I’ve got my main house server that has a number of dockerized applications, including nextcloud-aio. Nextcloud-AIO comes with a built-in backup system using BorgBackups. I’ve had this running and doing my backups, it is probably fine. Notable, it does encrypt the backup.

Now, I recently setup a separate machine to use rsnapshot to backup the things from the main machine that need backing up. It is SSHing on a schedule to do that, and backing up the folders I’ve listed.

When I set that up, I skipped the nextcloud borg backup, because that is already backing up; however, it is not a remote backup, so is of limited use (granted, my ‘official’ backup computer is using about 18 inches away from the main server, so also of limited use).

I can easily just include the nextcloud-borg-directory on the rsnapshot list, but does anyone know if it will properly handle just the updates?

That is, both Borg and Rsnapshot are set up so that each backup isn’t a complete backup but just incremental changes, so that you don’t fill your whole disk in two weeks. But if Borg does that first on the nextcloud data, will rsnapshot just not work and then try to backup the full 50GBs every day? Or just do the incremental changes? Will the borg encryption jack up the ability of rsnapshot to see the changes?

If no one knows, I will just do it anyway and report back in a few days if my disk is completely full or not.

Edit: it has been ~4 days, and I think it is not all busted (not going to say it is a good idea). The total space it is taking up on the second (backup) machine is what I expect - it hasn’t ballooned because it can’t properly grok the borg backup format or anything like that. Importantly, this is after ~4 days and very few changes (updates/deletions/edits) to anything on the nextcloud.

  • @florian
    link
    English
    011 months ago
    1. So, you do want to run rsnapshot on the Borg repository (the destination to which is backed up)? Both rsnapshot and Borg keep a history, so you are keeping a history of when the Borg repository had which history. This will not be particularly efficient nor “as intended”.
    2. be aware that Borg does incremental backups on file chunks, while rsnapshot works on whole files. So if a large file changes, rsnapshot will duplicate the storage used.
    3. a Borg repository is more like a database of chunks (similar to git), while rsnapshot recreates the original backup data.

    According to https://borgbackup.readthedocs.io/en/stable/internals/data-structures.html the repository storage consists of 500mb files. So using rsnapshot on those, will not be work as you want it to.