MAPREDUCE-5145. Changed default max-attempts to be more than one for MR jobs inline with YARN. Contributed by Zhijie Shen.

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1477889 13f79535-47bb-0310-9956-ffa450edef68

hadoop-mapreduce-project/CHANGES.txt

@@ -215,6 +215,9 @@ Release 2.0.5-beta - UNRELEASED
     MAPREDUCE-5069. add concrete common implementations of
     CombineFileInputFormat (Sangjin Lee via bobby)
 
+    MAPREDUCE-5145. Changed default max-attempts to be more than one for MR jobs
+    inline with YARN. (Zhijie Shen via vinodkv)
+
   OPTIMIZATIONS
 
     MAPREDUCE-4974. Optimising the LineRecordReader initialize() method 

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-app/src/test/java/org/apache/hadoop/mapreduce/v2/app/TestStagingCleanup.java

@@ -90,6 +90,7 @@ import org.junit.Test;
      JobId jobid = recordFactory.newRecordInstance(JobId.class);
      jobid.setAppId(appId);
      ContainerAllocator mockAlloc = mock(ContainerAllocator.class);
+     Assert.assertTrue(MRJobConfig.DEFAULT_MR_AM_MAX_ATTEMPTS > 1);
      MRAppMaster appMaster = new TestMRApp(attemptId, mockAlloc,
          JobStateInternal.RUNNING, MRJobConfig.DEFAULT_MR_AM_MAX_ATTEMPTS);
      appMaster.init(conf);
@@ -116,8 +117,9 @@ import org.junit.Test;
      appId.setId(0);
      attemptId.setApplicationId(appId);
      ContainerAllocator mockAlloc = mock(ContainerAllocator.class);
+     Assert.assertTrue(MRJobConfig.DEFAULT_MR_AM_MAX_ATTEMPTS > 1);
      MRAppMaster appMaster = new TestMRApp(attemptId, mockAlloc,
-         JobStateInternal.REBOOT, 4);
+         JobStateInternal.REBOOT, MRJobConfig.DEFAULT_MR_AM_MAX_ATTEMPTS);
      appMaster.init(conf);
      appMaster.start();
      //shutdown the job, not the lastRetry
@@ -144,7 +146,7 @@ import org.junit.Test;
      attemptId.setApplicationId(appId);
      ContainerAllocator mockAlloc = mock(ContainerAllocator.class);
      MRAppMaster appMaster = new TestMRApp(attemptId, mockAlloc,
-         JobStateInternal.REBOOT, MRJobConfig.DEFAULT_MR_AM_MAX_ATTEMPTS);
+         JobStateInternal.REBOOT, 1); //no retry
      appMaster.init(conf);
      appMaster.start();
      //shutdown the job, is lastRetry
@@ -201,8 +203,7 @@ import org.junit.Test;
      JobId jobid = recordFactory.newRecordInstance(JobId.class);
      jobid.setAppId(appId);
      ContainerAllocator mockAlloc = mock(ContainerAllocator.class);
-     MRAppMaster appMaster = new TestMRApp(attemptId, mockAlloc,
-         MRJobConfig.DEFAULT_MR_AM_MAX_ATTEMPTS);
+     MRAppMaster appMaster = new TestMRApp(attemptId, mockAlloc, 1); //no retry
      appMaster.init(conf);
      //simulate the process being killed
      MRAppMaster.MRAppMasterShutdownHook hook = 

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/MRJobConfig.java

@@ -677,6 +677,6 @@ public interface MRJobConfig {
    */
   public static final String MR_AM_MAX_ATTEMPTS = "mapreduce.am.max-attempts";
 
-  public static final int DEFAULT_MR_AM_MAX_ATTEMPTS = 1;
+  public static final int DEFAULT_MR_AM_MAX_ATTEMPTS = 2;
   
 }

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/resources/mapred-default.xml

@@ -816,10 +816,11 @@
 
 <property>
   <name>mapreduce.am.max-attempts</name>
-  <value>1</value>
+  <value>2</value>
   <description>The maximum number of application attempts. It is a
   application-specific setting. It should not be larger than the global number
-  set by resourcemanager. Otherwise, it will be override.</description>
+  set by resourcemanager. Otherwise, it will be override. The default number is
+  set to 2, to allow at least one retry for AM.</description>
 </property>
 
 <!-- Job Notification Configuration -->