From TN 109833034
This issue has been reported to Lotus Quality Engineering in the form of an enhancement request; however, there are currently no plans to address it.
Basically, scheduled agents run only on the server on which they are scheduled to run. If Server A is clustered with Server B and Server A goes down, any scheduled agents on Server A are not going to run on Server B, even though they are clustered.
One way around this is to code the agents so that they poll the other server for availability. If the server is available, they do not run; if it is not available, they do run. It is important to note that this workaround is not supported by Lotus.
An optional workaround is:
1. Create a single agent (AgentCore) with the core processing code you need and set the trigger to "Agent list selection".
2. Create a second agent (AgentTriggerPrimary) with code that calls the AgentCore agent. Schedule this agent to run on the Primary Cluster Server.
3. Create a third agent (AgentTriggerBackup) with code that checks to see if the Primary Cluster Server is available. If it is not, it calls the AgentCore agent, otherwise it exits. Schedule this agent to run on the Failover Cluster Server.