| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2025-09-09 | |||
| 16:28:39 | fwiesel | Uggla: So, back to you | |
| 16:28:40 | Uggla | fwiesel, thx, I just hope you pto were good. ;) | |
| 16:29:12 | Uggla | #topic Gibi's news about eventlet removal. | |
| 16:29:20 | Uggla | #link Blog: https://gibizer.github.io/categories/eventlet/ | |
| 16:29:23 | gibi | no news | |
| 16:29:37 | gibi | or rather no actionable new | |
| 16:29:39 | gibi | news | |
| 16:29:45 | gibi | I'm waiting for master to open for G | |
| 16:30:01 | Uggla | 👍 | |
| 16:30:09 | gibi | in the background I'm brewing a rally job that might be able to compare the performance of nova between eventlet and threading | |
| 16:30:09 | dansmith | gibi: I was trying to get my s-g patch set revised yesterday | |
| 16:30:17 | dansmith | the tooling was making that hard, but I'm hoping to be able to get that up soon in G | |
| 16:30:29 | gibi | dansmith: cool. I will review when it is up | |
| 16:31:02 | gibi | that is it from me | |
| 16:31:32 | Uggla | thx gibi | |
| 16:31:41 | Uggla | #topic Open discussion | |
| 16:32:09 | Uggla | I know bauzas wants to discuss about open infra topic. | |
| 16:32:16 | Uggla | bauzas please go ahead. | |
| 16:33:05 | bauzas | nothing important, this was a question whether we would want Nova presence at the OIF Summit pavilion https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/MM4OJO5CCC4GHMYUN3OJBDHTSZNB32H2/ | |
| 16:35:14 | gibi | I cannot be there | |
| 16:35:43 | opendevreview | Merged openstack/osc-placement stable/2025.2: [CI] Drop unsupported py39 based functional job https://review.opendev.org/c/openstack/osc-placement/+/959874 | |
| 16:36:37 | Uggla | bauzas, I don't manage to open the link is it something we could do ? | |
| 16:36:59 | stephenfin | Uggla: The OpenInfra Summit 2025 will host an Open Source Pavilion booth space where open source projects can book an hour or two to showcase their work, engage with attendees, and share updates on their latest developments at the Summit. If you are interested in taking advantage of this opportunity for your project please fill out this form [0] and the event planners will get back to you with scheduling details. | |
| 16:37:00 | bauzas | gibi: I'll be there, my question was whether you were wanting me to represent the Nova project | |
| 16:37:04 | stephenfin | That's the bulk of it | |
| 16:37:14 | stephenfin | (It took ages to load for me too) | |
| 16:37:16 | gibi | bauzas: ahh I see | |
| 16:37:32 | stephenfin | bauzas: What are you thinking? Something AI/GPU-flavoured? | |
| 16:37:39 | gibi | bauzas: showing the latest updates might make sense | |
| 16:37:41 | Uggla | thx stephenfin | |
| 16:37:45 | bauzas | previously we were providing project updates | |
| 16:37:54 | gibi | yeah someting like that^^ | |
| 16:37:56 | bauzas | looks like the pavilion is the new form | |
| 16:38:08 | stephenfin | Ah, nice one | |
| 16:38:23 | bauzas | cool, then that's settled | |
| 16:38:37 | bauzas | and Uggla, will you be at the summit ? | |
| 16:38:46 | Uggla | yes | |
| 16:38:47 | bauzas | fancy co-presenting ? | |
| 16:38:58 | gibi | bauzas: if you could also mention to the folks that we are removing eventlet and ask them to try threading mode and report any issues that would be awesomej | |
| 16:39:16 | dansmith | gibi: yeah, definitely | |
| 16:39:16 | bauzas | gibi: that will certainly be part of the project updates :) | |
| 16:39:24 | Uggla | gibi 👍 | |
| 16:39:35 | dansmith | definitely try to get some people on board for testing and providing feedback | |
| 16:39:43 | Uggla | bauzas, will be happy to copresent with you | |
| 16:40:11 | bauzas | cool, settled as I said | |
| 16:40:12 | gibi | ahh I see it will be in France hence we will have nice representation | |
| 16:40:14 | bauzas | moving on ? | |
| 16:41:04 | Uggla | anything else you want to discuss ? | |
| 16:41:20 | gibi | o/ | |
| 16:41:22 | gibi | pyroute2 | |
| 16:41:39 | Uggla | sure, sorry I was about to forget about it! Thx | |
| 16:42:08 | gibi | so the previous issue about async loops are reproduces with 0.9.4 as well | |
