Wednesday, 2026-07-22

opendevreviewMerged openstack/kolla master: rabbitmq: Update to 4.3  https://review.opendev.org/c/openstack/kolla/+/98842506:04
opendevreviewMichal Nasiadka proposed openstack/kolla master: CI: copr - stringify part 2  https://review.opendev.org/c/openstack/kolla/+/99826606:22
opendevreviewMichal Nasiadka proposed openstack/kolla master: CI: copr - replace secrets without newlines  https://review.opendev.org/c/openstack/kolla/+/99826606:47
opendevreviewMichal Nasiadka proposed openstack/kolla master: CI: copr - replace secrets without newlines  https://review.opendev.org/c/openstack/kolla/+/99826606:52
opendevreviewMerged openstack/kolla master: CI: copr - replace secrets without newlines  https://review.opendev.org/c/openstack/kolla/+/99826607:24
opendevreviewPierre Riteau proposed openstack/kayobe stable/2026.1: Add missing reconfigure.yml for manage-containers  https://review.opendev.org/c/openstack/kayobe/+/99827107:34
opendevreviewPierre Riteau proposed openstack/kayobe stable/2026.1: CI: Install kernel-modules-extra  https://review.opendev.org/c/openstack/kayobe/+/99827307:41
opendevreviewPierre Riteau proposed openstack/kayobe stable/2025.2: CI: Install kernel-modules-extra  https://review.opendev.org/c/openstack/kayobe/+/99827407:42
opendevreviewPierre Riteau proposed openstack/kayobe stable/2025.1: CI: Install kernel-modules-extra  https://review.opendev.org/c/openstack/kayobe/+/99827507:42
mnasiadkabbezak, blanson[m], frickler: I noticed a lot of our CI jobs are having RMQ message timeouts in logs, I think the qos_prefetch of 1 is not really helping anybody - posted https://review.opendev.org/c/openstack/kolla-ansible/+/99821908:15
opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: rabbitmq: Bump default rabbit_qos_prefetch_count from 1 to 10  https://review.opendev.org/c/openstack/kolla-ansible/+/99821908:16
blanson[m]interesting 08:17
blanson[m]we're not using quorum queues cause we use ancient (actual fossiles) versions of openstack and rmq, but the docs seems clear enough that this would help08:18
opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: rabbitmq: Bump default rabbit_qos_prefetch_count from 1 to 10  https://review.opendev.org/c/openstack/kolla-ansible/+/99821908:22
opendevreviewRafal Lewandowski proposed openstack/kolla-ansible master: Add container state check after restart  https://review.opendev.org/c/openstack/kolla-ansible/+/96905308:24
opendevreviewRafal Lewandowski proposed openstack/kolla-ansible master: Add container state check after restart  https://review.opendev.org/c/openstack/kolla-ansible/+/96905308:25
opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: rabbitmq: Bump default rabbit_qos_prefetch_count from 1 to 10  https://review.opendev.org/c/openstack/kolla-ansible/+/99821908:26
*** jhorstmann is now known as Guest1392408:47
opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: rabbitmq: Bump default rabbit_qos_prefetch_count from 1 to 50  https://review.opendev.org/c/openstack/kolla-ansible/+/99821908:49
mnasiadkablanson[m]: actually I think a good real-world production env value should be 50, we can lower that in CI if we see issues (but I doubt)08:49
mnasiadkaRMQ recommends it between 50 and 20008:49
opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: Bump ansible-core to 2.20/2.21  https://review.opendev.org/c/openstack/kolla-ansible/+/99674509:05
opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: Bump ansible-core to 2.20/2.21  https://review.opendev.org/c/openstack/kolla-ansible/+/99674509:07
opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: Bump ansible-core to 2.20/2.21  https://review.opendev.org/c/openstack/kolla-ansible/+/99674509:08
opendevreviewMerged openstack/kayobe master: Remove leftover device mapper references from the code  https://review.opendev.org/c/openstack/kayobe/+/99771409:10
opendevreviewMerged openstack/kolla master: CI: run copr periodic job only on master  https://review.opendev.org/c/openstack/kolla/+/99798509:11
opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: Bump ansible-core to 2.20/2.21  https://review.opendev.org/c/openstack/kolla-ansible/+/99674509:12
opendevreviewPierre Riteau proposed openstack/kayobe stable/2026.1: Add missing reconfigure.yml for manage-containers  https://review.opendev.org/c/openstack/kayobe/+/99827109:50
opendevreviewPierre Riteau proposed openstack/kayobe stable/2026.1: Remove leftover device mapper references from the code  https://review.opendev.org/c/openstack/kayobe/+/99828709:52
opendevreviewPierre Riteau proposed openstack/kayobe master: [WIP] Bump ansible version  https://review.opendev.org/c/openstack/kayobe/+/99502109:55
