| *** elodilles_pto is now known as elodilles | 07:34 | |
| *** sfinucan is now known as stephenfin | 10:58 | |
| *** tobias-urdin9 is now known as tobias-urdin | 13:03 | |
| noonedeadpunk | hi there! we have just branched OSA roles, but for some repos we do not see branches created in Zuul | 13:15 |
|---|---|---|
| noonedeadpunk | ie: https://zuul.opendev.org/t/openstack/project/opendev.org/openstack/openstack-ansible-os_horizon missing 2023.2, thus jobs are not even started in Zuul | 13:15 |
| fungi | noonedeadpunk: i think we've seen zuul miss ref-updated events when multiple branch creations occur in a very short span of time (not sure if gerrit fails to emit them or zuul fails to process them). were there multiple repos branched somewhat simultaneously? i think we've addressed it in the past with an online scheduler reconfiguration | 13:56 |
| fungi | i'm not really around today (or the first part of tomorrow), but hopefully someone else can help soon | 13:56 |
| noonedeadpunk | yeah, quite some branches were created at the same time indeed | 14:01 |
| noonedeadpunk | and around 6 of them were left without update on zuul side | 14:01 |
| tonyb | fungi: I know you aren't really around. I can do the reconfigure, I'm gussing it's seomthing like sudo docker exec -it --user zuul zuul-scheduler_scheduler_1 zuul-scheduler tenant-reconfigure openstack on zuul01 | 15:38 |
| clarkb | tonyb: there was discussion of this the last time it came up in #opendev. We should be able to dig up what the command was from there | 15:55 |
| clarkb | would've been around the recent openstack release. but ya your command is close. The thing I can't remmber is if we need to give flags for extra reconfiguration or not | 15:56 |
| tonyb | clarkb: Okay. I'll look | 15:58 |
| clarkb | tonyb: 2023-10-16T22:13:00+0000 docker exec zuul-scheduler_scheduler_1 zuul-scheduler tenant-reconfigure openstack that is in zuul02s root history | 15:59 |
| clarkb | I suspect that if we look at the #opendev log on that day it will show this is the command we settled on and that also matches your command | 16:00 |
| tonyb | clarkb: LOL | 16:00 |
| tonyb | [2023-10-16 17:13:12] <clarkb> `docker exec zuul-scheduler_scheduler_1 zuul-scheduler tenant-reconfigure openstack` is what I ran on zuul02 | 16:01 |
| clarkb | there is also discussion around that time about why this happens in zuul | 16:02 |
| clarkb | noonedeadpunk: ^ if you are curious about details | 16:02 |
| noonedeadpunk | it's happening in zuul matrix? | 16:08 |
| tonyb | noonedeadpunk: No, that was a discussion from Oct/16 | 16:09 |
| noonedeadpunk | ah, ok, now I realized :) | 16:09 |
| tonyb | clarkb: I the same command from the history on zuul02 | 16:10 |
| tonyb | clarkb: I guess the reconfig takes a little while | 16:10 |
| noonedeadpunk | well, I kinda have no means to control how actively branching will happen as it's applied on patch merge | 16:11 |
| clarkb | tonyb: yes it takes $minutes | 16:12 |
| clarkb | the suggestion we made to the release team was to batch them iirc | 16:12 |
| tonyb | noonedeadpunk: got it. It's something for us to look into. It's an annoyance for imapacted users but at this point not something you need to worry about | 16:12 |
| tonyb | clarkb: Yeah. We could do that. | 16:13 |
| noonedeadpunk | yeah, that would be fair thing to do I guess. | 16:13 |
| noonedeadpunk | But not sure that applying semaphore would be enough, as in our case we have just huge list of repos that are getting branched in one go | 16:14 |
| noonedeadpunk | so it's indeed smth to look in tooling itself | 16:15 |
| tonyb | Yes. | 16:15 |
| tonyb | the release tooling and/or adding debugging to zuul to dig into this | 16:16 |
| clarkb | from memory (haven't looked at the logs yet) the issue is when we have a bunch of reconfigurations that happen during a reconfiguration | 16:17 |
| clarkb | by batching them we can try and avoid the number of sequential reconfigurations | 16:17 |
| noonedeadpunk | ok, branches have appeared | 16:37 |
| noonedeadpunk | thanks for the help! | 16:43 |
| tonyb | np | 16:43 |
| *** pmatulis_ is now known as pmatulis | 17:04 | |
| pmatulis | can anyone nail this doc PR for the OpenStack Charms project? https://review.opendev.org/c/openstack/openstack-manuals/+/902695 | 17:04 |
| clarkb | pmatulis: I think you need someone from this group https://review.opendev.org/admin/groups/238e9f33453744139facac78a6eb17de63c46a00,members | 17:06 |
| tonyb | pmatulis: you can try poking on #openstack-doc | 17:08 |
| pmatulis | my understanding was that this channel took over from #openstack-doc , which is now shuttered | 17:14 |
| tonyb | pmatulis: oh that could be. I apologise if I sent you in the wrong direction | 17:16 |
| frickler | pmatulis: tonyb: actually it is the tc taking over reviews for that mostly nowadays, so asking in #openstack-tc is even better. I also have had that patch open for a day already, but didn't get round to actually verifying that the docs links are working | 17:20 |
| *** tobias-urdin34 is now known as tobias-urdin | 17:27 | |
| tonyb | frickler: Ahh okay. I wonder if it's worth updating the topic and/or setting a channel redirect | 17:32 |
| JayF | going to be honest, I didn't even realize TC were review owners on that repo | 17:34 |
| JayF | good news, review incoming | 17:34 |
| JayF | also jamespage is on TC And can review changes, and he's charm project adjacent | 17:34 |
| tonyb | JayF: neither did I TIL | 17:42 |
| frickler | JayF: tonyb: this explains it a bit, not sure if there was some official tc decision around it https://review.opendev.org/c/openstack/project-config/+/880235 | 17:50 |
| JayF | I mean, contextually, there's no docs team so it makes sense | 17:50 |
| frickler | regarding branching, if someone creates like 60 branches in a single review, there's not much the release team can do about that, either https://review.opendev.org/c/openstack/releases/+/901959/4/deliverables/bobcat/openstack-ansible-roles.yaml | 18:01 |
| frickler | although that was actually elodilles proposing that patch ... hmm | 18:05 |
| tonyb | yeah it's automated so we could update the tools to do no more than $n per review as a workaround | 18:07 |
| pmatulis | oh #openstack-tc ok | 18:17 |
| gmann | JayF: frickler: doc SIG is officially moved under TC since there was no maintainer left in doc team. all the doc repo are in TC owned repo since then https://github.com/openstack/governance/blob/master/reference/technical-committee-repos.yaml | 18:38 |
| gmann | that is why i made gerritbot change so that more TC start looking into those repo review. i keep doing for incoming one but hope more TC can start taking look | 18:39 |
| gmann | pmatulis: +A | 18:45 |
| elodilles | frickler: yes, unfortunately openstack-ansible-roles have that many repos in its collection. if this is the same issue as with the high number of eol'd branch deletion, then the solution is (as clark wrote) to batch them somehow in the release tooling | 19:00 |
| elodilles | it's good to know that this can occur in case of branching as well, not just branch deletion :S | 19:01 |
| clarkb | elodilles: yes its the same code paths in zuul iirc | 19:02 |
| elodilles | clarkb: ack | 19:02 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!