Some advice on performing backups to some 8TB archival disks

[Nicola Worthington]

I'm running the most recent patched FreeNAS 9.3 release.

I have a pool with a few datasets inside it made up of 3x vdevs (each vdev is 7x3TB disk raidz2), with a likelihood that I will add a 4th vdev of the same layout to the pool in the future.

I also have 5x 8TB archival disks in the same chassis which I would like to use for making periodic backups to of my main pool. In the past before I started migrating my setup to FreeNAS, I have had some hacky rsync scripts that simply copy smaller portions of my data on to each of these disks, so there was wasted space at the end of each archival disk from having to split the data up logically by directory.

My question is, what's the best way to do this? Should I configure each 8TB disks for use with a Bacula Jail (I have some experience of using Bacula at work, but only using a single large volume for writing backups to, not individual disks), or should I do something like stripe all the 8TB disks together and do some sort of snapshot replication between the main pool and the backup pool (something I have no experience of).

I'd love to get some advice here.

 

[depasseg]

What is your intent? I'm not sure I understand the reason behind the hacky rsync scripts or the use of bacula.

Setting up the 8TB drives in a separate ZFS pool (RAID Z1 or Z2) and using snapshots and replication is the simplest way to get a backup of your data. Just set the snapshot at the pool dataset level (level above the "few datasets") and mark them for reecursion. If the drives are attached to the same server, set up a replication task from "PoolA" to "PoolB" and the destination host is 127.0.0.1.

 

[Glorious1]

Nicola, I routinely use the solution that depasseg suggested. The main pool replicates to the backup pool daily, both in the same machine. It was a little tricky to set up (for me) but you seem pretty savvy so might be easy for you. FYI, here is one of the snapshot tasks for a datset (Attic) on the main pool (Ark):

And the corresponding replication task to send it to the backup pool (Arkbak):

Let us know if you hit any snags.