You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@iceberg.apache.org by GitBox <gi...@apache.org> on 2022/02/25 06:40:36 UTC

[GitHub] [iceberg] kbendick commented on pull request #4210: Infra - Trigger CI using on pull_request, saving push for master branch only

kbendick commented on pull request #4210:
URL: https://github.com/apache/iceberg/pull/4210#issuecomment-1050570508


   > changes LGTM, but how do we want to deal with running CI on release branches, such as `0.14.x`? Once this change is in, merges to those branches won't trigger separate CI runs and we would then rely on the PR CI runs. If that's desired then we can go ahead and merge, but I just wanted to point out the implication of this change here.
   
   That’s a good point. We _do_ make a commit of the version.txt file as part of each release cycle, which I would say is the canonical “commit” of the release (it is the SHA released I believe).
   
   That said, GitHub Actions have gotten _alot_ better and so we can also have CI run on `tag` and then have everything run.
   
   **TLDR** - Let’s open a follow up issue to create a single CI suite that runs everything in CI / tests when we merge into master / on tag (really, regardless of the change set, we should just run all of CI or at least all parts if not all versions when we merge). We can limit the time it takes by changing the run to something simpler (I think we duplicate some build and test runs when we run `Java` and some of the other CI’s together if they share code).
   
   Or we can go all out and make a `controller` ci and convert everything else to run on workflow run. I started an attempt at it, but at the time GitHub Actions were somewhat more beta and I worried the knowledge wouldn’t be there. But we could save a good amount of CI time this way. We’d save CI time, as currently a failure in `Spark` will kill the other parallel Spark jobs (from the same matrix), but not `Flink` or other CI jobs.
   
   
   Java CI fails within minutes and is the most common failure (checkstyle, core, etc). So we could save _alot_ of unnecessary runs this way. And now we at least are already using on `workflow_call` so the syntax isn’t as unknown at a group level like when I did it before.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org