Wednesday, 2026-02-04

@clarkb:matrix.orgGitea09 backups failed again but that happend about 20 minutes before ^ merged so I think it is unrelated00:02
@bbezak:matrix.orgHi. not sure if that was reported, but git operations on opendev is terribly slow08:52
@bbezak:matrix.orgLLMs hammering again?08:54
@fungicide:matrix.orgi'm going to grab an early lunch out now that the roads are mostly clear again, back shortly (i hope!)15:46
@clarkb:matrix.orginfra-root the keycloak backup failure does appear to be related to the mysql backup streaming script failing. I have to pop out in a few for a doctors appointment so not sure I'll get to debugging that for a bit16:00
@clarkb:matrix.orgwe can in theory manually run the /etc/borg-streams/mysql script and see if it fails in a consistent manner that we can debug though16:01
@clarkb:matrix.orgsystem load seems reasonable across the giteas so any slowness right now is probably not bots crawling the site16:03
@clarkb:matrix.orglooking at etherpad, gitea, gerrit, and mailman their db backups seem to have succeeded to the vexxhost backup server. Backups to the rax server should start in an hour or two. Its possible that this was just a fluke or maybe it is something specific to how keycloak runs its database server?16:15
@clarkb:matrix.orgnote I only looked at the backup logs not the backups themselves on each of those hosts16:16
@fungicide:matrix.orgi'll take a look at the keycloak situation in a few minutes17:12
@fungicide:matrix.orgbut yeah, that's the only new failure i see17:13
@fungicide:matrix.orgmariadb is listening on localhost:3306 at least17:14
-@gerrit:opendev.org- Monty Taylor https://matrix.to/#/@mordred:inaugust.com proposed: [openstack/project-config] 975673: Add wandertracks-android repo for WanderTracks https://review.opendev.org/c/openstack/project-config/+/97567317:24
@mordred:waterwanders.comI grew an extra repo17:25
@clarkb:matrix.orgmordred: and I can't wait to learn what you find about android testing in zuul :)17:25
@mordred:waterwanders.comClark: me neither! :)17:27
@mordred:waterwanders.comlocally tests run against an emulator ... but that is kvm backed, and of course we all love our friend nested virt17:27
@mordred:waterwanders.comso maybe they'll just run against an emulator REALLY SLOWLY17:28
@fungicide:matrix.org`mysqldump: Got error: 2002: "Can't connect to server on '127.0.0.1' (115)" when trying to connect`17:30
@mordred:waterwanders.comport vs socket / default mysql db permissions issue?17:31
@fungicide:matrix.orgi think the container networking must have split family binding17:33
@fungicide:matrix.orgwe configure mariadb to listen on ::1 and can't reach it on 127.0.0.117:34
@fungicide:matrix.orgif i adjust the mysqldump to do `-h ::1` it works fine17:34
@fungicide:matrix.orgor the socket binding mariadb uses is af-specific17:35
@clarkb:matrix.orgWe do have nested virt labels now17:36
@mordred:waterwanders.comoh - nice! I'll try those out once I get that far17:36
@clarkb:matrix.orgfungi: oh weird. so if we update the script to use ::1 it will probably work fine then17:36
@fungicide:matrix.orgor maybe we should tell it to bind to `localhost` instead of litersl `::1`17:38
@fungicide:matrix.orgliteral17:38
@clarkb:matrix.orgI wonder if they treat localhost special when binding and only do the socket if you do that17:44
@clarkb:matrix.orgsimilar to how it works whne you connect with the client17:44
@clarkb:matrix.orgI think we can start with using ::1 for now to get things working again then decide what if any proper refactoring should happen from there? Do you want to write that chnge or should I?17:44
@fungicide:matrix.orgi can18:02
@clarkb:matrix.orghttps://review.opendev.org/c/opendev/system-config/+/975319 would be a good one to get in soon too so that we can start updating images into the newly refreshed trixie builds18:09
@fungicide:matrix.orgClark: are we still building bullseye images for uwsgi-base? if not, you could just drop the platform profiles from libffi8 instead of extending the list18:12
@fungicide:matrix.orgthough i'm okay with the change either way18:13
@fungicide:matrix.orgfor that matter, the platform:dpkg profile is likely unnecessary these days since we're not building centos images any more (did we ever?)18:14
@clarkb:matrix.orgfungi: we are not building bullseye anymore. But I don't think it hurts to have there18:14
@clarkb:matrix.orgI can't ever remember building anything but debian images for this stuff.18:15
@fungicide:matrix.orgyeah, i'm fine approving as-is18:15
@clarkb:matrix.orgI was trying to keep the diff small and bindep should ignore the non applicable data anyway18:16
-@gerrit:opendev.org- Jeremy Stanley https://matrix.to/#/@fungicide:matrix.org proposed: [opendev/system-config] 975696: Backup keycloak database over IPv6 loopback https://review.opendev.org/c/opendev/system-config/+/97569619:18