opendevreviewBartosz Bezak proposed openstack/kolla stable/2026.1: CI: run copr periodic job only on master  https://review.opendev.org/c/openstack/kolla/+/99828809:58
opendevreviewBartosz Bezak proposed openstack/kolla stable/2025.2: CI: run copr periodic job only on master  https://review.opendev.org/c/openstack/kolla/+/99828909:58
opendevreviewPierre Riteau proposed openstack/kayobe master: [WIP] Bump ansible version  https://review.opendev.org/c/openstack/kayobe/+/99502110:13
opendevreviewPierre Riteau proposed openstack/kayobe stable/2026.1: Remove leftover device mapper references from the code  https://review.opendev.org/c/openstack/kayobe/+/99828710:14
opendevreviewRafal Lewandowski proposed openstack/kolla-ansible master: Add container state check after restart  https://review.opendev.org/c/openstack/kolla-ansible/+/96905310:47
opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: rabbitmq: Bump default rabbit_qos_prefetch_count from 1 to 50  https://review.opendev.org/c/openstack/kolla-ansible/+/99821911:17
opendevreviewMerged openstack/kayobe stable/2026.1: CI: Install kernel-modules-extra  https://review.opendev.org/c/openstack/kayobe/+/99827312:04
opendevreviewVerification of a change to openstack/kayobe stable/2025.2 failed: CI: Install kernel-modules-extra  https://review.opendev.org/c/openstack/kayobe/+/99827412:04
opendevreviewKari Häkkinen proposed openstack/kolla-ansible master: cinder: Define backend variables in group_vars/all  https://review.opendev.org/c/openstack/kolla-ansible/+/99829612:31
opendevreviewGrzegorz Koper proposed openstack/kolla-ansible master: neutron: Split shared network agents into role  https://review.opendev.org/c/openstack/kolla-ansible/+/97561412:50
opendevreviewGrzegorz Koper proposed openstack/kolla-ansible master: neutron: Split shared compute agents into role  https://review.opendev.org/c/openstack/kolla-ansible/+/99760012:50
opendevreviewGrzegorz Koper proposed openstack/kolla-ansible master: neutron: Split ML2/OVS agents into role  https://review.opendev.org/c/openstack/kolla-ansible/+/99760112:50
opendevreviewGrzegorz Koper proposed openstack/kolla-ansible master: neutron: Split ML2/OVN agents into role  https://review.opendev.org/c/openstack/kolla-ansible/+/99760212:50
opendevreviewKari Häkkinen proposed openstack/kolla-ansible master: cinder: Define backend variables in group_vars/all  https://review.opendev.org/c/openstack/kolla-ansible/+/99829612:54
mnasiadka bbezak frickler kevko mmalchuk gkoper jovial mattcrees dougszu darmach pabloclsn ravlew salmankh amir58118 r-krcek blanson[m] fprzewozn - meeting in 1 minute12:59
* fungi is around again for his topic too13:00
Guest13507o/13:00
mnasiadka#startmeeting kolla13:00
opendevmeetMeeting started Wed Jul 22 13:00:58 2026 UTC and is due to finish in 60 minutes.  The chair is mnasiadka. Information about MeetBot at http://wiki.debian.org/MeetBot.13:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.13:00
opendevmeetThe meeting name has been set to 'kolla'13:00
mnasiadka#topic rollcall13:01
mnasiadkaGuest13507: I assume you’d like to change your nick ;)13:01
mnasiadkao/13:01
frickler\o13:01
bbezak\o13:01
blanson[m]hello !13:01
fungiahoy!13:01
owenjoneso/13:02
AlmaMC[m]Hi13:02
mnasiadka#topic agenda13:03
mnasiadka* Current cycle planning13:03
mnasiadka* Additional agenda (from whiteboard)13:03
mnasiadka* Open discussion13:03
mnasiadka#topic Current cycle planning13:03
mnasiadkaAnybody wants to discuss any patches for new features?13:04
*** Guest13507 is now known as ravlew13:04
ravlew\o/13:05
mnasiadkaSeems not13:05
mnasiadkaSo let’s go to additional agenda13:05
mnasiadka#topic Additional agenda (from whiteboard)13:06
mnasiadkafungi: you first13:06
mnasiadkaBridging the Gap Gazpacho Cycle Follow-Up Discussion13:06
fungibriefly revisiting this topic from last week now that folks have hopefully had time to digest it better, i wanted to see if anyone had additional thoughts13:06
fungiwe had takeaway observations noted... contributor survey responses reflected a more negative perception of review activity than are borne out by metrics collected from gerrit13:06
fungithe median time to merge for patches was twice as fast in gazpacho, it could help to understand if there was a shift in review activity that drove this like focusing on smaller fixes or clearer prioritization13:07
blanson[m]median time for kolla projects only or across the board ? I assume kolla ? 13:07
fungier, i had that backwards13:07
