The Java Exception sensor retrieves information about the exception's
Throwable class, its messages and optionally its stack trace elements. It also retrieves information about exceptions that do not get thrown but are created by
new. The transferred exception event is handled as a single event with no duration.
To edit the System Profile specific settings, open the System Profile Preferences dialog box and select Agent Group/tier with Java technology > Sensor Configuration and then select the Exceptions item in the Sensor Configuration list.
Exception aggregation is aimed to reduce the PurePath node count when many exceptions are thrown. The aggregation is done per instrumented node level. The first three exceptions are captured normally, beginning with the 4th exception, aggregation takes effect and only count occurrences of distinct class+message combinations.
You can define which classes to instrument. There are two possible scenarios:
- Instrument listed classes only.
- Instrument all classes, except for the listed ones.
Select the appropriate option button and list classes to be included/excluded from the instrumentation. You can do this manually or use the Class Browser.
Note that this configuration only applies to
Exception subclasses, not
Error subclasses. For example, a
java.lang.AssertionError always appears, even if a rule such as
java. is active. The
java.lang.NullPointerException does not appear in the same example, because it is excluded by the same rule.
The Java Exception Sensor captures data of the
Throwable class without using reflection. A delegation suppression mechanism is used for suppressing nested
In the Overhead Properties section in the Exception Sensor Properties dialog box, you can enable or disable capture of stack traces for exceptions and fine-tune the amount of data that is captured and transferred from the Agent to the Server.
|Overhead Property||Description||Overhead impact||Default value|
|Capture Stacktrace||Enables capturing of stack traces.
Capturing the stack and transferring the data to the server causes high CPU overhead on the instrumented application and consumes network bandwidth.
|Maximum Captured Stacktrace depth||Limits the number of elements per stacktrace. Only the topmost elements are transferred.
The allowed range is 0—256.
|Maximum Captured Stacktraces per second||Limits the number of stacktraces captured per second.
The allowed range is 0—1024.