kopia has been great in my usage so far but I have just started it on a problem task: making a backup of a time machine backup drive. Time machine replicates the filesystem over and over again in directories named for the date of the snapshot, with duplicate files being hard links.
If kopia is smart enough to recognize these duplicate inodes and avoid rehashing them, the backup will be very fast, because there’s only 1 terabyte of data on the drive and virtually all of its content is already in the kopia blob store. I know this because there is already a snapshot of the latest time machine backup in the store, so only files that were heavily modified or deleted will create any new blobs.
So, the question is: will kopia skip even checksumming duplicate inodes, or will it wind up checksumming every file repeatedly? On this particular drive, there are 76 time machine snapshots, meaning it’s the difference between hashing about 1.1 terabytes and hashing 76 terabytes.