Selectively delete file across all backups

[proxwolfe]

So I have PBS running to backup my home lab cluster. And it is working fine and I backup everything there and from there on tape as well.

As data protection is becoming more and more important, I am wondering whether there is a way to selectively remove individual files or folders from not just one backup but rather all backups

Say you are a doctor and need to remove a patient's file or you are a landlord and need to remove a former tenant's info, or you are a lawyer and need to remove a client's data or whatever. You could go in to the datastore, open each individual backup, click your way through to the respective drive, folder and file and delete it. And repeat that for every backup which could be a few, if you backup daily.

So is there maybe a better way to select a particular file (or folder) and have it removed from every backup?

If so, if there maybe even a way to remove that file (or folder) from tape backups?

If not, what is the best practice in the times of GDPR?

 

[AmateurProgrammer]

Good question, I don't think there exists such a feature. How do you remove the individual files? I couldn't find a way to do that. I don't think there is any easy way to even access vm disk filesystems on the pbs side (which would be nice) for file download purposes either, but this can be done on a pve node with access to the datastore. In the case of containers I can open their contents easily in the pbs datastore content view.

I hardly believe that deleting single files (especially in the case of vm:s) is possible, as the data is stored in deduped compressed chunks. For vm:s it would also mean that pbs should first mount the filesystem from those chunks. And when it comes to deleting files from a vm filesystem, it would need a filesystem level operation on those chucks and propagate that through the nested deduped chunk hierarchy. This would be a significant burden on the server and for the dev team to implement and I'm not even sure if that is possible at all with the current chunk storage system.

But yeah, if you currently have the need to actually delete files from PBS backups, afaik the only solution seems to be deletion followed by garbage collection of the backup snapshots that include the file. I think feature that would point out which backups include the said file could be implemented at least for the containers.

 

[czechsys]

That's not possible from definition of the backup. Backup is backup state from past time, modyfiing backup = creating new backup from current time.

Backup is read only. You never delete from backup. You delete backup usually via expiration parameters.

GDPR has nothing to do with it.

 

[AmateurProgrammer]

Yes, it seems that backups do not have to be deleted necessarily, read more here: Quantum about backups and GDPR deletion

 

[ubu]

The only possibility i see is to restore the backup, delete the files from the restored backup , delete the backup and recreate the backup without the files ...
The same problem applies to a backup Tape in a safe, etc