I’m guessing there’s no issue changing this on existing backups, but wanted to check before I do.
I normally always backup with -hash set, because I’ve had issue in some odd cases. Like, with Excel users that open files and time stamps don’t change, yet their in-file metadata will change something small, like “author” will change from “LAYNE” to “EMILY”. the time stamp will not change, the byte size is even the same, but hash will catch it. Not that I care about getting that one Excel file when no actual important data changes, but I don’t trust that I’m getting truely changed files. (had other issues with Cobol files not changing time stamps too, but the file was different with an md5 hash)
None of that matters for my question above, just explaining weird ways I see data at times. Anyway, I’m asking because I recently setup a client’s windows server to backup, and they have way more data than I’ve ever ran into with Duplicacy, something around 700Gb. (nearly 1/3 to 1/2 of which is deduped in storage too! what a mess they have!) When I first backed up with my standard use of -hash, it took nearly 4 hours. That’s just way too long, even on subsequent jobs. (makes sense, due to hashing all files). So I removed -hash from their backup scripts and now it takes about 30 minutes. Perfect.
That’s all great. But now I thinking maybe I could disable -hash use on some other systems and get faster backups, a bit more often. But with about a year of backups archived now, I’d don’t want to mess up those archives. They are on ZFS, I could snapshot it first, but I probably wont notice a problem until I restore at some random point in the future.
Sorry about the long explanations, just want to be sure switching between -hash or not on existing backup storages is ok. Thanks!
Also, if it is OK, I’m considering running a separate script on weekends with -hash enabled. Just to capture those edge case files.