in ci | 14:34 |
| sean-k-mooney | https://github.com/openstack/cyborg/commit/95515c5c4bde2280ae967bf35725a57858f9379b | 14:35 |
| sean-k-mooney | so ya the change was made 8 years ago to stop using it in the ci by default | 14:36 |
| jgilaber | that is a long time, we can consider it broken | 14:37 |
| sean-k-mooney | ok i think we can likely move on or wrap up | 14:40 |
| chandankumar | yes, | 14:41 |
| chandankumar | coming to last topic | 14:41 |
| chandankumar | #topic volunteer for next meeting | 14:41 |
| chandankumar | Anyone one up to lead next cyborg irc meeting? | 14:41 |
| chandankumar | *anyone | 14:42 |
| jgilaber | I can do it | 14:42 |
| chandankumar | thanks jgilaber | 14:42 |
| chandankumar | time to close the meeting | 14:43 |
| chandankumar | thank you everyone for joining! | 14:43 |
| chandankumar | #endmeeting | 14:43 |
| opendevmeet | Meeting ended Tue Oct 27 14:43:24 2026 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 14:43 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/cyborg/2026/cyborg.2026-10-27-14.01.html | 14:43 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/cyborg/2026/cyborg.2026-10-27-14.01.txt | 14:43 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/cyborg/2026/cyborg.2026-10-27-14.01.log.html | 14:43 |
| opendevreview | chandan kumar proposed openstack/cyborg master: [DNM] opendev github-check testing https://review.opendev.org/c/openstack/cyborg/+/1012510 | 14:51 |
| melwitt | sean-k-mooney: hey I was reading through some PTG etherpads and wondered what did you mean by "moving nova interacting with cyborg via the sdk proxy api" https://etherpad.opendev.org/p/cyborg-2027.1-ptg#L54 ? | 15:07 |
| sean-k-mooney | melwitt: so in last cycle you moved cybrog to use the sdk but didnt use the proxy method | 15:08 |
| melwitt | ohh | 15:08 |
| sean-k-mooney | you just move to using th raw rest client with micoverions | 15:08 |
| sean-k-mooney | so if we have time it woudl be nice to do that | 15:08 |
| melwitt | ok I see. I thought I only did that where we lack proxy methods but I need to go look through. I can help add proxy methods, I've been all over openstacksdk lately haha | 15:09 |
| sean-k-mooney | https://github.com/openstack/nova/blob/master/nova/accelerator/cyborg.py#L337-L411 | 15:10 |
| melwitt | this is what I have proposed for the microversion stuff https://review.opendev.org/c/openstack/openstacksdk/+/1011952 initially I had used pci-sim to test but had second thoughts that it would likely be considered overkill so I changed it to fake driver for testing bind | 15:10 |
| sean-k-mooney | i have not chekced but i suppect deleting and getting arq by uuid is ther | 15:10 |
| sean-k-mooney | in the sdk | 15:10 |
| melwitt | ok. if it is then that is my bad, sorry :( | 15:10 |
| sean-k-mooney | its not your fault we discussed this in the reviews and i asked you not to rewite everthign to minimise the impact on my changes | 15:11 |
| sean-k-mooney | but now we have 6 months to do it so we can | 15:11 |
| sean-k-mooney | well more like 4 ish | 15:11 |
| melwitt | ok sorry I forgot then /o\ | 15:11 |
| melwitt | yeah. I might do it soon while sdk is fresh in my mind | 15:12 |
| sean-k-mooney | cool tag me in and im happy to review on both sides | 15:12 |
| melwitt | I'm slowly building some actual familiarity haha | 15:12 |
| sean-k-mooney | well i guess all 3 | 15:12 |
| sean-k-mooney | using the pci-sim may be desirabel later | 15:13 |
| sean-k-mooney | but the fake driver is also ok | 15:13 |
| melwitt | oh really? I had predicted you would not like it for sdk job | 15:16 |
| melwitt | I convinced myself pci-sim is generally reserved for cases that will inspect deeper such as nvme or mdev series | 15:16 |
| sean-k-mooney | its pretty cheap to compile and enable | 15:18 |
| sean-k-mooney | so we can test the api serfucen more or less with the fake driver | 15:18 |
| sean-k-mooney | so we can start with that | 15:18 |
| sean-k-mooney | if we have a need to test more we can consider pulling in the pci-sim | 15:19 |
| sean-k-mooney | i woudl only add it if a test needs it | 15:19 |
| sean-k-mooney | the main downside to the fake driver is its only updating the db | 15:19 |
| sean-k-mooney | so there is little benifit to testign booting a vm with a fake device other then testing the bidning flow | 15:20 |
| sean-k-mooney | that fine but if we were to say do resize | 15:20 |
| sean-k-mooney | or cold/live migrate | 15:20 |
| sean-k-mooney | then we proably woudl want the pci-sim at that point | 15:20 |
| sean-k-mooney | sicne you cant do that today we dont really need it today | 15:21 |