opendevreviewSeunghun Lee proposed openstack/kolla-ansible stable/2026.1: [stable-only] Split CA of ProxySQL internal TLS and backend TLS  https://review.opendev.org/c/openstack/kolla-ansible/+/99425813:07
fungiblanson[m]: for kolla13:07
fungii was looking at the wrong tab just now, the *median* time doubled (half as fast) but the mean/average was twice as fast13:08
fungioh, and that was time to review not time to merge13:08
fungiwe saw there was reasonable alignment in priority tracking methods in the survey responses13:09
frickler"Gazpacho Cycle" is the general openstack timeframe, not the shifted kolla one, right?13:09
AlmaMC[m]Is there some metrics on the time to merge ?13:09
fungifrickler: yes, rc1 thursday in flamingo to rc1 thursday in gazpacho, specifically13:10
fungifor the coordinated release13:10
fungibut the statistics were focused on the kolla deliverable repositories13:10
fungiAlmaMC[m]: no that's where i was confused, i forgot we didn't manage to get the mean and median times to merge this round because of the switch from bitergia to lfx insights, we're still working with the insights developers to try to surface those numbers for the next round13:11
fungiwe did get review response times, just not merge times yet13:11
opendevreviewMerged openstack/kayobe stable/2026.1: Add missing reconfigure.yml for manage-containers  https://review.opendev.org/c/openstack/kayobe/+/99827113:11
mnasiadkafungi: would be useful to get the metrics on cycle-trailing cycle, which is a bit different, and actually majority of the code lands between coordinated release and the cycle trailing release (or at least I think it lands then)13:12
fungisomething else slightly anomalous is that the median number of patchsets per review increased from 1 to 2, and the average climbed from 4.5 to 613:12
fungimnasiadka: that's an excellent point, and something i hadn't considered until frickler asked just a minute ago13:12
fungiit's an interesting challenge since the cycle-trailing projects aren't aligned to a common timeframe13:13
fungithere's a deadline, but they don't all release on the deadline they could "end" their work on the cycle any time between the coordinated release and the trailing deadline13:13
fungialso since we're trying to gauge cycle end by when the project's deliverables create their stable branches, we'd need to know somewhat in advance what date that would be for a cycle-trailing project's deliverables13:14
mnasiadkaKolla usually fits within the deadline, but I agree that you would need to measure merge dates in openstack/releases13:15
mnasiadkaWhich is probably non-trivial13:15
opendevreviewSeunghun Lee proposed openstack/kolla-ansible stable/2026.1: [stable-only] Split CA of ProxySQL internal TLS and backend TLS  https://review.opendev.org/c/openstack/kolla-ansible/+/99425813:15
fungiit may be doable though, i'll confer with ildiko and clark13:15
fungibut also that would delay our release of the openstack-wide statistics since they're built on the aggregate of all teams' deliverables too13:16
fungianyway, it's an excellent callout and i'll see what our options are around that13:16
mnasiadkaThat’s sort of two sets of statistics released on different dates13:16
mnasiadkaThanks fungi13:16
mnasiadkaAnybody has any questions to fungi around Bridging the Gap?13:17
fungii think the main observation we had is the declining maintainer count, you had 12 active maintainers up through epoxy, it fell to 9 in flamingo and hasn't recovered13:18
fungii know every team wants to find new maintainers, but obviously identifying volunteers and having time to mentor them aren't trivial13:19
mnasiadkaOh yes, the maintainer count is falling13:19
fungistill might be worthwhile to focus on as a theme for the next cycles13:19
fungiespecially with the workload itself showing no signs of slowing13:20
fungi(change volume is steadily increasing over the past several cycles)13:20
AlmaMC[m]How is considered a maintainer active from the stats ?13:20
fungithe number of gerrit accounts leaving code-review -2/+2 or workflow +1 votes in that timeframe13:21
fungion any of the team's deliverable repositories13:21
fungiobviously for teams that have split/subdivided responsibilities for core reviewer subteams that can be misleading13:21
mnasiadkaOh yes, Kayobe is a different thing amongst all of our repos13:22
fungilike different sets of core reviewers on separate repositories, you're being counted together as maintainers for the team overall13:22
fungiseems like there may be no other questions/suggestions for now? i'll yield the floor, but you know where to find me13:24
blanson[m]thanks fungi13:25
mnasiadkaNext one blanson[m]13:26
