Merge -c r1470243 from trunk to branch-2 for YARN-594. Update test and add comments in YARN-534 (Jian He via bikas)

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/branches/branch-2@1470245 13f79535-47bb-0310-9956-ffa450edef68

hadoop-yarn-project/CHANGES.txt

@@ -207,6 +207,8 @@ Release 2.0.5-beta - UNRELEASED
     YARN-547. Fixed race conditions in public and private resource localization
     which used to cause duplicate downloads. (Omkar Vinit Joshi via vinodkv)
 
+    YARN-594. Update test and add comments in YARN-534 (Jian He via bikas)
+
 Release 2.0.4-alpha - UNRELEASED
 
   INCOMPATIBLE CHANGES

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/RMAppManager.java

@@ -340,6 +340,8 @@ public class RMAppManager implements EventHandler<RMAppManagerEvent>,
       } else {
         maxAppAttempts = individualMaxAppAttempts;
       }
+      // In work-preserve restart, if attemptCount == maxAttempts, the job still
+      // needs to be recovered because the last attempt may still be running.
       if(appState.getAttemptCount() >= maxAppAttempts) {
         LOG.info("Not recovering application " + appState.getAppId() +
             " due to recovering attempt is beyond maxAppAttempt limit");

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/TestRMRestart.java

@@ -364,7 +364,6 @@ public class TestRMRestart {
     Assert.assertNotNull(attemptState);
     Assert.assertEquals(BuilderUtils.newContainerId(attemptId1, 1), 
                         attemptState.getMasterContainer().getId());
-    rm1.stop();
 
     // start new RM   
     MockRM rm2 = new MockRM(conf, memStore);
@@ -382,7 +381,12 @@ public class TestRMRestart {
     Assert.assertNull(rm2.getRMContext().getRMApps()
         .get(app1.getApplicationId()));
 
-    // stop the RM
+    // verify that app2 is stored, app1 is removed
+    Assert.assertNotNull(rmAppState.get(app2.getApplicationId()));
+    Assert.assertNull(rmAppState.get(app1.getApplicationId()));
+
+    // stop the RM  
+    rm1.stop();
     rm2.stop();
   }
 }