Using Arcserve Backup in a Cluster-aware Environment › Cluster Overview › How Failover Works

How Failover Works

Failover is the process of having cluster resources migrate (or transfer) from an unavailable node to an available node. Failover is automatically initiated when a failure is detected on one of the cluster nodes. The cluster monitors resources to determine when a failure has occurred and then takes action to recover from the failure by moving the clustered resource(s) to another node in the cluster.

In a Arcserve Backup HA cluster environment, Arcserve Backup is installed in each cluster node, but only one instance will be running. In this cluster, the active node will automatically take control of the backup resources and is referred to as the backup server. Other instances of Arcserve Backup that are hosted in a passive node are referred as the standby (or failover) server and the cluster system will only activate one of them in case of failover. If the active node fails, then all backup resources will migrate to a passive node, which then becomes the new active node. The new active node begins to function as the backup server, and continues the original backup operations and maintains all previous job scheduling and media management services.

Arcserve Backup provides the following types of failover protection:

• Planned Failovers--Planned failovers occur when it is necessary to perform maintenance on the active node within a cluster and you want Arcserve Backup to migrate the cluster resources from that active node to a passive node within the cluster. Examples for planned failovers are system maintenance, disaster recovery tests, and training. A planned failover can only be executed when no jobs are running and no other Arcserve Backup-related services (such as media operations, reporting, etc.) are occurring on both the primary and member backup servers.
• Unplanned Failovers--Unplanned failover can occur because of hardware or software failures. When the active node in a cluster fails, jobs are dispersed from the failed server and critical data (such job information) is saved into a shared disk. When failover occurs, the cluster system will move the shared disk into a passive node and activate the Arcserve Backup instance in that node. After the Arcserve Backup services are resumed in the failover server, any failed jobs from the previous server are rerun in a new active cluster node. If checkpoint information was created by the job before failover occurred, the restarted job will resume from the checkpoint.