| @chgans:matrix.org | I made a some progress (discovery really) with gerrit topic across repos: I've changed my base job to get the detailed job info (heavily inspired by https://gerrit.googlesource.com/zuul/config/). job wise i only associated a noop job (so no more `required-projects`) and it turned out that the inventory shows that the 2 projects affected by the gerrit topic are checked out, i can see 2 `items` and 2 `projects`: | 02:47 |
|---|---|---|
| @chgans:matrix.org | ``` | 02:48 |
| items: | ||
| - branch: master | ||
| bundle_id: 7b4d8b257cac4f3b9d46a63d1da6316a | ||
| change: '2' | ||
| change_url: http://gerrit:8080/2 | ||
| patchset: '1' | ||
| project: | ||
| canonical_hostname: gerrit | ||
| canonical_name: gerrit/aosp_device | ||
| name: aosp_device | ||
| short_name: aosp_device | ||
| src_dir: src/gerrit/aosp_device | ||
| - branch: master | ||
| bundle_id: 7b4d8b257cac4f3b9d46a63d1da6316a | ||
| change: '1' | ||
| change_url: http://gerrit:8080/1 | ||
| patchset: '1' | ||
| project: | ||
| canonical_hostname: gerrit | ||
| canonical_name: gerrit/aosp_build | ||
| name: aosp_build | ||
| short_name: aosp_build | ||
| src_dir: src/gerrit/aosp_build | ||
| ``` | ||
| @chgans:matrix.org | ``` | 02:48 |
| projects: | ||
| gerrit/aosp_build: | ||
| canonical_hostname: gerrit | ||
| canonical_name: gerrit/aosp_build | ||
| checkout: master | ||
| checkout_description: zuul branch | ||
| commit: efaac64a2f8f41fd120da7d3fbb43a0decdccd1d | ||
| name: aosp_build | ||
| required: false | ||
| short_name: aosp_build | ||
| src_dir: src/gerrit/aosp_build | ||
| gerrit/aosp_device: | ||
| canonical_hostname: gerrit | ||
| canonical_name: gerrit/aosp_device | ||
| checkout: master | ||
| checkout_description: zuul branch | ||
| commit: b6d27533f067ee6cec0b1d9135fa2e1964433edd | ||
| name: aosp_device | ||
| required: false | ||
| short_name: aosp_device | ||
| src_dir: src/gerrit/aosp_device | ||
| ``` | ||
| @chgans:matrix.org | The problem is still the same, 2 different gate jobs get triggered, one get +2, then the other one get a -2 and post a -2 to the former one. The reason being "merge failure". | 02:50 |
| @chgans:matrix.org | I wonder if it's a bug in zull code. It could as well be a bug in the gerrit batch plugin, as the port to 3.6 has not been merged yet: https://gerrit-review.googlesource.com/c/plugins/batch/+/338997 | 02:52 |
| @chgans:matrix.org | Maybe I should report it as a bug? Is this use case tested? Is it testable? not sure this is covered by `test_submitted_together_git` for example | 05:05 |
| @chgans:matrix.org | maybe `test_gerrit_crd.py` is a better candidate | 05:06 |
| @chgans:matrix.org | I've enabled debug mode and i've looked at the logs. | 05:42 |
| @chgans:matrix.org | IMO, core issue is that the deduplication is not working. I'm expecting the job to run once, and then approve both changesets. | 05:47 |
| @chgans:matrix.org | Instead, 2 jobs are running, and they both try to approve both changesets, creating mayhem | 05:47 |
| @vlotorev:matrix.org | > <@chgans:matrix.org> I wonder if it's a bug in zull code. It could as well be a bug in the gerrit batch plugin, as the port to 3.6 has not been merged yet: https://gerrit-review.googlesource.com/c/plugins/batch/+/338997 | 05:48 |
| Christian Gagneraud: Gerrit 'batch' plugin is not required for zuul to work with gerrit topics. | ||
| @chgans:matrix.org | hmm, interesting, i didn't know that. Do you think it is causing issues in my case? | 05:48 |
