| @iwienand:matrix.org | so moving everything in the container to it's own cgroup does seem to make podman happy; if you move things the right way (processes one by one) | 03:30 |
|---|---|---|
| @iwienand:matrix.org | this might be best practice to do at init time anyway? | 03:31 |
| @iwienand:matrix.org | another reason why i hate the github model | 04:03 |
| @iwienand:matrix.org | https://patch-diff.githubusercontent.com/raw/containers/podman/pull/14904.patch | 04:03 |
| @iwienand:matrix.org | it patches the same file multiple times, so you can't just apply it normally | 04:04 |
| -@gerrit:opendev.org- Ian Wienand proposed: [zuul/nodepool] 849273: [wip] test moving into separate cgroup https://review.opendev.org/c/zuul/nodepool/+/849273 | 05:15 | |
| -@gerrit:opendev.org- Ian Wienand proposed: [zuul/nodepool] 849273: [wip] test moving into separate cgroup https://review.opendev.org/c/zuul/nodepool/+/849273 | 07:09 | |
| -@gerrit:opendev.org- Artem Goncharov proposed: [zuul/zuul] 849720: Implement gitea comment trigger https://review.opendev.org/c/zuul/zuul/+/849720 | 07:55 | |
| -@gerrit:opendev.org- Albin Vass proposed wip: [zuul/zuul] 849795: WIP: Use debian testing version of libc-bin https://review.opendev.org/c/zuul/zuul/+/849795 | 08:36 | |
| @avass:vassast.org | We see better when running ansible 5 and using libc 2.33 from debian testing. But we're gonna run with that for a bit more to see if we have other issues as well. Not sure if it's something we want to upstream, but I pushed a change for it anyway. If it's wanted I can write a better commit message for it :) | 08:42 |
| @avass:vassast.org | * We see better results when running ansible 5 and using libc 2.33 from debian testing. But we're gonna run with that for a bit more to see if we have other issues as well. Not sure if it's something we want to upstream, but I pushed a change for it anyway. If it's wanted I can write a better commit message for it :) | 08:51 |
| -@gerrit:opendev.org- Albin Vass proposed wip: [zuul/zuul] 849795: WIP: Use debian testing version of libc-bin https://review.opendev.org/c/zuul/zuul/+/849795 | 09:15 | |
| -@gerrit:opendev.org- Ian Wienand proposed: [zuul/nodepool] 849273: [wip] test moving into separate cgroup https://review.opendev.org/c/zuul/nodepool/+/849273 | 09:39 | |
| -@gerrit:opendev.org- Artem Goncharov proposed: [zuul/zuul] 849720: Implement gitea comment trigger https://review.opendev.org/c/zuul/zuul/+/849720 | 10:18 | |
| -@gerrit:opendev.org- Artem Goncharov proposed: [zuul/zuul] 849727: Add gitea pr dyn_reconf test https://review.opendev.org/c/zuul/zuul/+/849727 | 10:19 | |
| -@gerrit:opendev.org- Artem Goncharov proposed: [zuul/zuul] 849728: Implement gitea refs https://review.opendev.org/c/zuul/zuul/+/849728 | 10:24 | |
| @benzizbit:matrix.org | Hi all, I'm new to Zuul , and would like to use it in our CI infrastructure. I like to integrate it to replace our current Gerrit plugin for Jenkins. I have noticed that Gearman support was removed in release 5.0. And I did not find any documentation on how to "connect" (trigger Jenkins jobs) on Gerrit event via Zuul. my understanding was that this is where Gearman will step in. I would highly appreciate if someone will refer me to doc or any other place where I can understand how can I integrate Zuul into our CI (currently working with Gerrit + Jenkins + Jenkins gerrit plugin). thanks in advance 😀 | 14:06 |
| @jim:acmegating.com | Albin Vass: is there an explanation somewhere that connects that error with libc versions? the bug reports linked don't seem to explain how that causes an error with ansible | 14:07 |
| @avass:vassast.org | I'm probably missing information in that commit | 14:08 |
| @jim:acmegating.com | Benzi Zbit: hi! around zuul version 3.0 (a few years ago), zuul started running its own jobs and no longer uses jenkins for that. while it's possible to write a zuul job that uses the jenkins api to run a job, it can be quite complex to do so correctly since zuul puts a lot of effort into preparing git repos for a job. most zuul users use zuul without jenkins (or if they do use jenkins, use it as a separate ci system) | 14:09 |
