Check the event logs on both nopde - System, Application and Security.
Check the cluster.log file on the failed node immediately after the failed start.
Verify that the account used to start the cluster service is a domain user account that is in the Local Administratoras group of each node.
Try a reboot of the failed node before failing over to it.
RegardsRudy KomacsarSenior Database Administrator"Ave Caesar! - Morituri te salutamus."