-@gerrit:opendev.org- Zuul merged on behalf of Clark Boylan: [opendev/system-config] 975319: Update python base images https://review.opendev.org/c/opendev/system-config/+/97531919:19
@clarkb:matrix.orgfungi +2 from me on that one but we should probably go ahead and approve it before the 0500 ish backups early tomorrow19:21
@clarkb:matrix.orgcool all of those container images promoted successfully according to zuul19:23
@clarkb:matrix.orgfungi: have time for a quick review on https://review.opendev.org/c/opendev/grafyaml/+/975334 I think that is the most trivial of the trixie updates19:23
@fungicide:matrix.orglooking19:27
@fungicide:matrix.orglgtm, let's see how it fares on deploy19:28
-@gerrit:opendev.org- Zuul merged on behalf of Clark Boylan: [opendev/grafyaml] 975334: Update to python3.12 trixie container https://review.opendev.org/c/opendev/grafyaml/+/97533419:37
@clarkb:matrix.orgthat may not deploy directly. We may need to do something to force the infra-prod-service-grafana job to run. but the ci jobs for grafyaml include testing against grafana so I 'm sure its fine19:40
@fungicide:matrix.orgoh right, it can't trigger that job because it's a change for another repository19:56
@fungicide:matrix.orgthough are dispatch jobs a possible solution to that, longer term?19:57
@fungicide:matrix.orgi'm still a little fuzzy on how those work19:57
@fungicide:matrix.orgi guess the real issue is that the secret for connecting to the bastion to run our nested deploy ansible is for system-config, so grafyaml doesn't have clearance to use it even if it could run the job from its own pipeline config19:58
@fungicide:matrix.orgor, not secret, project ssh key?19:59
@fungicide:matrix.orgregardless, roughly the same problem19:59
@fungicide:matrix.orgyeah, i guess a dispatch job could still only run other jobs for the same project pipeline? which makes sense security-wise20:01
@clarkb:matrix.orgyup its the project ssh key20:06
@clarkb:matrix.orgfungi: since diablo_rojo says there isnt' any need for ptgbot for a bit should we proceed with https://review.opendev.org/c/openstack/ptgbot/+/975325 ?21:43
@fungicide:matrix.orgyes. yes we should21:44
@clarkb:matrix.orgfungi: it failed to fetch quay.io/opendevorg/python-base:3.12-trixie on what looks like a short read22:06
@fungicide:matrix.org`ERROR: failed to build: failed to solve: quay.io/opendevorg/python-base:3.12-trixie: failed to resolve source metadata for quay.io/opendevorg/python-base:3.12-trixie: short read: expected 1814 bytes but got 0: unexpected EOF`22:06
@fungicide:matrix.orgyeah, just saw that22:06
@clarkb:matrix.orgfungi: I am able to fetch and run that image locally so I don't think the problem is the image or quay, but maybe related to the thing sean pointed out earlier22:06
@clarkb:matrix.orgI think we can recheck it22:06
@fungicide:matrix.orgagreed, done22:07
@clarkb:matrix.orgthis build ran in rax ord not dfw though22:07
@clarkb:matrix.orgbut possible they are rolling out network updatse across the clouds or something22:07
@fungicide:matrix.orgor intentionally degrading performance on rax classic in order to encourage people to move to flex if the price increase hasn't already motivated them (only half joking)22:08
@clarkb:matrix.orgonce the change lands and image promotes the hourly eavesdrop deploy should update the bot22:21
@jim:acmegating.comwhat's the thing sean pointed out?22:22
@fungicide:matrix.orgcorvus: https://meetings.opendev.org/irclogs/%23openstack-infra/%23openstack-infra.2026-02-04.log.html22:24
@fungicide:matrix.orglooked like a truncated read downloading a package22:24
@fungicide:matrix.orgso superficially similar22:24
@fungicide:matrix.orgthough was in a different rackspace region and communicating with our mirror server22:25
@clarkb:matrix.orgfungi: but the mirror server was proxying to pypi outside of the cloud network22:26
@fungicide:matrix.orgright, so could have been cascading from the other side of the proxy too22:26
-@gerrit:opendev.org- Zuul merged on behalf of Monty Taylor https://matrix.to/#/@mordred:inaugust.com: [openstack/project-config] 975673: Add wandertracks-android repo for WanderTracks https://review.opendev.org/c/openstack/project-config/+/97567322:38
@clarkb:matrix.orgAssuming nothing comes up between now and then I'll probably target friday for updating gerrit to 3.11.8. If anyone wants to weigh in on switching to tags for builds now would be the time22:38
@clarkb:matrix.orgI followed up in the openinfra events irc channel where ptgbot lives, but it reconnected as part of eavesdrop hourly deploys and the debug log for the bot indicates it connected using tls 1.3 so  Ithink that is working well23:11
@fungicide:matrix.orgyeah, lgtm too23:23

Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!