Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2026-10-05
13:47:41 opendevreview Matěj Mudra proposed openstack/nova master: Add --verbose to nova-manage commands https://review.opendev.org/c/openstack/nova/+/999956
13:58:44 opendevreview sean mooney proposed openstack/nova-specs master: Specify atomic PCI passthrough groups for Nova https://review.opendev.org/c/openstack/nova-specs/+/975635
14:01:10 opendevreview Lajos Katona proposed openstack/nova master: libvirt: Preserve source MTU during live migration https://review.opendev.org/c/openstack/nova/+/995809
14:01:21 sean-k-mooney gibi: no rush but i think ^ has adressed any remaining opens while maintianing the core or the v5 desgin intact, the addtion of `exclude` come directly form my testing with modern nvida gpus, they all seam to use the same pci vendor id and product id for PF and mig VFs making it a pain to list them otherwise.
14:25:41 bauzas dansmith: I don't know if you have either will or capacity for reviewing https://review.opendev.org/c/openstack/nova/+/1005727/ but it would be nice if you could do so for at least the first two patches in the series
14:26:11 dansmith bauzas: yes sorry I've had it open and keep getting pulled into other things
14:26:17 dansmith bauzas: keep nagging
14:26:23 bauzas no worries, we all feel the same
14:26:56 bauzas semi-related I got the chance to be free for 1.5h so I eventually was able to sit down and read your spec
14:32:16 dansmith ack, thanks
14:54:33 gibi folks I've closed https://bugs.launchpad.net/nova/+bug/2165149 as the last hit was 13 days ago. If you see functional failure with "qlalchemy.exc.OperationalError: (sqlite3.OperationalError) database is locked" then please reopen the bug and ping me
14:56:14 sean-k-mooney gibi: nice
14:56:48 dansmith bauzas: sorry I thought more had changed here but it looks like maybe unintentional change of a lot of nearby lines in the second patch that looks intentional/functional but isn't?
14:58:01 sean-k-mooney dansmith: we might eb able to add a signle line to agetns.md to prevent that
14:58:18 sean-k-mooney the issue with the wrapping of the comments is because pep8 requires we use a shortor 72 column limit
14:58:20 dansmith sean-k-mooney: yeah I think we should.. I have an instruction in my local one
14:58:23 sean-k-mooney but nova does nto follow that convention
14:58:58 dansmith sean-k-mooney: well, there are other such things which are purely style preference that get changed
14:59:26 dansmith sean-k-mooney: I'll propose an agents change and tag you
14:59:56 sean-k-mooney acki didnt really look closely at the other changes but ya it the second paragrah im 90% for the comments at least https://peps.python.org/pep-0008/#maximum-line-length
15:00:13 sean-k-mooney ```For flowing long blocks of text with fewer structural restrictions (docstrings or comments), the line length should be limited to 72 characters.```
15:04:02 opendevreview Dan Smith proposed openstack/nova master: Add agent instructions about gardening https://review.opendev.org/c/openstack/nova/+/1008817
15:04:07 dansmith sean-k-mooney: ^
15:04:19 sean-k-mooney :) gardening
15:04:49 dansmith a human referred to this using that term years ago, so it's always what I call it now
15:05:17 sean-k-mooney ack the phasing in the guardrails looks good to me
15:07:54 sean-k-mooney i have started experimenting with https://github.com/DietrichGebert/ponytail/blob/main/skills/ponytail/SKILL.md and https://github.com/cursor/plugins/blob/main/pstack/skills/unslop/SKILL.md lately in my own use
15:10:05 sean-k-mooney i have not used either enouch to say it definitivly helps but anicdtally they have. im reseving judgement unitll i atempt larger work with them
15:11:49 opendevreview Balazs Gibizer proposed openstack/nova master: Run periodics outside compute test callbacks https://review.opendev.org/c/openstack/nova/+/1007150
15:13:27 sean-k-mooney gibi: hum that intereesting i didnt think about the fact that the preodic woudl be in a diffent treadh/db context in that case
15:13:47 dansmith oh I like the unslop one
15:14:18 dansmith normally I just stop what I'm doing to verbally berate the agent for being so agentic in is verbiage.. makes me feel better.
15:14:22 sean-k-mooney that one i have found useful even if you jsut copy paste part fo it in your personal agent md
15:16:39 gibi sean-k-mooney: we basically sometimes triggering the periodic job of compute B in the computeA RPC handeler
15:16:51 sean-k-mooney i have only been using those for about 2-3 weeks so again i can say it impoved but it feels better.
15:17:15 sean-k-mooney gibi: oh ok ya that woudl be incorrect
15:17:36 sean-k-mooney escially if a and b were in differnet cells ectra
15:18:27 gibi yeah these tests were complicated :)
15:20:03 gibi gmaan: btw I finally reviewed https://review.opendev.org/c/openstack/nova/+/1006351 I have a bunch of small things that together might make sense to fix
15:20:24 sean-k-mooney general oslo.servce questsion https://docs.openstack.org/nova/latest/configuration/config.html#DEFAULT.run_external_periodic_tasks, i dont think nova really supprot changing that form its defautl as to my knowlage we dont have any external perodic like that
15:20:41 sean-k-mooney do we really need to expsoe thsoe in our config?