| @avass:vassast.org | corvus: I added a backtrace as a comment in the change | 14:09 |
| @jim:acmegating.com | Albin Vass: thanks. that seems like maybe a good candidate for an ansible bug on its own? | 14:13 |
| @avass:vassast.org | and two links to openpty in ansibles source. That was the second deadlock we mitigated btw, the first one we saw had a different backtrace and was caused by pwd which was removed in some ansible release after v2.9: | 14:13 |
| https://github.com/ansible/ansible/pull/72412 | ||
| @avass:vassast.org | > <@jim:acmegating.com> Albin Vass: thanks. that seems like maybe a good candidate for an ansible bug on its own? | 14:13 |
| yeah probably, but we needed a quicker fix than waiting for an ansible release + zuul to use it :) | ||
| @benzizbit:matrix.org | > <@jim:acmegating.com> Benzi Zbit: hi! around zuul version 3.0 (a few years ago), zuul started running its own jobs and no longer uses jenkins for that. while it's possible to write a zuul job that uses the jenkins api to run a job, it can be quite complex to do so correctly since zuul puts a lot of effort into preparing git repos for a job. most zuul users use zuul without jenkins (or if they do use jenkins, use it as a separate ci system) | 14:15 |
| many thanks for your reply ! | ||
| Do you think going back to (lets say 4.10) where Gearman interface is still present will be a good path to explose? | ||
| @jim:acmegating.com | Albin Vass: oh sure, it would just be a good place to track progress | 14:15 |
| @benzizbit:matrix.org | > <@jim:acmegating.com> Benzi Zbit: hi! around zuul version 3.0 (a few years ago), zuul started running its own jobs and no longer uses jenkins for that. while it's possible to write a zuul job that uses the jenkins api to run a job, it can be quite complex to do so correctly since zuul puts a lot of effort into preparing git repos for a job. most zuul users use zuul without jenkins (or if they do use jenkins, use it as a separate ci system) | 14:18 |
| * many thanks for your reply ! | ||
| Do you think going back to (lets say 4.10) where Gearman interface is still present will be a good path to explore? | ||
| @avass:vassast.org | > <@jim:acmegating.com> Albin Vass: oh sure, it would just be a good place to track progress | 14:19 |
| Yep, gonna see if I can submit some before i go on vacation | ||
| @jim:acmegating.com | Benzi Zbit: i would not recommend running older versions; we're continually improving zuul and older versions aren't maintaned (and some have security issues). if you are interested in migrating from jenkins to zuul, some folks here can offer advice about how to go about that. but if you really want to stick with jenkins in the long run, then i don't think zuul will be helpful. | 14:19 |
| @jim:acmegating.com | Benzi Zbit: here is an overview talk i recently gave on zuul in case it's helpful: https://www.youtube.com/watch?v=aWA8HFxFpos | 14:19 |
| @avass:vassast.org | * Yep, gonna see if I can submit something before i go on vacation | 14:20 |
| @benzizbit:matrix.org | > <@jim:acmegating.com> Benzi Zbit: i would not recommend running older versions; we're continually improving zuul and older versions aren't maintaned (and some have security issues). if you are interested in migrating from jenkins to zuul, some folks here can offer advice about how to go about that. but if you really want to stick with jenkins in the long run, then i don't think zuul will be helpful. | 14:22 |
| Ok, got it! | ||
| Again , many thanks for replay ! it's appreciated | ||
| @benzizbit:matrix.org | > <@jim:acmegating.com> Benzi Zbit: i would not recommend running older versions; we're continually improving zuul and older versions aren't maintaned (and some have security issues). if you are interested in migrating from jenkins to zuul, some folks here can offer advice about how to go about that. but if you really want to stick with jenkins in the long run, then i don't think zuul will be helpful. | 14:23 |
| * Ok, got it! | ||
| Again , many thanks for your replay ! it's appreciated | ||
| @jim:acmegating.com | Benzi Zbit: you're welcome; let us know if you have any more questions :) | 14:23 |
