| -@gerrit:opendev.org- Zuul merged on behalf of Benedikt Löffler: [zuul/nodepool] 925320: Support AWS IMDSv2 in fleet API https://review.opendev.org/c/zuul/nodepool/+/925320 | 08:46 | |
| -@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 925476: Fix wrong blob cleanup for items w/o job graph https://review.opendev.org/c/zuul/zuul/+/925476 | 12:13 | |
| @tristanc_:matrix.org | Hello folks, our users are worried that a change with a depends-on on GitHub is able to trigger a rate limit, effectively causing a zuul denial of service. This just happened because of https://review.opendev.org/c/openstack/neutron/+/925446 , even though the neutron repository is configured to not include any zuul config in our tenants config. | 12:14 |
|---|---|---|
| @tristanc_:matrix.org | Would it be possible to improve the scheduler to avoid consuming api credits in such a case, or limit the impact of a rate limit to this single buildset? | 12:15 |
| -@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 925476: Fix wrong blob cleanup for items w/o job graph https://review.opendev.org/c/zuul/zuul/+/925476 | 12:18 | |
| -@gerrit:opendev.org- Benjamin Schanzel proposed: [zuul/zuul] 924067: Add filtering support for pipeline status views https://review.opendev.org/c/zuul/zuul/+/924067 | 12:36 | |
| @jim:acmegating.com | mordred: usually just racy tests; the whole suite tends to fail catastrophically if some of the individual tests have a timeout. finding and fixing those is difficult. | 13:45 |
| @jim:acmegating.com | Clark: https://review.opendev.org/921637 is needed for ansible 9; the js lint jobs (and likely others) are broken without it | 13:45 |
| @jim:acmegating.com | that's failing with this: https://zuul.opendev.org/t/zuul/build/9704247d140a44e68fa789f27b1e6e62/console | 14:00 |
| @jim:acmegating.com | it's failing to install ansible 8 in a tox environment on an ubuntu focal node | 14:01 |
| @jim:acmegating.com | https://zuul.opendev.org/t/zuul/builds?job_name=zuul-jobs-test-fetch-sphinx-tarball-ubuntu-focal&project=zuul/zuul-jobs | 14:01 |
| @jim:acmegating.com | the build history for that job says it worked until early june, then stopped | 14:01 |
| @jim:acmegating.com | i can reproduce that in a local focal docker container | 14:01 |
| @jim:acmegating.com | was there a setuptools/pip/something change around then that might be involved? | 14:02 |
| @clarkb:matrix.org | corvus: I guess that was an optimization to speed up those tasks. One odd thing is my example from yesterday ran nox jobs successfully but the nox role also gets this cleanup so maybe it isn't always fatal? | 14:03 |
| @jim:acmegating.com | `8.0.0 Requires-Python >=3.9;` (from the failing focal job -- that seems unlikely to have changed retroactively?) | 14:03 |
| @jim:acmegating.com | Clark: there's a "when nox_constraints_file is defined" on the nox task | 14:04 |
| @jim:acmegating.com | probably tox too | 14:04 |
| @jim:acmegating.com | so that may be why we're only seeing it on js jobs in the zuul project | 14:04 |
| @clarkb:matrix.org | corvus: the error in that job appears to come from pip and ya it's saying the python version isn't valid | 14:05 |
| @clarkb:matrix.org | Ah that makes sense for why bindep was fine | 14:05 |
| @jim:acmegating.com | right, and the python version being invalid seems like a really weird thing to start happening 2 monhs ago | 14:05 |
| @clarkb:matrix.org | Maybe we changed the Ansible version. In fact I bet that's it | 14:06 |
| @clarkb:matrix.org | But this job didn't run when we did so? | 14:06 |
| @jim:acmegating.com | oh yeah, bet so | 14:06 |
| @clarkb:matrix.org | We could use python_version markers to run with older Ansible on focal/python3.8 | 14:07 |
| @jim:acmegating.com | yep Icd563def65dcfd40b174218cc4e2b94e0230c374 | 14:07 |
| @clarkb:matrix.org | Or drop the job and call jammy good enough | 14:07 |
| @jim:acmegating.com | ansible>=6,<9;python_version<'3.9' | 14:09 |
| ansible>=8,<9;python_version>='3.9' | ||
| @jim:acmegating.com | something like that? | 14:09 |
| @clarkb:matrix.org | Ya assuming Ansible 6 is old enough to work on python 3.8 | 14:10 |
| @jim:acmegating.com | yeah i just took a guess | 14:10 |
| @jim:acmegating.com | i think i'll squash that into lukas's change | 14:11 |
| @clarkb:matrix.org | Works for me | 14:12 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed on behalf of Lukas Kranz: [zuul/zuul-jobs] 921637: Remove get_md5 parameter from stat module. https://review.opendev.org/c/zuul/zuul-jobs/+/921637 | 14:13 | |
| @jim:acmegating.com | yay we will run many more tests now. :| | 14:15 |
| @clarkb:matrix.org | oh heh I just approved it so I guess we'll see if those extra tests pass | 14:16 |
| -@gerrit:opendev.org- Zuul merged on behalf of Lukas Kranz: [zuul/zuul-jobs] 921637: Remove get_md5 parameter from stat module. https://review.opendev.org/c/zuul/zuul-jobs/+/921637 | 14:28 | |
