Description:
How to reduce the high frequency of performance statistics in the BEMM (TIBCO BusinessEvents Monitoring and Management) Cluster.
Symptoms:
The BEMM Cluster has a BE performance metrics collection memory leak. The issue causes an increase in the memory consumption for BEMM, which can fluctuate. Collect information about the cluster which is being monitored using BEMM. If possible, provide a heap dump or "jmap -histo:live <pid>" output for further analysis.


Cause:
This is caused due to a high frequency collection of the performance statistics. The properties below can be added and adjusted to bring this issue under control. It is not possible to turn off the statistics completely.

How to reduce the high frequency of performance statistics in the BEMM (TIBCO BusinessEvents Monitoring and Management) Cluster.

Add the following properties to the MM.CDD file.

BEMM Node: Reduce metric generation on the node itself to once a day


        <property-group comment="" name="bemm-stats">

            <property name="be.stats.enabled" value="false"/>

            <property name="be.stats.threading.model" value="none"/>

            <property name="com.tibco.be.metric.publish.enable" value="false"/>

            <property name="com.tibco.be.metric.publish.interval" value="43200000"/>

            <property name="com.tibco.be.metric.resolution" value="21600000"/>

        </property-group>

 

Other Nodes (those actually monitored by BEMM):

You might want to set the collection interval to a less frequent 30 seconds - 1minute (Believe the defaults is 5 seconds)

        <property-group comment="" name="bemm-stats">

            <property name="com.tibco.be.metric.publish.interval" value="600000"/>

            <property name="com.tibco.be.metric.resolution" value="30000"/>

        </property-group>