Adding cluster-wide storage which is not available on all nodes. Possible problems?

[mailinglists]

Hi,

in my test cluster I just added a storage (zpool) which is only available on a few nodes.
It comes up in GUI, but with nodes that do not have it, it has a question mark.
I understand it obviously will not work on those nodes, but I really do not care as do do not plan to use it on those nodes that do not have that zpool.

The question is, are there any stability related issues that could arise from having added a non functioning storage on (some) nodes?

 

[mbaldini]

In the storage GUI, you can choose on which nodes that storage is enabled

 

[LnxBil]

The question is, are there any stability related issues that could arise from having added a non functioning storage on (some) nodes?

Just asking:
You are aware of the fact that your VMs will only resist on one of the nodes and there is - unless replication is enabled - no way that the data resists on multiple nodes. ZFS is a local filesystem and not in any way a shared filesystem.

 

[mailinglists]

@baldini, somehow I missed that option, but fund it later during the weekend.

@LnxBil, I know ZFS is not shared between the nodes. It also does not support multiple concurrent mounting like clustered LVM does.
My shared network storage includes DRBD backed storages via ISCSI targets or NFS exports configured via cman and pacemaker, as well as some expensive storage options from netapp all connected with switches in MC LAG.