| @vlotorev:matrix.org | I can't know. Removing unneeded plugins might help root-causing the actual issue. | 05:51 |
| @chgans:matrix.org | trying now... | 05:53 |
| @chgans:matrix.org | same, 2 jobs are triggered and then walk on each other toes... | 05:58 |
| @jsokolvewd:matrix.org | Christian Gagneraud: what version of zuul are you running? Job deduplication is still not released yet | 06:12 |
| @chgans:matrix.org | I'm using doc/sources/example from zuul master. I'm assuming this is running master in containers. | 06:14 |
| @chgans:matrix.org | ```root@7f420a594f15:/# zuul-executor --version | 06:15 |
| Zuul version: 6.0.1.dev63 | ||
| ``` | ||
| @chgans:matrix.org | Oops, looks like i'm running upstream zuul! 🤦 | 06:17 |
| @jsokolvewd:matrix.org | Hmm, this docker-compose refers to docker.io images, but one tagged as `latest` is clearly not the same one as `6.0` tag, so it indeed might be something from master branch | 06:19 |
| @chgans:matrix.org | The symtom i see is clearly that Job deduplication is not working/activated | 06:20 |
| @chgans:matrix.org | ther's no 6.0.1 tag in zull repo.... | 06:20 |
| @chgans:matrix.org | I remember corvus saying that this was an experimental feature. | 06:22 |
| @chgans:matrix.org | looks like i have this feature: | 06:29 |
| ``` | ||
| root@7f420a594f15:/# grep -w deduplicate /usr/local/lib/python3.8/site-packages/zuul/*.py | ||
| /usr/local/lib/python3.8/site-packages/zuul/configloader.py: 'deduplicate': vs.Any(bool, 'auto'), | ||
| /usr/local/lib/python3.8/site-packages/zuul/configloader.py: 'deduplicate', | ||
| /usr/local/lib/python3.8/site-packages/zuul/model.py: 'deduplicate', | ||
| /usr/local/lib/python3.8/site-packages/zuul/model.py: if 'deduplicate' not in data: | ||
| /usr/local/lib/python3.8/site-packages/zuul/model.py: data['deduplicate'] = 'auto' | ||
| /usr/local/lib/python3.8/site-packages/zuul/model.py: d['deduplicate'] = self.deduplicate | ||
| /usr/local/lib/python3.8/site-packages/zuul/model.py: deduplicate='auto', | ||
| /usr/local/lib/python3.8/site-packages/zuul/model.py: if job.deduplicate is False: | ||
| /usr/local/lib/python3.8/site-packages/zuul/model.py: if job.deduplicate == 'auto': | ||
| ``` | ||
| -@gerrit:opendev.org- yatin proposed: [zuul/zuul-jobs] 846546: [DNM] Test proposal playbooks https://review.opendev.org/c/zuul/zuul-jobs/+/846546 | 06:32 | |
| @jsokolvewd:matrix.org | Good, just wanted to make sure it's not the easy stuff :-) | 06:34 |
| -@gerrit:opendev.org- yatin proposed: [zuul/zuul-jobs] 846546: [DNM] Test proposal playbooks https://review.opendev.org/c/zuul/zuul-jobs/+/846546 | 06:40 | |
| -@gerrit:opendev.org- yatin proposed: [zuul/zuul-jobs] 846546: [DNM] Test proposal playbooks https://review.opendev.org/c/zuul/zuul-jobs/+/846546 | 06:45 | |
| -@gerrit:opendev.org- yatin proposed: [zuul/zuul-jobs] 846546: [DNM] Test proposal playbooks https://review.opendev.org/c/zuul/zuul-jobs/+/846546 | 07:13 | |
| -@gerrit:opendev.org- yatin proposed: [zuul/zuul-jobs] 846546: [DNM] Test proposal playbooks https://review.opendev.org/c/zuul/zuul-jobs/+/846546 | 07:25 | |
| -@gerrit:opendev.org- yatin proposed: [zuul/zuul-jobs] 846546: [DNM] Test proposal playbooks https://review.opendev.org/c/zuul/zuul-jobs/+/846546 | 07:58 | |
