Corrupted data in old archive

[Ploppz]

This is not something critical, as I have all data (I think) that I put into these backups.
I only did these backups with rdedup in 2020 as a redundant experiment.
I wanted to load one of the backups just to check whether a file I'm looking for is there.

I have ~/bin/rdedup binary that is unchanged since I archived these files in 2020 with the very same binary... I think (is there some way to verify version/commit?)

$ rdedup load archive-2020-07-09
Enter passphrase to unlock:
Error: 1e81d1929cdf179c611784acf1a778413952d2f656d1a9c8a2876969137e4d88 corrupted, data read: 6d1f87bc9c0bbf476990178238313b58b326b78607135f7efa1b0b9ec099d90b

[dpc]

Huh. That's the first case ever someone reports a data corruption, AFAIR. Though I don't know how many users actually use rdedup. The chunk is identified by its content's hash, so there's not a lot that can go wrong really - file gets written, fsynced, renamed to its final destination. There is a chance that for whatever reason the file was actually corrupted.

If you want and still have the original data, you could actually move this file someplace else, then backup the same (or even similar) data again and that should recreate the same file. Then you could compare the content of the corrupted and fresh version. Not sure how much it is going to tell us, but if it looks like a single bit being flipped or a localized change, then it could mean just disk corruption.

That's actually a weak point of rdedup - it does not do any error correction (it could be added).

Thanks for the report anyway!