15:21:11 sean-k-mooney i wonder if we shoudl propose taging that option as advanced
15:21:44 gibi yeah at least it is no-op as we don't have such tasks
15:22:37 sean-k-mooney there is also https://docs.openstack.org/nova/latest/configuration/config.html#DEFAULT.periodic_enable nova might work with them disabeld but we certenly dont test that
15:23:08 sean-k-mooney just somethign i have alwasy found odd in our config ref.
15:23:54 gibi yeah would be nice to hide them somehow
15:24:05 sean-k-mooney if we made them advanced https://docs.openstack.org/oslo.config/ocata/cfg.html#advanced-option that woudl at least make it more clear that you proably shoudl not mess it it
15:24:54 sean-k-mooney advanced does nto actully hide them but it puts them at the end of the section with a header waring agaisnt using them
15:25:50 sean-k-mooney i might just write a patch for that if i can confirm where those are defeiend
15:25:59 sean-k-mooney i think its oslo.service but i could be wrong
15:31:35 tkajinam o/ would be great if we can throw a few logic cleanups ( https://review.opendev.org/c/openstack/nova/+/997818 and https://review.opendev.org/c/openstack/nova/+/995160 ) before merge conflict hits these
15:32:31 gibi tkajinam: looking...
15:32:37 tkajinam gibi, thx !
15:32:46 Uggla Reminder: upstream meetin in ~30mn
15:38:07 sean-k-mooney tkajinam: by the way we really really shoudl raise the min version in 2027.1
15:38:17 bauzas dansmith: wait, what ? on my patch series ? doh, lemme raise my fist to Cursor/OpenCode
15:38:33 sean-k-mooney the next version are the version ubuntu 24.04 shiped with and this is the last release wehre that os is even supproted
15:38:57 tkajinam sean-k-mooney, yeah that's probably the good topic to be discussed at PTG.
15:39:28 tkajinam we don't really "discuss" it but may just get quick agreement (hopefully)
15:39:30 sean-k-mooney im not sure what the new "NEXT" version shoudl be but i suspsect if we took 26.04 as a baseline it woudl be fine
15:39:38 tkajinam yeah
15:40:14 sean-k-mooney i know centos 10 ships a newer version and i suspect trixie does as well but that really all we need to check
15:40:26 dansmith bauzas: yes, quite a bit. I noted them
15:40:31 bauzas just saw
15:40:48 bauzas usually I'm able to fix it directly
15:40:59 bauzas but this time I don't know why I missed them
15:41:05 bauzas I probably rushed, my bad
15:41:21 bauzas I'll upload a new rev
15:41:39 dansmith thanks
15:42:30 bauzas I'm rewrapping all my AI efforts with OpenCode, probably worth adding a new rule indeed
15:42:53 dansmith see my proposal above :)
15:42:55 bauzas Cursor is nice but too much autonomous
15:43:34 bauzas dansmith: I already have a bunch of instructions fwiw, and I'm pretty sure I said about it
15:43:58 bauzas not sure why it didn't followed the nova convention
15:44:11 dansmith my local instruction (which is basically what I proposed) seems to be working for me
15:44:34 dansmith before it would rewrap comments and change style of completely unrelated lines in the general vicinity of work I was doing
15:44:34 sean-k-mooney the nova convetion explictly is in conflict with pep8s convetnion in this specific case and we dont docuemnt that anywhere
15:44:41 sean-k-mooney but i think the guardrail dan proposed
15:44:53 sean-k-mooney is generic enouch to prevent this and other similer cases
15:45:09 dansmith sean-k-mooney: I don't think the pep8 thing has anything to do with one of his changes which was turning a string into an f-string but changing nothing else about it
15:45:23 dansmith ...in a line of code totally unrelated to the functional change he was maknig
15:45:47 sean-k-mooney ack i didnt look at those patches closely
15:46:05 sean-k-mooney so i dind see that trasnformation
15:48:32 tkajinam I wonder if we can discuss libvirt version bump in release planning or do we want a separate topic ?
15:49:11 opendevreview Sylvain Bauza proposed openstack/nova master: Prevent concurrent attach_volume with task_state guard https://review.opendev.org/c/openstack/nova/+/1006786
15:49:12 opendevreview Sylvain Bauza proposed openstack/nova master: Reject stale reserve_block_device_name RPC on compute https://review.opendev.org/c/openstack/nova/+/1006011
15:54:47 sean-k-mooney i think that prorbly fine it shoudl be quick
15:55:10 sean-k-mooney tkajinam: i woudl jsut push a patch to do the bump and add the link (not the cleanup just to change the constants)
15:55:32 tkajinam ok
15:56:23 tkajinam I've added the topic at the bottom just in case we want it separate but I'll ask Uggla if we can merge it into the release planning topic.
15:57:42 Uggla tkajinam, I have not followed the discussion, but feel free to add to the correct place.
15:58:00 tkajinam we can quickly talk about that after the meeting
15:58:54 Uggla tkajinam, up to you but there is no issue to add it to the agenda, if you wish.
16:00:17 Uggla #startmeeting nova
16:00:17 opendevmeet Meeting started Mon Oct 5 16:00:17 2026 UTC and is due to finish in 60 minutes. The chair is Uggla. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:17 opendevmeet Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:17 opendevmeet The meeting name has been set to 'nova'
16:00:22 fwiesel o/
16:00:26 Uggla Hello everyone
16:00:28 tkajinam o/

Earlier   Later