MAPREDUCE-6339. Job history file is not flushed correctly because

isTimerActive flag is not set true when flushTimerTask is scheduled.
Contributed by zhihai xu.

hadoop-mapreduce-project/CHANGES.txt

@@ -382,6 +382,9 @@ Release 2.7.1 - UNRELEASED
     MAPREDUCE-6334. Fetcher#copyMapOutput is leaking usedMemory upon
     IOException during InMemoryMapOutput shuffle handler (Eric Payne via jlowe)
 
+    MAPREDUCE-6339. Job history file is not flushed correctly because isTimerActive 
+    flag is not set true when flushTimerTask is scheduled. (zhihai xu via devaraj)
+
 Release 2.7.0 - 2015-04-20
 
   INCOMPATIBLE CHANGES

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-app/src/main/java/org/apache/hadoop/mapreduce/jobhistory/JobHistoryEventHandler.java

@@ -72,6 +72,7 @@ import org.codehaus.jackson.map.ObjectMapper;
 import org.codehaus.jackson.node.ArrayNode;
 import org.codehaus.jackson.node.ObjectNode;
 
+import com.google.common.annotations.VisibleForTesting;
 /**
  * The job history events get routed to this class. This class writes the Job
  * history events to the DFS directly into a staging dir and then moved to a
@@ -1259,6 +1260,7 @@ public class JobHistoryEventHandler extends AbstractService
           if (!isTimerShutDown) {
             flushTimerTask = new FlushTimerTask(this);
             flushTimer.schedule(flushTimerTask, flushTimeout);
+            isTimerActive = true;
           }
         }
       }
@@ -1378,4 +1380,9 @@ public class JobHistoryEventHandler extends AbstractService
     }
     return JobState.KILLED.toString();
   }
+
+  @VisibleForTesting
+  boolean getFlushTimerStatus() {
+    return isTimerActive;
+  }
 }

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-app/src/test/java/org/apache/hadoop/mapreduce/jobhistory/TestJobHistoryEventHandler.java

@@ -223,10 +223,12 @@ public class TestJobHistoryEventHandler {
       }
 
       handleNextNEvents(jheh, 9);
+      Assert.assertTrue(jheh.getFlushTimerStatus());
       verify(mockWriter, times(0)).flush();
 
       Thread.sleep(2 * 4 * 1000l); // 4 seconds should be enough. Just be safe.
       verify(mockWriter).flush();
+      Assert.assertFalse(jheh.getFlushTimerStatus());
     } finally {
       jheh.stop();
       verify(mockWriter).close();