[SOLVED] Replication not possible ? Cluster + ZFS

[alchemy24]

Hi,

I'm trying to do replication but it's not working, here's my setup.

Cluster of 3 nodes.
Local ZFS storage on each node.
Each ZFS pool has to have a different pool name, this seems to be a consequence of each node being part of the cluster.
When doing the replication it asks that the storage name (pool name) be the same on each node.

This is the log entry from the failed replication job:
2019-07-23 08:53:03 100-0: end replication job with error: command '/usr/bin/ssh -e none -o 'BatchMode=yes' -o 'HostKeyAlias=prox03' root@192.168.140.23 -- pvesr prepare-local-job 100-0 pool1:subvol-100-disk-0 --last_sync 0' failed: exit code 255

If I run the command manually, I get this:
root@prox01:~# /usr/bin/ssh -e none -o 'BatchMode=yes' -o 'HostKeyAlias=prox03' root@192.168.140.23 -- pvesr prepare-local-job 100-0 pool1:subvol-100-disk-0 --last_sync 0
storage 'pool1' is not available on node 'prox03'

Is there any way to do replication, in a cluster, using local ZFS storage ?
Technically, the process just need to accept that the storage name be different on the target.
Worst case, could we just modify the process to accept different storage name ?

Thanks

 

[aaron]

Local ZFS storage on each node.
Each ZFS pool has to have a different pool name, this seems to be a consequence of each node being part of the cluster.
When doing the replication it asks that the storage name (pool name) be the same on each node.

Why do you think this is so?

In fact the storage ID/name needs to be the same across all nodes in the cluster for replication to work.
If you create a new ZFS pool on a node take a look at the "Add Storage" checkbox right below the name field. If you created it already on another node you need to make sure to disable it.
Once you have created the ZFS pool on all nodes edit the storage properties on the datacenter level and select all nodes on which you have the pool.

 

[alchemy24]

Works brilliantly !
I have to say I find it counter-intuitive, but in th end it makes total sense and it works fine.

Here are the steps:
node 1:
create ZFS "pool1" - "Add Storage" checked ON - add available disk to pool
node 2:
create ZFS "pool1" - "Add Storage" checked OFF - add available disk to pool
node 3:
create ZFS "pool1" - "Add Storage" checked OFF - add available disk to pool
In Datacenter:
Storage, "pool1", edit, add nodes 2 and 3.

Then the replication works fine.

Thanks a lot Aaron !

 

[aaron]

Great. Please be so kind and mark the topic as solved