Bitrot mitigation

[zatricky]

I'm considering using borg (capitalisation?) for remote backup, hence this question.

Does borg have a built-in feature to mitigate bitrot?

It is one of the few features that I'm unable to verify. Admittedly, bup seems to be the only other backup tool that specifically mentions the feature, though bup doesn't seem appropriate for my use-case anyway.

[enkore]

Borg has strong integrity checking, to detect bitrot and tampering.

check --repair can repair files with corrupted data (replace it with runs of zero), and heal them later should the data become available in the repo again.

Forward error correction (FEC) which would mean that bitrot can be corrected in place is not implemented at this time (#225). #225 hasn't been implemented yet for a few reasons; there is no one in the project with the necessary expertise, the storage format doesn't support adding it "out of the box", it's unclear how a correct implementation would look like that is actually efficient against rust bitrot and so on. (FAQ)

We looked into it a while ago, but it's just non-trivial with so many parameters to choose and get right/wrong. We also found that most FEC algorithms are designed to mitigate transmission errors and are mostly meant to be efficient in hardware, not software, making them rather slow. Storage FEC mostly deals with distributing data among a group of drives assuming total loss of a number of drives - not partial corruption of a drive.

[ThomasWaldmann]

We already have #225, so close this one?

[enkore]

I guess. ftr: duplicate / closed shouldn't stop anyone from asking follow up questions :)