Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
None
-
None
Description
Some jobs throw NullPointerExceptions in reef-tests on Mesos after the protobuf to POJO changes. On the previous Mesos tests this was not caught because a sanity check was not performed to see if jobs had timed out.
For reference, driver.stderr reads:
Apr 30, 2015 2:04:54 AM org.apache.reef.runtime.mesos.driver.REEFScheduler statusUpdate SEVERE: Task Status Update: Exception in thread "Thread-12" java.lang.NullPointerException at org.apache.reef.runtime.common.driver.evaluator.EvaluatorManagerFactory.getNewEvaluatorManager(EvaluatorMana gerFactory.java:93) at org.apache.reef.runtime.common.driver.evaluator.Evaluators.put(Evaluators.java:110) at org.apache.reef.runtime.common.driver.resourcemanager.ResourceAllocationHandler.onNext(ResourceAllocationHan dler.java:60) at org.apache.reef.runtime.common.driver.resourcemanager.ResourceAllocationHandler.onNext(ResourceAllocationHan dler.java:32) at org.apache.reef.runtime.mesos.driver.REEFEventHandlers.onResourceAllocation(REEFEventHandlers.java:59) at org.apache.reef.runtime.mesos.driver.REEFScheduler.handleNewExecutor(REEFScheduler.java:389) at org.apache.reef.runtime.mesos.driver.REEFScheduler.statusUpdate(REEFScheduler.java:199) I0430 02:04:54.248558 25813 sched.cpp:772] Aborting framework '20150429-222950-167815360-5050-20608-0091'