I just want to confirm if my understanding is correct: In order to move build artifacts between workflows, we have to bring our own externally hosted S3 and use the Crow/Woodpecker S3 plugin, correct? There’s no small S3 bucket provided somewhere that each repo could use?
Yes, Crow doesn’t have an integrated generic caching system yet. See https://codefloe.com/crowci/crow/issues/1248 for the open discussion.
A shared S3 bucket would be a security issue, so we can’t do that. What we would need to do is to bootstrap a unique bucket per org. However, we can’t really control the cleanup in there in a smart way, so the best way is to use your own bucket for this through the S3 Woodpecker plugin.
Thanks, yes, I didn’t mean shared as in physically the same between all tenants. More like a small dedicated bucket each project gets that would be enough to shift a few MB of artifacts here and there to avoid repeating steps between e.g. build and publish.
But in that case I’ll research some hosted S3 options.
We would need to self-host a tool here to be able to set size limits.
At some point, Crow will have a central caching option which most likely will also run with an S3 backend. For such we would have more control then and can set size limits and retention times.