The Java agent is experiencing any or all of these conditions:
1. A JVM with extremely high CPU or even crashes.
2. Detection of a large amount of Automatic Entry Points/Backends. This can be seen in the Agent log.
Example:
3. Autodetected Entrypoint/Backend methods have extremely high Responses Per Interval.
1. Automatic Entry Point/Backend Detection thread dump analysis could be very CPU expensive in some JVMs (IBM).
2. Automatic Entry Point/Backend Detection detected a class/method which has extreme number of hits (i.e. 1 million+ Responses Per Interval). Monitoring such very high traffic class/method could be very expensive.
3. Automatic Entry Point/Backend Detection keeps taking thread dump even after an Entrypoint/Backend has already been detected in the transaction.
#2 is improved in 10.6.
#3 has been fixed in 10.3 HF#18 and 10.5.
Disable Automatic Entry Point Detection
1. On the IntroscopeAgent.profile, set introscope.agent.deep.entrypoint.enabled=false
for details on this property
2. If disabled, ensure that the AutoPersist.pbd file in /hotdeploy (if any) is also removed.
This does not require a managed application restart
Disable Automatic Backend Detection
1. On the IntroscopeAgent.profile, set introscope.agent.automatic.backend.detection.enabled=false
for details on this property
2. If disabled, ensure that the AutoPersist.pbd file in /hotdeploy (if any) is also removed.
This does not require a managed application restart.
Various fixes in APM 10.6 should help improve this issue.
Also see https://www.ca.com/us/services-support/ca-support/ca-support-online/knowledge-base-articles.tec1984947.html Entry points are still reported in the investigator even though the feature is disabled.