I’m mainly curious because we have a growing problem on our restic nodes, where the cache gets a bit out of hand (>200GB’s) on expensive SSD’s.
If it would be safe to delete local cache blobs, we could run a command like the following to remove locally available blobs that were not accessed for 5 days (and are in my eyes not really required in a cache):
find /root/.cache/restic/*/data -atime +5 -delete
At first we tried to run with
--no-cache but this resulted in backup times that were not acceptable. Now I read this, where it is mentioned that
--no-cache is slower also slower then running with a empty cache directory. Which, when thinking about it and knowing the implementation difference for
--no-cache, makes sense.
Now I was just wondering if we’re able to keep our cache sizes small by safely removing not accessed cache blobs, at which point I would expect they’re just fetched again from the backend if needed.