HI,This is Ryotaro
I want to ask about Memory Sensor(Collection,Map).
The other day (July 27) JavaAgent for function verification,
After changing the default setting in Tire on the DynatraceServer side as follows,
When Java Agent is restarted to activate this setting, an it took 30 to 40 minutes (usually several minutes) at WAS startup time.
<Setting change details>
■ Sensor arrangement
Enable EJB Invocation
■ Configuration of sensor
JDBC Sensor Properties
Change to "Enable aggregation" → "Enable aggregation of bind values"
Map sensor properties
All existing checks of "assign class" are off
Add conditions starting with "java.lang." To "assign class" and turn on check
Check "Exclude all but the following"
Collection Sensor Properties
Set as same as "Properties of Map Sensor" above
Component：Dynatrace Java Agent
Application：IBM WebSphere Application Server7.0
We confirmed the phenomenon that the heap of the Java Agent server side WAS was used up to 1.5 GB (usually about 0.5 GB) and GC processing occurred frequently.
Would you teach the necessary resource requirements for each sensor in order to avoid the above phenomenon concerning the arrangement and configuration of sensors set this time?
Please note that memory sensors are not for production use, as they allocate a huge amount of memory. They keep track of every map/collection instantiation, which makes the JVM slow.