i’m using restic on linux server with rclone + google drive as repo (i have shared drive with unlimited space , only 400000 files+directory limit).
so i would keep many backup, and create a script that backup then keep 365 daily backup.
works all fine, but on a server with an hdd 300gb and 200gb of mail / php files i have a cache bigger 50gb (about)
so my question is.
edit script with --no-cache (i don’t have problem with bandwidth), but i think is a longer backup
mount a remote (another datacenter) “sftp/webdavs/google drive” folder and put cache in that directory?
Do not disable the cache. You don’t have a bandwidth problem but you are probably not within 1 ms of latency of the Google Drive endpoint, so this would affect performance probably in an unacceptable way, see:
I certainly don’t like the idea.
If you delete the cache and than make a single backup, is the cache already that big or does it grow to this number of time?
You could delete the cache after every backup run, this is till better than running with --no-cache as per the thread linked above.
Having a cache with low latency helps a lot! Deleting the cache after each backup may cause other problems, like always downloading non-necessary blobs but may be worth a try.
About the large cache size: Make sure you have compressed the metadata. All metadata is in json and compresses really well. Also keeping fewer snapshots might reduce the cache size - if lots of trees are changing quite often.
as a bit of offtopic here, how was the migration process? I have a similar setup than you (using rclone and Google Drive) but I’m guessing the migration will make restic re-upload everything again, isn’t it? So it would be similar as creating another repo or I’m wrong in this?