Description
First of all thanks for this action 😄
Today I had my pre-commit job failing, with the following error, which isn't fixable by rerunning the job:
Error: reserveCache failed: Cache already exists. Scope: refs/pull/80/merge, Key: pre-commit-2-b76e203571a33b83fd824866e8e8daaeda30555cba9e70eb155d9e5310e578b8-7365e065f5f4ed7e591c7e54ee4f8c40d559f8ff4d40fb8fc9eec5d1a9d2e11c, Version: 28cdb9f5496f334116f23e86f0063f5d3a9348c2e22425a33171e071aadada7e
My best guess is, that this caused by a corrupted cache, where a cache entry with the key exists, but cache.restoreCache
is unable to restore it properly (maybe on: push
and on: pull_request
wanted to save the cache at the same time? 😵 ).
So this might be a very rare and quite impossible to reproduce case.
Anyway, since the cache is nice and all, but not a key feature, IMHO it would be reasonable to catch errors caused by cache.saveCache
and give users a warning ( core.warning
), instead of letting it bubble up to core.setFailed
.
Line 87 in 763fbaf
Also, there have been some changes to @actions/cache
since version 0.2.1 which you use, so running npm upgrade
might improve this action as well.