This behavior (Figure 1) will occur if all nodes in the cluster have not been seeded and Global Atomic Broadcast (GAB) has not been modified from the default VERITAS Cluster Server (tm) 2.0 configuration.

Figure 1
 

By default, GAB is configured to seed the cluster after all nodes in the cluster have joined the cluster. This configuration is read from gabtab.txt (Figure 2), which is  located by default in the C:\Program Files\VERITAS\comms\gab\ directory. In this case, this is a two node cluster that will not start until two nodes have joined. If both nodes go down for any reason and one node does not come online, the cluster will not start. The configured service groups will not start either.

Figure 2
 

This can be modified on demand by typing the following command (Figure 3) :

gabconfig -c -x

Note: This is not a permanent change.

Figure 3
 

The "-x" switch tells the High Availability Engine (HAD) to start the cluster on all available nodes, regardless of how many nodes are online. Once this command is issued, the HAD service should start and cluster will seed (Figure 4).

Figure 4