This has been discussed a few times in the forums.
The way uroni (the urbackup dev) recommends:
He use zfs/btrfs builtin sync features.
So at the block level of the urbackup partition, only changes are copied.
And that would sync everything including the conf and whatnot
I dont think you could have the two instances running at the same time, for backup cleanup , that wouldn’t help.
That should lead to less traffic than using mutiple backup servers.
The way i do it, i use two completely different instances of urbackup.
The annoying thing about 2 different servers; is that you need to manually configure each server, add each client in each servers. The settings aren’t replicated between servers.
- If you use the same key on both server, the client will connect to both transparently (if you use different key, you need to install the key on each client).
Local servers are autodiscovered if they have an interface on the same lan than the client.
If you want a remote/internet server as redundancy, this was discussed in a recent topic :
- You also set the internet server adress (server side config) as the name of your remote/internet server.
do this in local and remote server.
The local server will autodiscover the client, the client will install it’s key. Because the key is the same, the client will be able to connect to the internet server that you configured on your local server.
Internet mode transfers only file differences (cost some cpu to compute), compress transferred data, and urbackup in general doesn’t sends 2 times the same file. Still it should use more bandwidth than a zfs/btrfs sync.
With two server you can use different retention, backup periods and whatnot, for example :
- Like one server backup Monday, Wednesday, Friday
- the other Tuesday, Thursday, Saturday each keep backup for 60 days
- An online server that backup only Sunday and keep 30 backups.