mnasiadkablanson (2026/07/13)13:27
mnasiadkaincreasing neutron rpc workers fixed most of the port creation timeouts we had in CI (if not all). How do we scale that properly for users ?13:27
mnasiadkaWe have encountered this ourselves in production with ovs, once there is sufficient pressure on neutron, workers start "missing" commands and some ports never get created because the creation message is never put into rabbitmq.13:27
mnasiadkaOfficial neutron recommendations are 1 rpc worker per 2 cpu cores (https://docs.openstack.org/neutron/latest/configuration/samples/neutron.html , search "rpc_workers")13:27
mnasiadkaInitial idea was: add floor and ceiling values (like 6 and 32, idk), scale to 1 per 2 cores between these values, cap the bottom portion so that small controllers might still have a decent amount, cap the ceiling so that we don't scale this to infinity in case people have absurdly large controllers13:27
mnasiadkause custom values in CI because it doesn't need to be long-term stable ? biggest problems wqith too many workers is memory starvation long term. 13:27
mnasiadkadocumenting this is probbly a good idea aswell.13:27
blanson[m]I was supposed to work on it this week but then our DC went awol so13:28
mnasiadkaOk then13:28
blanson[m]I will send something this week, I'll have to remind myself of what we agreed on, but it didn't seem too complicated to do 13:28
mnasiadkaI think the rabbit_qos_prefetch thing is sort of related13:29
mnasiadka#link https://review.opendev.org/c/openstack/kolla-ansible/+/99821913:29
mnasiadkaBut we should improve our api workers stance13:29
mnasiadkablanson[m]: I just realised neutron has a good calculation manner if you don’t supply rpc and api workers count13:30
mnasiadkaMaybe other services have that as well?13:30
blanson[m]werll the thing is that with uwsgi move, api workers aren't managed by nova anymore 13:31
blanson[m]nor neutron13:31
blanson[m]nor anybody really 13:31
blanson[m]and I'm not sure the wsgi conf has as nice of a default machenism as neutron for rpc_workers13:31
mnasiadkatrue13:32
mnasiadkaWhat about RPC?13:32
blanson[m]rpc should be fine 13:32
blanson[m]idk if nova has the same "smart" defaults 13:32
blanson[m]but I can look it up 13:32
mnasiadkaWe would need to analyse it, if not we could maybe do some smarter default than we currently have13:32
mnasiadkaAnd encourage people in the docs to fine tune them by themselves13:33
mnasiadkaAnyway, just thoughts13:33
mnasiadkaHopefully you can have more insight next week13:33
blanson[m]nova conductor default to cpu count, scheduler worker just says "read the docs" so idk :D 13:34
mnasiadkaThat’s fantastically universal13:34
mnasiadkaOk, let’s move on13:34
mnasiadka#topic Open discussion13:34
mnasiadkaAnybody anything?13:34
opendevreviewMichal Nasiadka proposed openstack/ansible-collection-kolla master: docker: Use ansible_facts for rpm subdir  https://review.opendev.org/c/openstack/ansible-collection-kolla/+/99830213:37
blanson[m]lots of RP+1 review are almost over the line 13:39
mnasiadkaAs in they need a second reviewer?13:40
mnasiadkaOr what is the line?13:40
blanson[m]maybe we push a few of them over this week huh ? :D 13:40
blanson[m]yh I think there are like 5-8 reviews awaiting a second review 13:40
blanson[m]I will try to get some done this queue keeps growing it scares me 13:40
mnasiadkaOk, I can have a look13:42
mnasiadkaIf bbezak or frickler can do the same it would be nice13:42
blanson[m]9/30 have a +2 already. 5/30 don't pass the gate and 7/30 have not been reviewed or have been vote-reset 13:43
blanson[m]I'll get some work done this week hopefully I have time 13:44
mnasiadkaIf something is broken beyond comprehension and there are no updates from author - we should remove the RP vote (there should be a bin icon available)13:44
mnasiadkaOk then, let’s finish for today13:46
mnasiadkaThank you all for coming!13:46
mnasiadka#endmeeting13:46
opendevmeetMeeting ended Wed Jul 22 13:46:36 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)13:46
opendevmeetMinutes:        https://meetings.opendev.org/meetings/kolla/2026/kolla.2026-07-22-13.00.html13:46
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/kolla/2026/kolla.2026-07-22-13.00.txt13:46
opendevmeetLog:            https://meetings.opendev.org/meetings/kolla/2026/kolla.2026-07-22-13.00.log.html13:46
blanson[m]thank you13:48
opendevreviewMichal Nasiadka proposed openstack/ansible-collection-kolla master: docker: Store yum gpg file in the role  https://review.opendev.org/c/openstack/ansible-collection-kolla/+/99830613:53

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