Manual instrumentation is usually not the right way to instrument larger applications. However, to inspect smaller portions of an application in an ad-hoc manner or in cases aspect-weaving is not possible, manual instrumentation can be a viable option.
To use Kieker with an Java application, you have to add the dependency to your build system, e.g., (in gradle)
compile 'net.kieker-monitoring:kieker:1.14'
See also https://mvnrepository.com/artifact/net.kieker-monitoring/kieker
Instrumentation requires three key elements:
- A MonitoringController
- Data collection
- Logging of the data
The MonitoringController provides basic facilities for monitoring
and logging, including a source for timestamps. It can be obtained in
any class by
private static final IMonitoringController MONITORING_CONTROLLER =
MonitoringController.getInstance();
This returns a singleton instance of the monitoring controller.
Usually in data collection you gather all information you want to store and put that data into an instance of an event type.
final long tin = MONITORING_CONTROLLER.getTimeSource().getTime();
final String operationName = "public void exampleOp()"
final String className = this.getClass().getName();
In this example, the first line uses the time source facility of the
MonitoringController to gain the current time. Subsequent, two
strings are defined which represent the name of the operation (method)
and the name of the class the method resides in. Finally, the data must
be packed into a event type.
final BeforeOperationEvent e =
new BeforeOperationEvent(tin, className, operationName);
The last step uses the logging facility of the MonitoringController.
MONITORING_CONTROLLER.newMonitoringRecord(e);