| @benzizbit:matrix.org | > <@jim:acmegating.com> Benzi Zbit: you're welcome; let us know if you have any more questions :) | 14:23 |
| sure ! 😀 | ||
| @fungicide:matrix.org | Benzi Zbit: there's also a blog post here explaining some of the reasons we dropped jenkins: https://zuul-ci.org/blog/#20200207a | 15:01 |
| @clarkb:matrix.org | > <@avass:vassast.org> We see better results when running ansible 5 and using libc 2.33 from debian testing. But we're gonna run with that for a bit more to see if we have other issues as well. Not sure if it's something we want to upstream, but I pushed a change for it anyway. If it's wanted I can write a better commit message for it :) | 16:08 |
| Side note we tried alpine based images with musl when we first containerized and found that musl performance wasn't very good. There is some precedent for adjusting images for libc that suits the workload | ||
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 849887: Add Nodepool in Zuul spec https://review.opendev.org/c/zuul/zuul/+/849887 | 16:28 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Matthieu Huin https://matrix.to/#/@mhuin:matrix.org: [zuul/zuul] 841610: Handle jwt decoding error, fix exception default messages https://review.opendev.org/c/zuul/zuul/+/841610 | 18:29 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Matthieu Huin https://matrix.to/#/@mhuin:matrix.org: [zuul/zuul] 831222: GUI: Do not show sign-in button if no IdP is available https://review.opendev.org/c/zuul/zuul/+/831222 | 18:51 | |
| @vlotorev:matrix.org | Hi, there are no labels on https://zuul.opendev.org/t/opendev/labels, is this expected? I've upgraded to zuul 6.1.0 and zuul web also doesn't show any labels. | 22:37 |
| @clarkb:matrix.org | vlotorev: looks like the server is returning a json doc with an empty list for labels | 22:41 |
| @clarkb:matrix.org | Probably a bug | 22:42 |
| @vlotorev:matrix.org | Clark: thanks for confirming. | 22:48 |
| @blaisep-sureify:matrix.org | A lot of python modules are compiled with `glibc` , it takes a looooog time to build images based on Alpine if glibc is required. | 22:50 |
| I can immediately tell which k8s admins are new because they tend to build everything off Alpine. | ||
| @blaisep-sureify:matrix.org | pythonspeed.com has great articles on using Docker with python projects. | 22:51 |
| @clarkb:matrix.org | Blaise Pabon: our observations were runtime not build time. musl tends to sacrifice memory footpring as it is used in more constrainted environments aiui. Whereas glibc is happy to throw memory at the problem to increase performance | 22:55 |
| @clarkb:matrix.org | But ya I can imagine that you end up building a lot of wheels too | 22:56 |
| @jim:acmegating.com | > <@vlotorev:matrix.org> Hi, there are no labels on https://zuul.opendev.org/t/opendev/labels, is this expected? I've upgraded to zuul 6.1.0 and zuul web also doesn't show any labels. | 23:00 |
| i think i broke that. i can fix it (maybe not today). | ||
| -@gerrit:opendev.org- Ian Wienand proposed: [zuul/zuul-jobs] 849903: upload-pypi: always test upload https://review.opendev.org/c/zuul/zuul-jobs/+/849903 | 23:47 | |
| @iwienand:matrix.org | so ... https://review.opendev.org/c/zuul/nodepool/+/849273/4/Dockerfile works to get podman able to run under docker and do enough to make containerfile elements happy | 23:49 |
| @iwienand:matrix.org | i do not know how i feel about it :) | 23:50 |
| @clarkb:matrix.org | ianw: couple thoughts posted on that | 23:52 |
| @iwienand:matrix.org | yeah i do think the name is arbitrary | 23:53 |
| @iwienand:matrix.org | i might reopen the upstream issue because i don't think the chagne they merged actually fixed the issue in this case | 23:54 |
| @iwienand:matrix.org | i've tried testing it, and i got an error, but it's a big pita to get installed | 23:54 |
| @iwienand:matrix.org | considering the number of issues we've hit, which is now about 5 i think, it might almost be worth the effort of doing a 3rd-party ci thing and figuring out how to pull podman master into the nodepool image | 23:55 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!