Tracker is a project mainly written in Java, based on the View license.
A lightweight library for Java application performance monitoring: shows application's threads and call stacks with your methods' elapsed times, various call statistics, slowest application calls, etc. etc.
AT A GLANCE This library is not a substitute for profiler. However, when it's impossible to connect with profiler (production/customer site/whatever), tracker provides a great insight on what your application is doing right now, your applicative call stack, how much time it takes, what the most called methods are, etc. etc.
Tracking data can be accessed programmatically via TrackerFactory. The library also provides a JMX which shows tracking data:
Applicative threads and call stacks example -
Tracker: Default "demo-thr" Id=12 TIMED_WAITING java.lang.Thread.sleep(Native Method) org.drift.tracker.utils.ThreadUtils.sleep(ThreadUtils.java:168) org.drift.tracker.DemoClass.midLevelCall_aroundBody2(DemoClass.java:19) org.drift.tracker.DemoClass$AjcClosure3.run(DemoClass.java:1) org.aspectj.runtime.reflect.JoinPointImpl.proceed(JoinPointImpl.java:149) org.drift.tracker.aop.MonitoringAspect.monitor(MonitoringAspect.java:36) org.drift.tracker.DemoClass.midLevelCall(DemoClass.java:19) org.drift.tracker.DemoClass.topLevelCall_aroundBody0(DemoClass.java:13) org.drift.tracker.DemoClass$AjcClosure1.run(DemoClass.java:1) org.aspectj.runtime.reflect.JoinPointImpl.proceed(JoinPointImpl.java:149) org.drift.tracker.aop.MonitoringAspect.monitor(MonitoringAspect.java:36) org.drift.tracker.DemoClass.topLevelCall(DemoClass.java:12) org.drift.tracker.DemoApp$1.run(DemoApp.java:25) java.lang.Thread.run(Thread.java:619)
Calls: Time: started at 2011-06-08 16:06:04.546, elapsed: 60360 ms Call: DemoClass.topLevelCall
Time: started at 2011-06-08 16:06:05.562, elapsed: 59359 ms
Call: DemoClass.midLevelCallCall statistics example -
Call: DemoClass.midLevelCall => invocations: N, total elapsed: ..., average elapsed: ..., max elapsed: ... Call: DemoClass.topLevelCall => invocations: M, total elapsed: ..., average elapsed: ..., max elapsed: ...
You can use jconsole (not the best job Sun ever did ...) or incorporate this lovely JMX web console written in Groovy (https://github.com/rimar/groovy-jmx-console) in your application.
HOW IT WORKS Tracker is using load-time weaving to instrument classes with methods you choose to track.
To incorporate tracker in your application:
See tiny demo application (org.drift.tracker.DemoApp) example. You can configure sleep intervals via VM args for your convenience.
You can also track methods programmatically. See the org.drift.tracker.examples.AbstractPerfMonitoringFilter example.
AOP Configure your weaving in aop.xml by adding
If you use Spring, annotate MonitoringAspect class with @Component. You will probably need to adjust the library to your Spring AOP configuration. If it's a standard proxy-based AOP, you're good to go and can even throw away the aop.xml.
Spring also provides its own instrumentation agent.
LOGGING The library is not integrated with any particular logger. Choose any logger you like and add pieces of logging, for example in org.drift.tracker.aop.MonitoringAspect:45