Hi,
I have set up a cluster of 3 ESX 5.5 hosts, connected to a SAN. Within this ESX cluster I want to run a 2-node-cluster which depends on RDMs to access LUNs of the SAN.
My idea has been, to create two datastores and distribute the cluster nodes across them: each node at a separate datastore (the datastores will contain more VMs in the end, but this is out of scope here). Should one datastore fail, I can still work with the cluster node stored at the other datastore.
To provide access to the cluster disks (LUNs), I have to create RDMs. I decided to go with physical RDMs. I wanted to create all RDMs twice: one set within datastore1 (for cluster node 1) and one set within datastore2 (for cluster node 2).
While setting this up, I was able to add the RDMs to the first cluster node without problems (using the RDMs created within datastore1). But when I tried to add the RDMs to the second cluster node (using the RDMs created within datastore2), I received an error message, stating that a locking error occured. I was able to add the RDMs of datastore1 to the second cluster node without problems and the setup works fine.
But this is not what I want: now cluster node 2 depends on datastore1 also.
Isn't it possible to have two sets of RDM files? Otherwise it would mean that I have to place all cluster nodes within the same datastore, making it an single point of failure...