A failover occurs when cluster resource services responds to a failure of a primary node by switching the access point for cluster resources normally accessed from that node to the first available backup node. For each CRG in which the failing node is the primary node, the access point to the CRG resources is switched to a backup node according to the recovery domain. If the first backup node is not available, the next backup node in the recovery domain is used.
When multiple CRGs are involved in a failover, device CRGs are processed first, the data CRGs second, followed by application CRGs.
If the cause of the failover is resolved, the failover can be cancelled. The CRG message queue provides the mechanism to cancel a failover.