| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-cyborg - 2026-07-28 | |||
| 14:31:14 | sean-k-mooney | but we can discuss it in teh review | |
| 14:31:25 | jgilaber | yes, that would be a natural point as well | |
| 14:31:34 | sean-k-mooney | its partly because fo the release note https://review.opendev.org/c/openstack/cyborg/+/997050/7/releasenotes/notes/generic-mdev-driver-af107e247cb1c815.yaml | |
| 14:31:46 | sean-k-mooney | i dont really wanto to say "this is supproted" if only the cybrog part is done | |
| 14:31:54 | melwitt | I think that could be ok as long as the last patch demonstrates everything working end-to-end (ensure we will not need to backtrack) | |
| 14:32:18 | sean-k-mooney | melwitt: oh yes we can it just i want to make sure we have got some trackign on the nvoa side in paralle | |
| 14:32:26 | jgilaber | that's fair, I could move the depends-on there | |
| 14:32:38 | sean-k-mooney | no you dont need too | |
| 14:32:53 | chandankumar | I have added nova dependency as well as final patch where all are tested on review etherpad | |
| 14:32:57 | sean-k-mooney | this is more for use to have some check points | |
| 14:33:06 | sean-k-mooney | lets review this as normal | |
| 14:33:19 | sean-k-mooney | i jus tmay hold +w until we feel the nova changes are getting close | |
| 14:33:47 | sean-k-mooney | i think you have already demonstrated a lot fo this via the later patches | |
| 14:33:55 | melwitt | yeah I mean just saying that would be my criteria for merging the nova stuff first, seems ok idea to me as long as we have the end-to-end working demonstrated we know the nova stuff is 100% | |
| 14:34:18 | jgilaber | yes the last does have everything working in tempest tests | |
| 14:34:19 | melwitt | (that we won't need to change it) | |
| 14:34:27 | sean-k-mooney | melwitt: so on the nova part | |
| 14:34:29 | jgilaber | with the same lifecycle tests we have for pci devices | |
| 14:34:37 | sean-k-mooney | cybrog alrady has one driver that supprot mdev based gpus | |
| 14:34:49 | sean-k-mooney | and nova didnt merge teh nova change in the past | |
| 14:35:08 | sean-k-mooney | which is why i want to see us make some progress there as well | |
| 14:35:21 | melwitt | yes makes sense | |
| 14:35:22 | sean-k-mooney | so we dotn end up with 2 driver that cant work at the end of the cycle | |
| 14:35:40 | sean-k-mooney | by the way im not that worreid we have time | |
| 14:35:48 | sean-k-mooney | just trying to level set | |
| 14:36:04 | melwitt | are they all on the same gerrit topic? I find that helpful for not missing things personally | |
| 14:36:16 | melwitt | yeah, all good | |
| 14:36:52 | sean-k-mooney | the cybrog ones are the nova oens are on a diffent topic | |
| 14:36:52 | chandankumar | we have two topics https://review.opendev.org/q/topic:%22generic-mdev-driver%22 and https://review.opendev.org/q/topic:%22cyborg-vgpu-support%22 | |
| 14:36:55 | sean-k-mooney | we coudl use the same for both | |
| 14:37:09 | sean-k-mooney | but technially the bluepritn names are diffent | |
| 14:37:13 | sean-k-mooney | so two topic | |
| 14:37:43 | melwitt | we don't have to, was just saying I like using it to tie together patches for the same effort that are across multiple projects. just a personal thing I like to do helps me keep track of all together | |
| 14:38:16 | melwitt | https://review.opendev.org/q/topic:%22ipv6-multinode%22 is a good example haha | |
| 14:39:15 | melwitt | ok, matching blueprint names, no problem. ignore me :) | |
| 14:39:31 | sean-k-mooney | ya i like to do that too | |
| 14:39:45 | sean-k-mooney | its only an issue if you have 2 difefnt blurpints for it | |
| 14:39:58 | sean-k-mooney | then you have to chosoe wihc to use or use the hastag feature | |
| 14:40:13 | melwitt | yeah, I see | |
| 14:40:54 | chandankumar | https://review.opendev.org/q/topic:%22cyborg-vgpu-support%22+OR+topic:%22generic-mdev-driver%22 | |
| 14:41:10 | sean-k-mooney | that worsk as well :) | |
| 14:41:55 | chandankumar | any more questions on this review series | |
| 14:41:57 | sean-k-mooney | unles there are other queston on this we can likely move on | |
| 14:42:10 | jgilaber | yes, I think we can move one, thanks | |
| 14:42:13 | sean-k-mooney | ill try to review the nova parts in the next day or two | |
| 14:42:23 | chandankumar | Under Improvement section, we have guest reboot we already talked about that | |
| 14:42:24 | melwitt | that doesn't fit how I use it fwiw, bc I click on the topic from one of the patches to discover the others. but the link is helpful nonetheless :) | |