| melwitt | yeah, makes sense | 15:21 |
| melwitt | agreed it's cheap to compile and enable. initially I had thought it would be required for testing binding but realized it isn't after I had done with pci-sim | 15:21 |
| melwitt | pci-sim ran so fast though, it did not seem to add any cost. the only downside is requiring the debian nodeset but soonish I'm guessing we will have ubuntu 26.04 available in nodeset which will also work obvs | 15:23 |
| melwitt | for running locally I used 26.04 | 15:23 |
| sean-k-mooney | it works on 26.04 | 15:23 |
| sean-k-mooney | it also works on c10s | 15:23 |
| sean-k-mooney | and proably 24.04 but we never tested that | 15:23 |
| sean-k-mooney | the inital version was writen ageinst the master kernel | 15:24 |
| sean-k-mooney | which was roughly 7.0.0.rc3 i think | 15:24 |
| sean-k-mooney | so i have tried not to keep it fairly moderen | 15:24 |
| sean-k-mooney | and only add the minimal supprot we needed for older distos | 15:24 |
| sean-k-mooney | but ya 26.04, debian 13 or c10s shoudl all work today and all 3 are aviabel in ci | 15:25 |
| melwitt | oh really? LLM thought 24.04 wouldn't work which I thought was based on kernel version but admittedly I did not check myeslf | 15:25 |
| sean-k-mooney | it might work because of the c10s supprot we added but no one tested it | 15:26 |
| sean-k-mooney | it didnt work before that | 15:26 |
| melwitt | oh I didn't realize 26.04 was already available, I guess it's just not the default | 15:26 |
| sean-k-mooney | correct, i added the devstack supprot back around june | 15:26 |
| sean-k-mooney | we are using it in nova-next already | 15:27 |
| melwitt | whoa nice | 15:27 |
| sean-k-mooney | we could move cyrbog to it if we wnated but debian 13 has been workign fine | 15:27 |
| melwitt | are nodesets subject to any scarcity or no? I didn't know how that worked | 15:28 |
| sean-k-mooney | not that im aware of | 15:28 |
| sean-k-mooney | i think we have it aviabel on all the providers | 15:29 |
| melwitt | ok, that's good to know | 15:29 |
| sean-k-mooney | for both debian and 26.04 | 15:29 |
| melwitt | I wasn't sure the connection if any to the providers | 15:29 |
| sean-k-mooney | in case your wonderign this is where the supprot for multipel kernels lives https://github.com/openstack/cyborg/blob/master/pci-sim/fake_pci_sriov_compat.h | 15:29 |
| sean-k-mooney | its based on kernel flags rather then versions | 15:29 |
| sean-k-mooney | i just dont knwo if 24.04 passes those checks or not | 15:30 |
| melwitt | ah I see. tha makes sense | 15:30 |
| melwitt | chandankumar: just a heads up that I realized we don't have 100% parity with python-cyborgclient due to the default microversion behavior (python-cyborgclient defaults to latest microversion but OSC does _not_), until the following work is merged and released in SDK https://review.opendev.org/c/openstack/openstacksdk/+/1011952 | 15:53 |
| chandankumar | melwitt: thank you!, I will block you downstream patches until openstacksdk changes merged and released. | 15:54 |
| chandankumar | s/you/my | 15:54 |
| sean-k-mooney | we shoudl not be depending on the microvsrion defaultignin our ci either way | 17:14 |
| sean-k-mooney | not in ci or other determinsitc scripts | 17:15 |
| melwitt | true | 17:48 |
| melwitt | just wanted to be transparent about the remaining gap between the in-tree vs plugin behavior | 17:49 |
| sean-k-mooney | i reviewed you sdk change overall it looks corect but we have a bug we shoudl fix in the cybrog side | 17:49 |
| sean-k-mooney | well the gap is really on the sdk side | 17:50 |
| sean-k-mooney | osc just delicates in anycase | 17:50 |
| melwitt | yes, I mean as perceived to a CLI user | 17:50 |
| melwitt | hm I seem to be missing gerrit notifications for certain things. didn't get one for the openstacksdk review | 18:16 |
| sean-k-mooney | there might be a place to filter that in gerrit | 18:26 |
| sean-k-mooney | btu im not sure | 18:26 |
| sean-k-mooney | https://review.opendev.org/settings/#Notifications | 18:26 |
| sean-k-mooney | i only have os vif and netowrking-ovs-dpdk there | 18:27 |
| sean-k-mooney | so i think thsi is for extra notificatons | 18:27 |
| sean-k-mooney | when your not on cc or reviewer lines | 18:28 |
| melwitt | I think it may be connected to the "wip" mode, I had started that patch in wip mode and then changed it to active when ready | 18:44 |
| sean-k-mooney | oh maybe i never use that feature | 19:01 |
| sean-k-mooney | i tend to find it harmful because it has a bunch of weird edgecases | 19:01 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!