General questions about the purpose of users within bbs

[ovizii]

I've fired up a test instance of bbs and created a user, assigned a client to the user and noticed:

• repositories are per client so no deduplication possible if similar data is being backed up. Is this a limitation of borg or of bbs?
• there seems no option to enable quotas on repositories. Is this possible or maybe even on the roadmap?
• each user seems to be able to add repositories but also local ones, meaning if the user is another family member, the risk is having too many repos created and possibly abandoned. Is there an admin view to see all repos and possible clean them up or delete empty ones?

I was assuming users were meant to give i.e. family members a tool to back up their machines, but this seems risky: no control over quota, no control over randomly created repos by users.

So leaving aside the family members scenario, if I were the only user wanting to back up 3 different machines I wouldn't need the “user feature” I think?
There wouldn't be any deduplication happening across the 3 machine's data, right?

I have read the docs about S3 Offsite Sync. Question incoming:

• there seems no option to create a backup in parallel on several repositories or even with different retention. The only option seems to be to sync a full repo to S3, right?
• again, this seems to be “per client” meaning there is no easy way to sync all repos to S3, it must be done per client and repo, right? The docs mention: Method A: Global S3 Settings (Recommended for Single-Tenant) => Settings → Offsite Storage tab with S3 configuration form but there is no such setting. Found it. Its under storage => S3 Sync Settings

General question: there seems no support for using rclone as a storage provider with whatever remote rclone supports. I mean I can set this up outside bbs and mount a path served by rclone via a docker bin mount but if that could be done from within bbs that would be even more awesome.

EDIT: mistakenly pressed ENTER before I was done writing.

EDIT2: Bonus question: how do I switch bbs and the clients to borg v2 for testing?

[marcpope]

This isn't really designed to be a "home" solution. This was originally designed to be used in a hosting company that was backing up dozens of servers. That's not to say home users don't use it, but it might be a little overkill. Adding users was something we added later and the main intention of it was so that those hosting customers could do self-restores. I wouldn't call this a true multi-tenant solution, that would take much more work.

Quota would be tricky, a ton of people have this in Docker and also bare metal and many other storage loations. Enforcing a quota would be complex. Maybe an artificial one would be possible instead of linux style quotas. It would take some doing though. You don't know how large the backup is going to be until its done. And do you want your backups to stop when the quota goes over or just charge those customers for the overage? I'd probably be more on the side of having a number you could put and a report would tell you if someone was over their limit with an option to stop backups (seems dangerous).

BBS isn't designed to have a combined repo for multiple clients. Borg does allow this, but this software does not because each client was intended to have it's own backup space for security. You can achieve de-duplication through the underlying file system if you want that, e.g. using zfs for example.

The docs may need updating there's been 100s of changes lately so menu items might not be in the same places. You can create global S3 Settings and use them on every repo, but you have to specifically mark each repo to sync. It just uses shared credentials. You can also set S3 settings per client/repo. S3 Sync happens after the backup. Borg 1.x does not support backup to S3. This software is designed for 1.x only, not 2.x. 2.x is beta and shouldn't be used in production. Once it is released, we'll look at it, but it will require major parts of the software to be re-built. It's not backwards compatible. Rclone method you mention is not recommended or supported in 1.x. Rclone is used post-backup to sync after the backup. This is generally for off-site backups of local backups. If you have something like rsync.net, it's already off-site.