| -@gerrit:opendev.org- yatin proposed: [zuul/zuul-jobs] 846546: [DNM] Test proposal playbooks https://review.opendev.org/c/zuul/zuul-jobs/+/846546 | 08:06 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Tobias Henkel: [zuul/zuul] 845983: Improve cycles paragraph on gating page https://review.opendev.org/c/zuul/zuul/+/845983 | 08:50 | |
| @q:fricklercloud.de | it seems that setting default-branch via a regex doesn't work? I have https://github.com/osism/zuul-config/blob/main/zuul.d/projects.yaml#L14-L16 but still see | 10:39 |
| Error: Project github.com/osism/ansible-collection-commons does not have the default branch master | ||
| https://zuul.osism.xyz/t/osism/build/59dab755abcc4d4b99f913723dcb9159 | ||
| @fungicide:matrix.org | > <@chgans:matrix.org> ther's no 6.0.1 tag in zull repo.... | 11:13 |
| right, 6.0.1.dev63 is a development version after the 6.0.0 tag. it's "guessing" that the next version will be at least 6.0.1 so is autogenerating a prerelease version number for that (though the next tag could be 6.1.0 or even 7.0.0, it will be at least 6.0.1). since zuul is developed in python, it follows the python community's pep 440 versioning standard, which indicates that versions like 6.0.1.dev63 sort after 6.0.0 but before 6.0.1 | ||
| @q:fricklercloud.de | > <@q:fricklercloud.de> it seems that setting default-branch via a regex doesn't work? I have https://github.com/osism/zuul-config/blob/main/zuul.d/projects.yaml#L14-L16 but still see | 11:35 |
| > Error: Project github.com/osism/ansible-collection-commons does not have the default branch master | ||
| > https://zuul.osism.xyz/t/osism/build/59dab755abcc4d4b99f913723dcb9159 | ||
| if I read the code correctly, regex project definitions are treated like project templates and silently discard a whole list of attributes. maybe it would be better to throw a config error in that case? also update the documentation? also how about being able to set a tenant-wide default-branch? at least until the world is ready to switch the default default from "master" to "main"? | ||
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/nodepool] 846625: Suppress component registry logging in command https://review.opendev.org/c/zuul/nodepool/+/846625 | 14:09 | |
| @jim:acmegating.com | frickler: seems like those attrs may have been added to the wrong class | 14:12 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/nodepool] 846801: Avoid collisions after ZK image data import https://review.opendev.org/c/zuul/nodepool/+/846801 | 20:00 | |
| @clarkb:matrix.org | corvus: see comment on https://review.opendev.org/c/zuul/zuul/+/846469 (I think it is approvable just wanted someone else to double check alex's upstream comment ebfore we do that) | 22:53 |
| @jim:acmegating.com | Clark: thanks; i agree that doesn't change my thinking on our security posture there. | 22:57 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/nodepool] 846806: Support deleting DIB images while builder is offline https://review.opendev.org/c/zuul/nodepool/+/846806 | 22:57 | |
| @clarkb:matrix.org | corvus: couple of thoughts/questions on https://review.opendev.org/c/zuul/zuul/+/846472 too | 23:11 |
| @jim:acmegating.com | Clark: replied, thanks! | 23:19 |
| @clarkb:matrix.org | corvus: and comment on https://review.opendev.org/c/zuul/nodepool/+/846801 | 23:29 |
| @jim:acmegating.com | Clark: yeah i don't love it either, but i haven't found anything better yet (replied) | 23:49 |
| @jim:acmegating.com | i really wish that were just settable using the api :/ | 23:50 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!