Linux Enterprise Cluster: Build a Highly Available Cluster with Commodity Hardware and Free Software

A high-availability LVS cluster has two LVS Directors: a primary and a backup. When the two Directors first boot, the primary Director owns the cluster load-balancing resource (the VIP and the LVS forwarding and scheduling rules). The secondary (backup) Director listens to the heartbeats coming from the primary Director. If the secondary Director does not hear the primary server's heartbeat, it initiates a failover and takes ownership of the cluster load-balancing resource by doing the following:

• Adding the VIP address to one of its NICs.

• Sending GARP broadcasts to let other machines on the network know that it now has the VIP.

• Creating an IPVS table to load balance incoming requests for (virtual) services.

• Possibly shutting off the power (Stonith) to the primary Director.