| -@gerrit:opendev.org- Andrei Dmitriev proposed: [zuul/nodepool] 916728: Introduce use-internal-ip parameter to OpenStack https://review.opendev.org/c/zuul/nodepool/+/916728 | 06:40 | |
| -@gerrit:opendev.org- Bernhard Berg proposed: [zuul/zuul-jobs] 917539: prepare-workspace-git: Add allow deleting current branch https://review.opendev.org/c/zuul/zuul-jobs/+/917539 | 07:27 | |
| -@gerrit:opendev.org- Andrei Dmitriev proposed: [zuul/nodepool] 916729: Cmd: add metastatic backing nodes removal https://review.opendev.org/c/zuul/nodepool/+/916729 | 08:45 | |
| -@gerrit:opendev.org- Andrei Dmitriev proposed: [zuul/nodepool] 916729: Cmd: add metastatic backing nodes removal https://review.opendev.org/c/zuul/nodepool/+/916729 | 09:25 | |
| -@gerrit:opendev.org- Andrei Dmitriev proposed: [zuul/nodepool] 916729: Cmd: add metastatic backing nodes removal https://review.opendev.org/c/zuul/nodepool/+/916729 | 09:31 | |
| -@gerrit:opendev.org- Andrei Dmitriev proposed: [zuul/nodepool] 916729: Cmd: add metastatic backing nodes removal https://review.opendev.org/c/zuul/nodepool/+/916729 | 09:33 | |
| -@gerrit:opendev.org- Bernhard Berg proposed: [zuul/zuul-jobs] 917539: prepare-workspace-git: Add allow deleting current branch https://review.opendev.org/c/zuul/zuul-jobs/+/917539 | 11:27 | |
| @bridgefan:matrix.org | Doe anyone know what user ansible playbooks run as inside, of quay.io/zuul-ci/zuul-executor:10.0.0 ? | 17:16 |
|---|---|---|
| @bridgefan:matrix.org | I recently attempted to do an upgrade from zuul 8.3.2dev to 9.0.0 then 10.0.0 | 17:17 |
| @bridgefan:matrix.org | and after I do that I seem to be getting issues where the executor can't access the zuul-output log server | 17:17 |
| @bridgefan:matrix.org | 2024-04-30 15:36:04.390093 | TASK [upload-logs : Create log directories] | 17:17 |
| 2024-04-30 15:36:04.803953 | localhost | ERROR | ||
| 2024-04-30 15:36:04.804130 | localhost | { | ||
| 2024-04-30 15:36:04.804163 | localhost | "msg": "There was an issue creating /srv/static/logs/81 as requested: [Errno 13] Permission denied: b'/srv/static/logs/81'", | ||
| 2024-04-30 15:36:04.804191 | localhost | "path": "/srv/static/logs/81/681/1/check/appimage-build-emacs-nox/aef2896" | ||
| 2024-04-30 15:36:04.804216 | localhost | } | ||
| @bridgefan:matrix.org | the /srv/static/logs folder is mounted into the executor (using docker-compose) and uses the UID 1000 user | 17:18 |
| @bridgefan:matrix.org | If I exec into the executor I see: | 17:19 |
| @bridgefan:matrix.org | drwxrwxr-x 4 1000 1000 4096 Jun 26 2023 01 | 17:19 |
| @bridgefan:matrix.org | and I can mkdir there just fine (seems I exec as root) | 17:20 |
| @bridgefan:matrix.org | Any idea why I might still be getting permissions issues? | 17:20 |
| @clarkb:matrix.org | the container image wasn't changed between those versions as far as I know | 17:21 |
| @clarkb:matrix.org | perhaps the bind mount side is what chnaged affecting the perms in the container? | 17:22 |
| @bridgefan:matrix.org | the strange thing is i can manually create directories, it's just ansible doesn't like it for some reason | 17:23 |
| @bridgefan:matrix.org | I'll see if the docker-compose for the example change the mounting | 17:23 |
| @bridgefan:matrix.org | I don't think I see the volume mount changing. Though I convert from a docker volume to a folder myself. | 17:31 |
| @bridgefan:matrix.org | Thank you for the idea at least | 17:31 |
| @clarkb:matrix.org | did you check the permissions at the deeper path? | 17:32 |
| @clarkb:matrix.org | the top level perms look good but maybe thei ssue is deeper in the tree | 17:32 |
| @bridgefan:matrix.org | In the above example, /srv/static/logs all exist already but 81 (and everything after) needs to be created | 17:34 |
| @bridgefan:matrix.org | I have checked /srv/static/logs has 775 and just quickly looking through a few other example directories (not 81) I seem to see they also have 775 with same user (UID 1000, not root) | 17:35 |
| @bridgefan:matrix.org | actually you know what I misread /srv/static/logs was only 755 but all subdirectories were 775 | 17:40 |
| @bridgefan:matrix.org | let me see if making /src/static/logs 775 as well, helps | 17:40 |
| @bridgefan:matrix.org | Still no luck | 18:06 |
