I’m owning 2 identical TrueNAS physical servers, each 10 hard drives.
On the first one, i’m exposing NFS and SMB shares to my users.
I would like to use the second one as a backup, in case something goes wrong with the first one ( hack, fire, flood, etc…).
They are in different locations.
What would be, according to you, the best set-up for :
Thanks for your advices
A single RAIDZ2 vdev.
ZFS replication.
Start here and then replication as next step.
https://www.truenas.com/docs/scale/25.04/gettingstarted/configure/setupbackupscale/
Thanks Dan, so i have to install both TrueNAS, both RAIDZ2 pool. Can you explain why you’re picking the RAIDZ2 choice over the others ?
If i use zfs replication, do i have to take care of shares / users / groups / rights on the backup server ?
thank you so much, i will take a look a this link
The other realistic options would be RAIDZ1 and RAIDZ3; striped mirrors would lose half of your capacity to redundancy. If you were needing high-performance block storage, striped mirrors would be your best bet, but for routine file-sharing, parity RAID (i.e., RAIDZn) is generally the way to go. Two disks’ worth of redundancy is generally the “sweet spot” with modern-capacity drives.
No.
Permissions will come across as can dataset properties however whether your IDs match across both servers is another question. Are you creating local users or binding to an Active Directory service?
Shares are not auto created so if needed they would also need attention on the backup server but it depends on how you wish to use this backup if shares are needed at all.
So far, i dont have an AD. MY users are are on external LDAP server, but as i understood, this doesnt work. I managed to sync my users, but not their passwords. Therefore i will create them only locally in a first time. Later i will use a FreeIPA server for users
Is the idea of the backup server to also act as a failover in the event of a disaster?
If so you’ll want your local users and groups to match and have your shares created and ready to roll.
I’d also suggest you spend a bit of time becoming familiar with a DR scenario. Things like how to disable read-only on the backup system if needed and how to make sure you don’t bring your primary back online weeks later and it wipe out all the changes with replication.
I think you made a point. I didn’t think so far yet, but i have to agree that acting as a failover, would be awesome. Nowadays, the waiting time to get new equipements count in monthes… So maybe a failover is the best idea !
a little bonus optional tip, with an eye on mantainance: give a look into zfs bookmarks, they are not integrated in the GUI (feature request here 
), but they can save you from a replication from scratch that can be a pain when system in different networks are involved