| 16:42:27 | gibi | so the supposed fix wasn't a real fix, it probably decreased the chance of happening | |
| 16:42:54 | gibi | I pinged the release channel asking if we can revert back to 0.8.x so late in the cycle. I got no answers yet | |
| 16:43:19 | tkajinam | you mean requirements channel ? | |
| 16:44:01 | gibi | I pinged the release channel I'm not aware of having a separate channel to requirements | |
| 16:44:14 | elodilles | #openstack-requirements o:) | |
| 16:44:20 | tkajinam | yeah | |
| 16:44:26 | elodilles | but i think i pinged the release folks | |
| 16:44:26 | sean-k-mooney | gibi: is there a workaround | |
| 16:44:31 | elodilles | on release channel | |
| 16:44:32 | sean-k-mooney | beyond downgrading | |
| 16:44:46 | gibi | sean-k-mooney: try again and it will succeed | |
| 16:44:56 | elodilles | (yeah, i did ping them) | |
| 16:45:02 | gibi | the issue happens pretty infrequently | |
| 16:45:34 | gibi | https://bugs.launchpad.net/os-vif/+bug/2109396 | |
| 16:45:37 | sean-k-mooney | i guess you could also use the legacy vsctl driver if we have not deleted that yet | |
| 16:45:40 | gibi | this is the bug it has the links | |
| 16:45:55 | sean-k-mooney | oh actully no | |
| 16:46:25 | gibi | I also don't have clear answer if we can revert the requirements | |
| 16:46:28 | gibi | later on stable | |
| 16:46:47 | tkajinam | though I've been struggling to get any feedback there. Maybe sending an email to ml would be a good option, hoping that Tony and Sean may be reading it. though I know release team has full +2 access to requirements, too. | |
| 16:47:03 | tkajinam | I mean I've been struggling to get any feedback in #openstack-requirements | |
| 16:47:25 | gibi | I will drop a mail to the ML | |
| 16:47:26 | sean-k-mooney | gibi: i guess the reals concern with the cap | |
| 16:47:32 | tkajinam | frickler has been the only one there for some time (though he is not a req team member but a release team member) | |
| 16:47:34 | sean-k-mooney | is has neutron started depending on it | |
| 16:47:39 | sean-k-mooney | and what are distos shiping | |
| 16:48:45 | sean-k-mooney | if we cant cap it later we mgith eb able to work around it with a retry in os-vif by cathcing the RuntimeError | |
| 16:48:52 | gibi | at least ykarel is aware of the issue from neutron as he discovered that the issue is happening with 0.9.4 | |
| 16:49:22 | sean-k-mooney | meaning they likely would not object to the cap? | |
| 16:49:36 | gibi | I don't want to solve it here. I will drop a mail to ML asking for guidance from requirements team | |
| 16:49:36 | sean-k-mooney | i.e. its impacting them also | |
| 16:49:42 | sean-k-mooney | ack | |
| 16:49:49 | gibi | I think they saw it in CI | |
| 16:49:57 | gibi | making interface attach failing | |
| 16:50:14 | gibi | but the failure is during the call from nova | |
| 16:50:22 | gibi | not a call form neutron to os-vif | |
| 16:50:34 | sean-k-mooney | ack well as i said we may be able to catch that excption and retry a limtied number of tiem in os-vif | |
| 16:50:43 | sean-k-mooney | kind of like we do for powering off cores | |
| 16:50:48 | gibi | sure. I rather want pyroute2 to fix the async code | |
| 16:50:49 | sean-k-mooney | btu i know neutron uses it more widely | |
| 16:51:09 | gibi | anyhow I have nothing further on this | |
| 16:52:17 | elodilles | (i added this to release team's weekly meeting's agenda for Friday, but i really hope that this will be sorted out until then :S) | |
| 16:53:02 | gibi | no worries I think we can go out with Flamingo even if the bug is there. Occasional failures that are fixed by the human retrying is not an end of the world | |
| 16:53:27 | gibi | elodilles: thanks!j | |
| 16:53:54 | elodilles | gibi: thanks too in advance for your mail to ML about this | |
| 16:55:03 | Uggla | good, anything else ? | |
| 16:55:43 | tkajinam | I was wondering if I could get any thoughts for https://bugs.launchpad.net/nova/+bug/2122296 . we don't have enough time and I likely skip the meeting next week but we can discuss it later here or the meeting in 2 weeks | |
| 16:56:20 | tkajinam | that's about sateless firmware but I also want to discuss similar problem with sev/sev-es, too | |