| @clarkb:matrix.org | ya if it was already owned by 1000:1000 then changing the group perms shouldn't matter | 18:13 |
| @bridgefan:matrix.org | I was poking around the container and noticed that there is a user "zuul". When I try to make the directory with it, I see the original permissions denied error | 18:17 |
| @bridgefan:matrix.org | at least I may be able to repro the issue now without ansible | 18:18 |
| @clarkb:matrix.org | looking at the container images the useradd sets uid to 10001. I wonder where 1000 is coming from | 18:21 |
| @clarkb:matrix.org | perhaps from the hosting side and you just need to explicitly chown to 10001 | 18:21 |
| @bridgefan:matrix.org | 1000 should be coming from the host side user (default user of 1000) | 18:22 |
| @bridgefan:matrix.org | since the logging container also needs the same folder, I wonder if chowning to 10001 would be safe. It is an interesting idea to try though | 18:22 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 917782: Speed up merger git resets https://review.opendev.org/c/zuul/zuul/+/917782 | 18:31 | |
| @bridgefan:matrix.org | Finally solved the issue. I needed to chown to root:root. Interestingly enough 10001:10001 didn't work for me. I suppose it makes sense given that I found the zuul.conf for the example zuul docker-compose which has a default_username=root line. | 19:13 |
| @bridgefan:matrix.org | Thanks for your help Clark | 19:13 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 916142: Ignore /COMMIT_MSG in files matchers even more https://review.opendev.org/c/zuul/zuul/+/916142 | 20:29 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 916142: Ignore /COMMIT_MSG in files matchers even more https://review.opendev.org/c/zuul/zuul/+/916142 | 20:29 | |
| @clarkb:matrix.org | corvus: why is the function in https://review.opendev.org/c/zuul/zuul/+/917782/1/zuul/merger/merger.py called `getPackedRefs` when it doesn't seem to list only packed refs? | 22:19 |
| @clarkb:matrix.org | or maybe for_each_ref is packed ref specific? | 22:19 |
| @clarkb:matrix.org | The reason I ask is that I'm worried we're skipping some refs potentially. The old code that used that method fell back to looking up refs directly if they didn't end up in that list which should be safe but we don't seem to do that anymore | 22:21 |
| @clarkb:matrix.org | and if we don't do that anymore because it is unnecessary because we're actually looking at all refs then maybe the function should be renamed? | 22:21 |
| @clarkb:matrix.org | 2e785ac7b69b8730525a4fc3c7463799e19baf60 is the commit that simplified thigns and stopped doing the fallback | 22:22 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: | 22:36 | |
| - [zuul/zuul] 915514: Refactor branch cache to support more queries https://review.opendev.org/c/zuul/zuul/+/915514 | ||
| - [zuul/zuul] 915515: Add a github graphql query for branch protection https://review.opendev.org/c/zuul/zuul/+/915515 | ||
| - [zuul/zuul] 915516: Add support for excluding locked branches https://review.opendev.org/c/zuul/zuul/+/915516 | ||
| @jim:acmegating.com | Clark: i see nothing in the docs for "git for-each-ref" that would suggest that it's only packed-refs | 22:42 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 917782: Speed up merger git resets https://review.opendev.org/c/zuul/zuul/+/917782 | 22:47 | |
| @jim:acmegating.com | Clark: additionally, my local testing had a mixture of packed and loose refs with no observed problems. | 22:51 |
| @clarkb:matrix.org | ya I would also expect the test suite to actually be almost 100% loose refs since those repos are created on the fly | 23:00 |
| @clarkb:matrix.org | and won't have time to repack during a test run | 23:00 |
| @clarkb:matrix.org | corvus: I think I identified why the tests on the commit message change are failing | 23:02 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 916142: Ignore /COMMIT_MSG in files matchers even more https://review.opendev.org/c/zuul/zuul/+/916142 | 23:09 | |
| @jim:acmegating.com | > <@clarkb:matrix.org> and won't have time to repack during a test run | 23:10 |
| yes -- though we do have at least one test that explicitly repacks | ||
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!