Closed max-sixty closed 5 years ago
This looks like expected behavior. (That's the first 8 chars of your build ID) If you don't add tags to your build definition, it falls back to using the ID. Otherwise, if you have multiple pipelines, they will incorrectly stack on each other.
Here's the code snippet that's doing that: https://github.com/stealthybox/container-builder-github-ci-status/blob/aa35321/index.js#L49-L52
Solution is to add tags to your cloudbuild.yaml:
tags:
- api
- frontend
- lint
I'm going to close this, but if you have suggestions on another way to automatically name the builds, let me know! Using the step names would probably be way too long.
Cheers.
Could we use the branch name?
No, you can run multiple build triggers that perform different operations on a single branch/commit. ( build, test, lint, deploy might all be run from different build triggers )
The status name is used to differentiate that.
Just to come back on this - using tags: ['${TAG_NAME}${BRANCH_NAME}']
works exactly as I had hoped.
Thanks for the pointer @stealthybox
OH WOW; I didn't know you could interpolate into tags! That's super useful for something I was just looking at.
I'm very glad that works :+1:
We recently deployed the most recent version of the function to one of our projects
Unfortunately we're now seeing earlier failures remain as a separate failing test, rather than collapsing all tests together, and only showing the latest result
This includes branches & PRs that were only created since the redeploy
potential ref https://github.com/stealthybox/container-builder-github-ci-status/issues/6. Though I hadn't seen this problem in earlier version of the function
Anything I could look at to get around this?
Thank you!