[MAPREDUCE-5837] MRAppMaster fails when checking on uber mode

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Fix Version/s: 2.5.0
• Target Version/s: 2.5.0
• Hadoop Flags: Reviewed
• Assignee: Haohui Mai
• Reporter: Haohui Mai
• Created: 15/Apr/14 20:24
• Updated: 15/Aug/14 05:47
• Resolved: 24/Apr/14 22:56

Description

When the MRAppMaster determines whether the job should run in the uber mode, it call Class.forName() to check whether the class is derived from ChainMapper:

 try {
      String mapClassName = conf.get(MRJobConfig.MAP_CLASS_ATTR);
      if (mapClassName != null) {
        Class<?> mapClass = Class.forName(mapClassName);
        if (ChainMapper.class.isAssignableFrom(mapClass))
          isChainJob = true;
      }
    } catch (ClassNotFoundException cnfe) {
      // don't care; assume it's not derived from ChainMapper
    }

The problem here is that Class.forName() can also throw NoClassDefError. It happens when the additional dependent jar is unavailable to the MRAppMaster. For example, the MRAppMaster complains about a MR job on Scala:

2014-04-15 11:52:55,877 FATAL [main] org.apache.hadoop.mapreduce.v2.app.MRAppMaster: Error starting MRAppMaster
java.lang.NoClassDefFoundError: scala/Function1
        at java.lang.Class.forName0(Native Method)
        at java.lang.Class.forName(Class.java:190)
        at org.apache.hadoop.mapreduce.v2.app.job.impl.JobImpl.isChainJob(JobImpl.java:1282)
        at org.apache.hadoop.mapreduce.v2.app.job.impl.JobImpl.makeUberDecision(JobImpl.java:1224)
        at org.apache.hadoop.mapreduce.v2.app.job.impl.JobImpl.access$3700(JobImpl.java:136)
        at org.apache.hadoop.mapreduce.v2.app.job.impl.JobImpl$InitTransition.transition(JobImpl.java:1425)
        at org.apache.hadoop.mapreduce.v2.app.job.impl.JobImpl$InitTransition.transition(JobImpl.java:1363)
        at org.apache.hadoop.yarn.state.StateMachineFactory$MultipleInternalArc.doTransition(StateMachineFactory.java:385)
        at org.apache.hadoop.yarn.state.StateMachineFactory.doTransition(StateMachineFactory.java:302)
        at org.apache.hadoop.yarn.state.StateMachineFactory.access$300(StateMachineFactory.java:46)
        at org.apache.hadoop.yarn.state.StateMachineFactory$InternalStateMachine.doTransition(StateMachineFactory.java:448)
        at org.apache.hadoop.mapreduce.v2.app.job.impl.JobImpl.handle(JobImpl.java:976)
        at org.apache.hadoop.mapreduce.v2.app.job.impl.JobImpl.handle(JobImpl.java:135)
        at org.apache.hadoop.mapreduce.v2.app.MRAppMaster$JobEventDispatcher.handle(MRAppMaster.java:1263)
        at org.apache.hadoop.mapreduce.v2.app.MRAppMaster.serviceStart(MRAppMaster.java:1063)
        at org.apache.hadoop.service.AbstractService.start(AbstractService.java:193)
        at org.apache.hadoop.mapreduce.v2.app.MRAppMaster$1.run(MRAppMaster.java:1480)
        at java.security.AccessController.doPrivileged(Native Method)
        at javax.security.auth.Subject.doAs(Subject.java:415)
        at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1606)
        at org.apache.hadoop.mapreduce.v2.app.MRAppMaster.initAndStartAppMaster(MRAppMaster.java:1476)
        at org.apache.hadoop.mapreduce.v2.app.MRAppMaster.main(MRAppMaster.java:1409)
Caused by: java.lang.ClassNotFoundException: scala.Function1
        at java.net.URLClassLoader$1.run(URLClassLoader.java:366)
        at java.net.URLClassLoader$1.run(URLClassLoader.java:355)
        at java.security.AccessController.doPrivileged(Native Method)
        at java.net.URLClassLoader.findClass(URLClassLoader.java:354)
        at java.lang.ClassLoader.loadClass(ClassLoader.java:425)
        at sun.misc.Launcher$AppClassLoader.loadClass(Launcher.java:308)
        at java.lang.ClassLoader.loadClass(ClassLoader.java:358)
        ... 22 more

The proposed fix is to catch NoClassDefError at the corresponding places.