| -@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 810413: Periodically run malloc_trim via apscheduler https://review.opendev.org/c/zuul/zuul/+/810413 | 09:12 | |
| -@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 810413: Periodically run malloc_trim via apscheduler https://review.opendev.org/c/zuul/zuul/+/810413 | 09:13 | |
| -@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 810413: Periodically run malloc_trim via apscheduler https://review.opendev.org/c/zuul/zuul/+/810413 | 11:27 | |
| @clarkb:matrix.org | swest: ^ left a couple of thoughts on that | 15:56 |
|---|---|---|
| @westphahl:matrix.org | Clark: thanks! We've been using this change in production for quite a while now and I just had to rebase it due to a merge conflict. Not sure if it's something we want to merge | 16:01 |
| @clarkb:matrix.org | oh I see | 16:01 |
| @clarkb:matrix.org | if you find it useful then maybe we should make it mergeable? | 16:02 |
| @clarkb:matrix.org | In opendev we restart weekly making this less important for us | 16:02 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: | 19:59 | |
| - [zuul/zuul] 902855: Index job map by uuid https://review.opendev.org/c/zuul/zuul/+/902855 | ||
| - [zuul/zuul] 903536: Store builds on buildset by uuid https://review.opendev.org/c/zuul/zuul/+/903536 | ||
| -@gerrit:opendev.org- Zuul merged on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/nodepool] 902428: Fix gpu parameter type in openshiftpods https://review.opendev.org/c/zuul/nodepool/+/902428 | 20:36 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 894839: Use a seeded PRNG when setting timer triggers https://review.opendev.org/c/zuul/zuul/+/894839 | 21:22 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 894839: Use a seeded PRNG when setting timer triggers https://review.opendev.org/c/zuul/zuul/+/894839 | 21:24 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 894839: Use a seeded PRNG when setting timer triggers https://review.opendev.org/c/zuul/zuul/+/894839 | 21:25 | |
| @jim:acmegating.com | zuul-maint: how does this look for a nodepool release? commit 42f9100d823ef96d164609011d3ede20413231d7 (HEAD -> master, tag: 9.1.0, origin/master) | 21:33 |
| @clarkb:matrix.org | I'll need to update the annual report again :) | 22:42 |
| @dancywmf:matrix.org | Hi Zuul folks! Hopefully a quick question: Is opendev.org's Zuul installation deployed using containers? | 22:56 |
| @dancywmf:matrix.org | (specifically, the containers created by the Dockerfile in the zuul/zuul repo). | 23:00 |
| @fungicide:matrix.org | dancywmf: yes, precisely those containers | 23:01 |
| @dancywmf:matrix.org | > <@fungicide:matrix.org> dancywmf: yes, precisely those containers | 23:01 |
| Excellent. Thank you fungi! | ||
| @fungicide:matrix.org | dancywmf: for reference, an example is https://opendev.org/opendev/system-config/src/commit/6ba06de/playbooks/roles/zuul-scheduler/files/docker-compose.yaml#L8 | 23:02 |
| @fungicide:matrix.org | and we perform an automated rolling restart early utc every saturday | 23:02 |
| @fungicide:matrix.org | i was about to say we don't build our own zuul container images, but i guess you could say we do since we host the zuul that builds those zuul container images | 23:04 |
| @fungicide:matrix.org | regardless, we don't customize the images for opendev, the published images are the ones we deploy | 23:05 |
| @jim:acmegating.com | `we` for some values of `we` | 23:05 |
| @fungicide:matrix.org | oui oui | 23:05 |
| @dancywmf:matrix.org | Is the rolling restart happening to deal with a particular problem? | 23:06 |
| @jim:acmegating.com | only for upgrades | 23:06 |
| @fungicide:matrix.org | yes, the problem of wanting to be close to the current state of zuul upstream | 23:06 |
| @fungicide:matrix.org | i should have phrased it "rolling upgrade" | 23:06 |
| @jim:acmegating.com | we've actually considered just continuously restarting, but we would effectively dedicate an executor to that job, and we haven't felt it worthwhile yet. we might get there though. :) | 23:07 |
| @fungicide:matrix.org | (the independent components restart onto the latest images in a coordinated fashion so that the service itself remains fully operational the entire time) | 23:07 |
| @jim:acmegating.com | but very little thought goes into it -- it's basically just automated continuous deployment at a 1 week resolution. | 23:07 |
| @fungicide:matrix.org | occasionally our users spot a regression that slipped past a blind spot in test coverage, otherwise the only way people even notice the upgrading is that new features appear | 23:08 |
| @dancywmf:matrix.org | OK. Makes sense. Thanks for the info! | 23:08 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!