Do I have to create iSCSI sessions / connections to the head (Head B) that doesn't run the live service to cater for Cluster Failover?
Nope, that's not required. In a fail-over situation the remaining cluster head will take over 'virtual' identity of the failed head with its igroups, LUN mappings, etc.
But there are some caveats:
- NIC ports from both heads need to have access to the same subnets / VLANs
- Addresses of IP interfaces on Head A need to be manually specified in the config of Head B, so it can take over these IP addresses, should the failure occur.