Failover Membership

The failover membership layer monitors and detects a node failure in real time.

It provides a fast and reliable election mechanism to dynamically failover the public IP address from the node that just failed to an active partner node.

The appliances in the failover configuration are organized in a ring architecture, with two members, or partners, in the ring. Each node monitors its successor through a TCP connection. A small network packet is sent at regular intervals to the successor node.

Failure to receive consecutive heartbeat packets triggers a recovery operation. A recovery operation consists of moving the public IP address to another node. After a node failure detection, a new active node is elected and it assumes the public IP address. In a normal situation, the total time for a failure recovery is 3 seconds. The public IP address is added to a network interface chosen by the user when configuring the HA feature. The HA feature also uses a network interface to monitor a partner node and to carry data replication traffic. The interface is also chosen by the user among those that already have an IP address assigned.

Note: The user is prompted to choose only if there is more than one choice.

In order to properly monitor that the public IP address is accessible, both interfaces need to be the same.