Closed JavierSegoviaCordoba closed 9 months ago
Thanks for the report @JavierSegoviaCordoba. The "Unhandled error" is very likely to be the problem, and this appears to be coming from the cache-cleanup
part of the action.
This problem during cache cleanup seems to be causing the rest of the post-action to abort, including writing the job summary. (Which means we can't see a caching summary like this for the MacOS/Windows run).
I suspect that you may have seen a similar issue with the build that should have written the data to the cache, but I'm unable to find that CI run. Can you repeat your experiment after deleting any GitHub Actions cache entries and share the links to both GitHub Actions runs (the one that writes the entries to cache and the one that should read them back).
In any case, the action should be more robust with respect to failures during cache-cleanup.
I confirm it works. But I had to manually delete 30 cache files one by one via GitHub UI 😅
Having old caches that should be unnecessary is happening in all my repositories.
I confirm it works. But I had to manually delete 30 cache files one by one via GitHub UI
Thanks. I was actually hoping to confirm the error-on-store, but I guess whatever transient state caused the error is now gone. I should be able to make the cache-cleanup more robust anyway.
Closing this in preference of #990
I am triggering a matrix with the three OS. Calling it two times in a row with a
workflow_dispatch
should avoid rerunning all cacheable tasks in the second run, but this is only happening on Ubuntu.Looks like there are some issues with the action related to the accessors (it shows Gradle 8.4, but the project is using Gradle 8.5).