Borg Backup with Proxmox

[goatonboard]

I was looking for a way to backup the Debian's I run inside proxmox with borg backup. I found it pretty challanging to set up. There is a new borgbackup version borgbackup2 now - quote from the repo:

This is borg2!​

Please note that this is the README for borg2 / master branch.

borg2 is currently in beta testing.

I tested it and it works quite nicely. I didnt see to many differences to the old one in terms of using it.

I found this excellent blogpost where i could just copy paste everything. I used a normal dedicated server that stands next to our proxmox as backup server.

 

[UdoB]

borg2 is currently in beta testing.

Stabilization is hard:

"borg2: it's coming! #6602 Open ThomasWaldmann opened this issue Apr 15, 2022 "

Borgbackp (version 1) is really great program and I do use it at two locations. But then there are 441 open issues listed...

I am really happy that Proxmox PBS offers a similar feature set

 

[enoch85]

I'm looking into Borg since I want to backup the PBS repos to an offsite location. Testing it as we speak, but have anyone else here on this forum used it in the same way? What I'm doing is basically:

Proxmox --> PBS
PBS --> QNAP (over SSH and VPN) where I use different borg repos for each repo on the PBS.

Now, I have two questions so far:

1. My existing backup is already running with rsync, and the folders are quite large. The biggest being almost 10 TB (which took 1 month to sync with rsync over WAN) - I'd rather not do that again, and are wondering if it's possible to index the existing files, or do I have to start over from 0 with borg?
2. Is it possible for borg to sync so that I actually see the files on the remote, or do I have to use borg extraxt, list and so on to see the actual files?

Thanks!

 

[michabbs]

The biggest being almost 10 TB (which took 1 month to sync with rsync over WAN)

With such a big datastore (at least relatively to bandwidth) - any rsync/rclone/borg method seems useless. Most likely you will never get consistent backup (remember the datastore changes all the time). The consistent copy should be "atomic", for example from a snapshot of original datastore. But anyway you will get a month old copy... Forget.

I suggest you set up another PBS in a remove location and use sync feature.

 

[enoch85]

With such a big datastore (at least relatively to bandwidth) - any rsync/rclone/borg method seems useless. Most likely you will never get consistent backup (remember the datastore changes all the time). The consistent copy should be "atomic", for example from a snapshot of original datastore. But anyway you will get a month old copy... Forget.

I suggest you set up another PBS in a remove location and use sync feature.

Well, that large repo only syncs once a month, as a disaster recovery if my house is burning up or similar.

The other ones aren't that heavy, maybe 1 TB at the most in total size, and since it only sync changes it only takes a couple of hours during the night to get even.

 

[michabbs]

Well, that large repo only syncs once a month, as a disaster recovery if my house is burning up or similar.

But after the disaster you will have to download back the whole datastore before you begin with recovery. If you had another PBS server remotely, then you could (1) begin immediately and (2) download only what you need.