NPE in drools rules resulting into CYCLIC_DEPENDENCY and OOM - out-of-memory

Question

We have set up a basic Drools framework with Guvnor UI. Rules are written by semi-technical guys and have scope of errors (scope is always there irrespective of who writes it :))
Yesterday, someone wrote a rule which started throwing NullPoiterException but Drool appears to wrap this exception into CYCLIC_DEPENDENCY_ERROR and showing the same.
Also, we ended up with OOM java.lang.OutOfMemoryError: Java heap space immediately(3-4 minutes) after this issue.
Stacktrace for details:
Feb 2013 08:06:55,447 [ERROR] (SWF Activity DroolsActivityTaskList-2.2 229) com.amazon.mobius.csi.workers.droolsengine.DroolsEngineActivitiesImpl:runRules:220 [CYCLIC_DEPENDENCY_ERROR]: Error
while executing Rules, Please validate all rules against any cyclic dependency
Exception executing consequence for rule "final NON EC" in com.amazon.mobius.csi.workers.droolsengine: java.lang.NullPointerException
at org.drools.runtime.rule.impl.DefaultConsequenceExceptionHandler.handleException(DefaultConsequenceExceptionHandler.java:39)
at org.drools.common.DefaultAgenda.fireActivation(DefaultAgenda.java:1101)
at org.drools.common.DefaultAgenda.fireNextItem(DefaultAgenda.java:1029)
at org.drools.common.DefaultAgenda.fireAllRules(DefaultAgenda.java:1251)
at org.drools.common.AbstractWorkingMemory.fireAllRules(AbstractWorkingMemory.java:708)
at org.drools.common.AbstractWorkingMemory.fireAllRules(AbstractWorkingMemory.java:672)
at org.drools.impl.StatefulKnowledgeSessionImpl.fireAllRules(StatefulKnowledgeSessionImpl.java:218)
at org.drools.impl.StatelessKnowledgeSessionImpl.execute(StatelessKnowledgeSessionImpl.java:295)
at com.amazon.mobius.csi.workers.droolsengine.DroolsEngineActivitiesImpl.runRules(DroolsEngineActivitiesImpl.java:216)
at com.amazon.mobius.csi.workers.droolsengine.DroolsEngineActivitiesImpl.executeRules(DroolsEngineActivitiesImpl.java:145)
at sun.reflect.GeneratedMethodAccessor173.invoke(Unknown Source)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25)
at java.lang.reflect.Method.invoke(Method.java:597)
at com.amazonaws.services.simpleworkflow.flow.pojo.POJOActivityImplementation.execute(POJOActivityImplementation.java:63)
at com.amazonaws.services.simpleworkflow.flow.generic.ActivityImplementationBase.execute(ActivityImplementationBase.java:46)
at com.amazonaws.services.simpleworkflow.flow.worker.SynchronousActivityTaskPoller.execute(SynchronousActivityTaskPoller.java:214)
at com.amazonaws.services.simpleworkflow.flow.worker.ActivityTaskPoller$2.run(ActivityTaskPoller.java:95)
at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:895)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:918)
at java.lang.Thread.run(Thread.java:662)
Caused by: java.lang.NullPointerException
at com.amazon.mobius.csi.workers.droolsengine.Rule_final_NON_EC.defaultConsequence(Unknown Source)
at com.amazon.mobius.csi.workers.droolsengine.Rule_final_NON_ECDefaultConsequenceInvoker.evaluate(Unknown Source)
at org.drools.common.DefaultAgenda.fireActivation(DefaultAgenda.java:1091)
... 18 more
Why does Drools treat it as CYCLIC_DEPENDENCY and result into OOM? Could we have some other issue? Any help would be appreciated.
Thanks,

Answer 1

I believe this to be a cyclic dependency between rules. The RHS of one rule is causing another to activate, which in turn has a RHS which causes the first rule to activate. Therefore there is an infinite loop of rule activations.
Without seeing the rules themselves, I would recommend setting up an agenda event listener to log which rules are firing. That way you should be able to see which 2 rules are activating in this manner. An example such logging listener is here:
https://github.com/gratiartis/scattercode-drools-scenarios/blob/master/src/main/java/uk/co/scattercode/drools/util/TrackingAgendaEventListener.java