Monitoring and Managing a TIBCO BusinessEvents Cluster : Cluster Explorer

Cluster Explorer
Active and inactive nodes are shown in Cluster Explorer for a quick view of system health.
The Cluster Explorer figure above shows the hierarchy of cluster members. Inactive agents (which could be standby agents or failed agents) are dimmed.
The structure of the cluster member hierarchy is as follows:
Site
   Cluster
      Machine (host name)
         Process (Processing Unit or Deployment Unit or JVM process ID)
            Agent (inference agent, query agent, or cache agent, dashboard agent,
                  or mm agent)
      Cache Objects
Where:
Process shows each of the JVM processes (TIBCO BusinessEvents engines) running on a machine. The label for a process that was predefined in the topology file is the process unit ID assigned in the file, concatenated with the process ID enclosed in parentheses. The label for an unpredefined process is the JVM process ID.
The Cache Objects panel shows all the objects stored in the cache, without regard to their physical location in the TIBCO BusinessEvents cluster.
Machines, TIBCO BusinessEvents engines, and agents are all members of the TIBCO BusinessEvents cluster.
Predefined and Unpredefined Members
Engines that are not defined in the site topology file are known as unpredefined engines. There are some differences between predefined and unpredefined engines.
You can "Purge Inactive" members that are unpredefined to remove them from the display. Predefined members always remain in the cluster explorer UI.
The label for a process that was predefined in the topology file is the process unit ID assigned in the file, concatenated with the process ID enclosed in parentheses. (The label for an unpredefined process is the JVM process ID.)
Note that if you start a predefined TIBCO BusinessEvents engine at the command line (outside of MM) and you use a different JMX port from the one specified in the topology file, the engine starts as an unpredefined engine.
Inactive Members
Part of cluster health is checking to see that all members are running. When a member becomes inactive, Cluster Explorer and other parts of the MM Console displays a visual indicator. Standby agents in a fault tolerant group display as inactive, as well as cluster members that have stopped operating.
How Inactive Members Display
In Cluster Explorer, the icons for inactive members display in a dimmed state.
If a machine is inactive, processes and agents on that machine are also marked as inactive. Similarly, if a process is inactive, agents running in that process are also marked as inactive.
When a cluster member is inactive, you can still view the last available data in the panel for that member, but overlaid with a gray panel with the label: "Entity Inactive".
When a cluster member is inactive, and a pane relating to that member has been promoted to the Cluster Overview panel, the pane displays in gray with a message:
The difference in the display inactive member’s promoted pane alerts you to the fact that the member is inactive.
How Inactivity is Determined
Inactivity is determined by the unsuccessful return of a health ping. Health pings are set up for machines and processes only. Process pings use JMX. Machine pings use TIBCO Hawk. If TIBCO Hawk is not available, cluster health status is determined using the health status of the processes (TIBCO BusinessEvents engines) running on each machine.
The property that controls the frequency of the health check ping is tibco.clientVar.healthCheckFreq. See MM Agent Basic Configuration Reference for details.