| @jim:acmegating.com | huzzah | 14:28 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 925506: Add test for blobstore cleanup fix https://review.opendev.org/c/zuul/zuul/+/925506 | 15:41 | |
| -@gerrit:opendev.org- Zuul merged on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/zuul] 925141: Don't log tracebacks on missing jobs in freeze https://review.opendev.org/c/zuul/zuul/+/925141 | 15:42 | |
| @jamesharrisg:matrix.org | > <@jim:acmegating.com> https://zuul-ci.org/docs/zuul/latest/job-content.html#var-nodepool.slot | 16:32 |
| Thanks. That actually would probably be the way to do it. | ||
| @jamesharrisg:matrix.org | though, looking at this again. Is there a way to tell ansible that the current working directory of the ansible jobs is something other than the ansible_user_dir? | 16:35 |
| @jim:acmegating.com | no, that's why the zuul-jobs have the workspace variable convention | 16:36 |
| @jamesharrisg:matrix.org | many of the default zuul roles use `zuul_workspace_dir: .` | 16:36 |
| @jamesharrisg:matrix.org | it doesn't appear to flow throughout the thing. I.e. some jobs have `zuul_work_dir: "{{zuul.project.src_dir}}"` without prepending `zuul_workspace_root` | 16:37 |
| @jim:acmegating.com | i think patches to flesh that out would probably be welcome; i suspect that's just oversight. | 16:38 |
| @jamesharrisg:matrix.org | ah. that makes sense there. | 16:39 |
| @jim:acmegating.com | hrm, maybe not -- i think it's zuul_work_dir i'm thinking of that should be universal in zuul-jobs | 16:40 |
| @jim:acmegating.com | yeah, zuul_workspace_root is just for the prepare workspace roles | 16:41 |
| @jamesharrisg:matrix.org | but since `zuul_workspace_root` sets the parent path for all `zuul.project.src_dir` shouldn't it inherit that on nodes? | 16:43 |
| @jamesharrisg:matrix.org | in the default cause? | 16:43 |
| @jamesharrisg:matrix.org | * in the default case? | 16:43 |
| @jamesharrisg:matrix.org | otherwise changing your zuul_workspace_root will require you to sett a zuul_work_dir for all of your tasks manually to point to where you actually cloned your stuff. | 16:44 |
| @jamesharrisg:matrix.org | * otherwise changing your zuul\_workspace\_root will require you to sett a zuul\_work\_dir for all of your roles manually to point to where you actually cloned your stuff. | 16:45 |
| @jim:acmegating.com | you could set it at a higher level than the tasks; they should be default vars | 16:45 |
| @jamesharrisg:matrix.org | I guess that is true, but it sucks that the default configuration doesn't automatically include the `zuul_workspace_root` | 16:45 |
| @jim:acmegating.com | also, i'm going to retract my statement about ansible_user_dir; you might be able to set or override that for a host. you may want to check that. | 16:47 |
| @jamesharrisg:matrix.org | technically ansible_user_dir is a fact you gather based on your user name. | 16:48 |
| @jim:acmegating.com | > <@jamesharrisg:matrix.org> I guess that is true, but it sucks that the default configuration doesn't automatically include the `zuul_workspace_root` | 16:48 |
| i'm not sure i'd say it sucks. it's not clear to me that you can't accomplish what you want without setting one or two variables. there's a lot of flexibility there to handle a lot of different cases. | ||
| @jamesharrisg:matrix.org | I guess. I would expect the default behavior of something executing in the project src directory to look somewhere relative to zuul_workspace_root for the project src directory. | 16:51 |
| @jamesharrisg:matrix.org | I also really don't want to change ansible_user_dir, because, many of the roles that effect things like ssh keys assume that that points to the user's home directory. | 16:52 |
| @jim:acmegating.com | i agree it's worth considering changing the default; it would probably benefit from some careful contemplation, and possibly review of the history of those 2 vars. might want to ask on the mailing list if anyone sees problems with that before you start. | 16:53 |
| @jamesharrisg:matrix.org | ok. that is fair. | 16:54 |
| @jamesharrisg:matrix.org | It may be some time before I can get to working directly on it though (or other colleagues of mine might take up the effort if they need it sooner.) but I think it is worth pursuing so I will ask about it. | 16:55 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 925506: Add test for blobstore cleanup fix https://review.opendev.org/c/zuul/zuul/+/925506 | 17:03 | |
| -@gerrit:opendev.org- Monty Taylor https://matrix.to/#/@mordred:inaugust.com proposed: [zuul/zuul] 925434: Combine reserved variable name lists in check_varnames https://review.opendev.org/c/zuul/zuul/+/925434 | 21:02 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: | 21:49 | |
| - [zuul/zuul-jobs] 925539: Update test-prepare-workspace-git to use a module https://review.opendev.org/c/zuul/zuul-jobs/+/925539 | ||
| - [zuul/zuul-jobs] 925540: Synchronize test-prepare-workspace-git to prepare-workspace-git https://review.opendev.org/c/zuul/zuul-jobs/+/925540 | ||
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!