-
Notifications
You must be signed in to change notification settings - Fork 104
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
Error: Cache upload failed because file read failed
should not cause step to fail
#120
Comments
Thanks for the report. I haven't been able to find any way to eliminate these 429 rate limiting errors, but they should only be logged as a warning and not cause the job to fail. (You can see these in your logs). However, the Error that cause the failure is new to me: https://github.com/detekt/detekt/runs/4201692342?check_suite_focus=true#step:6:15, and for some reason it's not being caught in via regular exception handling. It looks like there was a similar issue fixed for the Even with this fix, in many cases it's beneficial to use |
Cache upload failed because file read failed
should not cause step to fail
|
Thank you! Besides this (now hopefully fixed) issue the new action seems to work well, and it's definitely much nicer having Gradle Inc supporting the action and proper cache setup than having to figure it out ourselves.
I think that's a better default. I think on balance it would be better to cache main branch by default and increase cache hits on all PRs. Otherwise if multiple PRs are cached and evict the main branch cache, no new PRs will get the benefit of cache hits. |
I migrated the detekt project to use gradle-build-action when v2 final was released (detekt/detekt@c6f796a).
It appears that the change in cache usage is causing flakiness. I just opened a number of PRs in quick succession, and they all started failing with errors similar to this:
https://github.com/detekt/detekt/runs/4201692342?check_suite_focus=true
I'm guessing this is because the cache step is caching several artifacts at the end of each build job, so we're now hitting rate limits. It's not typical that multiple PRs are opened so quickly, but when it does happen and cache failures occur thanks to rate limiting it shouldn't fail the job.
A workaround would be to only write to cache on the main branch, but I assume if multiple PRs are merged one after the after in quick succession, the same issue would occur.
The text was updated successfully, but these errors were encountered: