EVS failover

At the user's discretion, the Application Manager can perform a graceful, or “soft”, failover. A soft failover means the source cluster remains up and running while the EVS is transferred to the target cluster. This can be accomplished by using the failover button in Application Manager. The steps the Application Manager takes to transfer the EVS from the source cluster to the target are the same as during a site failure failover, which is described in Site failure failover. The only difference is that in a soft failover, the source cluster resources are taken offline gracefully.

If the Application Manager detects that the source cluster has failed completely, it will display a prompt asking if a failover is desired. The time it takes for the Application Manager to realize a complete source cluster failure varies greatly. At times, the prompt can be seconds after the failure, or it might take several minutes. If you want the prompt to appear more quickly and you know that the source cluster has failed, you can decrease the amount of wait time by closing the Application Manager, re-opening it, and selecting the protected pair (if it is not automatically selected).

When a failover is initiated, the following steps occur:

1. The Application Manager waits on the target replication queue to empty.
2. The exchfailover.exe utility fails over the virtual protocols of the EVS.
3. The resources are created on the target cluster.

• The resources are created in the same order as they appeared on the source cluster.
• The resources are configured exactly as they were on the source cluster.

4. The resources are brought online.
5. The DNS failover utility is used to lock the source DNS record.

The process used to failback to the source cluster is the same as any other cluster failover method. The only difference is the EVS is now on both the source and target clusters.

To failback to the source cluster, make sure the Physical Disk resource(s) and the IP Address resource are online on the source cluster, then use the Double-Take Availability Application Manager to failback.

Note:

When you bring the source cluster online, an identical network name will still be active on the target. Because of this, when the source cluster tries to bring up the EVS on the source, the network name resource will fail and consequently the group will not come online on the source. You should allow the source cluster to finish trying to bring the resources online before using the Application Manager to failback. In a like-named environment with more than one DNS server, there may be a delay contacting the source server after failback. The DNS server used by the source cluster is updated on failback to point back to the source server. However, if the Application Manager is running on a machine that uses a different DNS server, it may not recognize the change until the next DNS zone refresh.