| 14:42:29 | chandankumar | Now jumping to bug | |
| 14:42:38 | chandankumar | #topic Bugs | |
| 14:42:56 | chandankumar | #link Devstack doesn't enable port_device_profile ML2 extension driver for Cyborg: https://bugs.launchpad.net/openstack-cyborg/+bug/2161365 | |
| 14:43:20 | chandankumar | It was opened by melwitt | |
| 14:43:23 | melwitt | I got some feedback on my patch for this, apparently not really a bug. so not sure if I should be thinking to close it? | |
| 14:43:36 | sean-k-mooney | whishlist is fine | |
| 14:43:43 | melwitt | ok | |
| 14:43:44 | jgilaber | +1 seems fine as is | |
| 14:43:48 | chandankumar | sean-k-mooney: chnaged it to wishlist | |
| 14:43:51 | sean-k-mooney | ist a mini enhancment | |
| 14:43:53 | melwitt | thanks | |
| 14:44:04 | sean-k-mooney | melwitt: since we are on this topic | |
| 14:44:04 | chandankumar | it is already triaged, thank you | |
| 14:44:24 | sean-k-mooney | i ased for some enhacned docs in teh review i left | |
| 14:44:42 | sean-k-mooney | but i was also thinking maybe we shoudl clean up https://github.com/openstack/cyborg/blob/master/cyborg.conf.intelnic.sample | |
| 14:45:04 | sean-k-mooney | its really weird to have a random config example at the root of the repo | |
| 14:45:16 | melwitt | ok sure | |
| 14:45:55 | sean-k-mooney | my prefence woudl be to add a local.conf exmaple instead https://github.com/openstack/cyborg/blob/master/devstack/local-conf.pci-sim.sample | |
| 14:46:03 | sean-k-mooney | but ya that basiclly what i said in the review | |
| 14:46:25 | sean-k-mooney | i am fine with adding a speicif driver doc later | |
| 14:46:29 | melwitt | I thhink that would make sense | |
| 14:46:34 | melwitt | *think | |
| 14:46:42 | sean-k-mooney | so im not really asking for fully docs just capture the impaont parts form the gist you rote | |
| 14:47:30 | melwitt | ok, can do | |
| 14:47:33 | sean-k-mooney | i.e. if your automating the cofniguration with devstack add a small devstack local.conf samepel adn a littel prose to descibe how to set it up. | |
| 14:47:35 | sean-k-mooney | cool | |
| 14:47:50 | melwitt | thanks | |
| 14:47:58 | sean-k-mooney | by the way the reaon i said its not entirly a bug | |
| 14:48:08 | sean-k-mooney | is until this cycle | |
| 14:48:19 | sean-k-mooney | there really wast any devstack automation for any fo the drivers | |
| 14:48:54 | sean-k-mooney | im fine with filign wishlist bugs as tracker but we are missing devstack supprot for most of the rest of the drivrs as well | |
| 14:49:12 | sean-k-mooney | and i dont thnk fixing that when we have time shoudl need a lot of paper work | |
| 14:49:34 | melwitt | yeah I mean for this, it's the general device_profile extension so seems required for any cyborg use no? | |
| 14:49:54 | sean-k-mooney | only if using cyborg with neutron ports | |
| 14:50:03 | melwitt | yeah, it's good we have Wishlist | |
| 14:50:07 | melwitt | ah right. ok | |
| 14:50:23 | sean-k-mooney | that why i was debting if this shoudl only be enabled if you enable the intel driver | |
| 14:50:36 | sean-k-mooney | or sriov nic agent but its harmless to alwasy enabel | |
| 14:50:52 | sean-k-mooney | neutron will ignore it if your usign ovn af far as im aware | |
| 14:51:14 | melwitt | yeah. I think a guard would make sense. I guess intel is the only nic driver under nic/? I forgot already | |
| 14:51:26 | sean-k-mooney | currently yes | |
| 14:51:46 | sean-k-mooney | https://github.com/openstack/cyborg/blob/master/pyproject.toml#L52-L63 | |
| 14:51:49 | sean-k-mooney | at least in tree | |
| 14:52:05 | melwitt | ok cool. bc really it would be if any nic driver enabled but if we only have the one, then yeah | |
| 14:52:13 | sean-k-mooney | its technically a stevador entriy point i think but i woudl have to check | |
| 14:53:17 | chandankumar | we have one more bug to discuss, then we can talk more about that after meeting | |
| 14:53:37 | sean-k-mooney | cool | |
| 14:53:52 | chandankumar | #link | |
| 14:53:52 | chandankumar | [cyborg-tempest-plugin] Cyborg tempest tests fail in role manager does not exist https://bugs.launchpad.net/openstack-cyborg/+bug/2161909 | |
| 14:54:06 | chandankumar | amoralej: opened it. | |
| 14:54:13 | sean-k-mooney | melwitt: https://github.com/openstack/cyborg/blob/master/cyborg/agent/resource_tracker.py#L53-L57 | |
| 14:54:19 | chandankumar | It comes only when manager role does not exist in the environment | |
| 14:54:29 | sean-k-mooney | chandankumar: ya so i dug into this a bit | |