3 x (3 x 3TB mirrored) freenas nodes mirrored

[diversity]

Is something like this possible?

3 FreeNAS or TrueNAS builds (let's call them nodes) each node having 3 x 3 TB mirrored. And have the nodes act as mirrored amongst them self.

So if one node fails and corrupts data then the other 2 nodes will know about it and repopulate the correct data on the failing node once it is back online and reporting healty

 

[Samuel Tai]

No, FreeNAS does not have any capability for clustering multiple servers together as active mirrors of each other. The best it can do is to have periodic replication syncs from A->B->C. Then if A fails, you have to manually switch to B, and likewise to C if B fails. However, at each switch, you'll only be able to access the last replica, not the real-time status of the pool on the replica sender.

 

[diversity]

Thanks for your answer.

But what if node A is the one with corrupt data but somehow not even aware of that (not sure if that is even possible)

Would then not B and C get fed corrupt data and it's all game over?

 

[Samuel Tai]

Theoretically possible, but very unlikely, with ZFS checksumming. If corruption is detected, A will alarm on a DEGRADED pool, and replication stops.

 

[sretalla]

So if one node fails and corrupts data then the other 2 nodes will know about it and repopulate the correct data on the failing node once it is back online and reporting healty

It sounds like you may be interested in the High Availability features offered by the bigger (Enterprise) brother of FreeNAS, TrueNAS (soon to be known as TrueNAS Enterprise when FreeNAS becomes TrueNAS Core).

TrueNAS (Enterprise) is only supported on iX Sytems hardware, so you will need to look into it on their website for more details. https://www.ixsystems.com/truenas/#a-tn-info

 

[HoneyBadger]

Is something like this possible?

3 FreeNAS or TrueNAS builds (let's call them nodes) each node having 3 x 3 TB mirrored. And have the nodes act as mirrored amongst them self.

So if one node fails and corrupts data then the other 2 nodes will know about it and repopulate the correct data on the failing node once it is back online and reporting healty

Sounds like the upcoming TrueNAS SCALE will be right up your alley; it's Gluster on top of ZFS nodes.