Advice on using mergerfs between two servers

[pilipit]

I have two storage servers, each with their own disks with different data.

On each server, I have set up the disks as accessible via nfs shares between each other. I have also set up mergerfs, on each, so both their local disks and the nfs mounts are mounted as one mergerfs filesystem on their own. Such that:

ServerA nfs share and local disk:
/mnt/serverA/disk1

On ServerB, this share is mounted to:
/mnt/serverA/disk1

On ServerB, local disk:
/mnt/serverB/disk1

On ServerB, mergerfs:
/mnt/serverB/disk1:/mnt/serverA/disk1 /mnt/mergerfs/fs1

Same is done on ServerA side.

My reasoning for this configuration is both servers have *arr containers handling their own content and I'm trying to avoid limiting available disk space based on what they have locally.

In addition, these separate drives are also shared to a remote media server, that mounts them as nfs shares then also has mergerfs configured to combine them as a single mount.

Is there a better approach to this kind of setup?

[trapexit]

A better setup would probably be to leverage a true distributed filesystem :)

But given that is a big lift for a small setup I'm not seeing a better alternative. If you didn't need the filesystems accessible to multiple clients then I'd suggest not using a network filesystem but SAN of some sort. iscsi for example. Though you could still perhaps set that up. Export block devices over the network to a system that mounts them, merges them, and exports them over NFS. Perf would probably be better. However, the system dynamics change a bit wrt errors.

[pilipit]

Thank you for your response.