YARN-165. RM should point tracking URL to RM web page for app when AM fails (jlowe via bobby)

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1404211 13f79535-47bb-0310-9956-ffa450edef68

hadoop-yarn-project/CHANGES.txt

@@ -187,6 +187,9 @@ Release 0.23.5 - UNRELEASED
     YARN-139. Interrupted Exception within AsyncDispatcher leads to user
     confusion. (Vinod Kumar Vavilapalli via jlowe)
 
+    YARN-165. RM should point tracking URL to RM web page for app when AM fails
+    (jlowe via bobby)
+
 Release 0.23.4 - UNRELEASED
 
   INCOMPATIBLE CHANGES

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

@@ -531,7 +531,7 @@ public class RMAppImpl implements RMApp {
 
     RMAppAttempt attempt = new RMAppAttemptImpl(appAttemptId,
         clientTokenStr, rmContext, scheduler, masterService,
-        submissionContext, YarnConfiguration.getProxyHostAndPort(conf));
+        submissionContext, conf);
     attempts.put(appAttemptId, attempt);
     currentAttempt = attempt;
     handler.handle(

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/rmapp/attempt/RMAppAttemptImpl.java

@@ -18,6 +18,8 @@
 
 package org.apache.hadoop.yarn.server.resourcemanager.rmapp.attempt;
 
+import static org.apache.hadoop.yarn.util.StringHelper.pjoin;
+
 import java.net.URI;
 import java.net.URISyntaxException;
 import java.util.ArrayList;
@@ -33,6 +35,7 @@ import java.util.concurrent.locks.ReentrantReadWriteLock.WriteLock;
 
 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;
+import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.http.HttpConfig;
 import org.apache.hadoop.yarn.api.records.ApplicationAttemptId;
 import org.apache.hadoop.yarn.api.records.ApplicationResourceUsageReport;
@@ -45,6 +48,7 @@ import org.apache.hadoop.yarn.api.records.ContainerStatus;
 import org.apache.hadoop.yarn.api.records.NodeId;
 import org.apache.hadoop.yarn.api.records.Priority;
 import org.apache.hadoop.yarn.api.records.ResourceRequest;
+import org.apache.hadoop.yarn.conf.YarnConfiguration;
 import org.apache.hadoop.yarn.event.EventHandler;
 import org.apache.hadoop.yarn.factories.RecordFactory;
 import org.apache.hadoop.yarn.factory.providers.RecordFactoryProvider;
@@ -128,7 +132,7 @@ public class RMAppAttemptImpl implements RMAppAttempt {
   private FinalApplicationStatus finalStatus = null;
   private final StringBuilder diagnostics = new StringBuilder();
 
-  private final String proxy;
+  private Configuration conf;
 
   private static final StateMachineFactory<RMAppAttemptImpl,
                                            RMAppAttemptState,
@@ -285,9 +289,9 @@ public class RMAppAttemptImpl implements RMAppAttempt {
       String clientToken, RMContext rmContext, YarnScheduler scheduler,
       ApplicationMasterService masterService,
       ApplicationSubmissionContext submissionContext,
-      String proxy) {
+      Configuration conf) {
 
-    this.proxy = proxy;
+    this.conf = conf;
     this.applicationAttemptId = appAttemptId;
     this.rmContext = rmContext;
     this.eventHandler = rmContext.getDispatcher().getEventHandler();
@@ -397,6 +401,7 @@ public class RMAppAttemptImpl implements RMAppAttempt {
     try {
       URI trackingUri = trackingUriWithoutScheme == null ? null :
         ProxyUriUtils.getUriFromAMUrl(trackingUriWithoutScheme);
+      String proxy = YarnConfiguration.getProxyHostAndPort(conf);
       URI proxyUri = ProxyUriUtils.getUriFromAMUrl(proxy);
       URI result = ProxyUriUtils.getProxyUri(trackingUri, proxyUri,
           applicationAttemptId.getApplicationId());
@@ -977,15 +982,13 @@ public class RMAppAttemptImpl implements RMAppAttempt {
             " due to: " +  containerStatus.getDiagnostics() + "." +
             "Failing this attempt.");
 
-        /*
-         * In the case when the AM dies, the trackingUrl is left pointing to the AM's
-         * URL, which shows up in the scheduler UI as a broken link. Setting it here
-         * to empty string will prevent any link from being displayed.
-         * NOTE: don't set trackingUrl to 'null'. That will cause null-pointer exceptions
-         * in the generated proto code.
-         */
-        appAttempt.origTrackingUrl = "";
-        appAttempt.proxiedTrackingUrl = "";
+        // When the AM dies, the trackingUrl is left pointing to the AM's URL,
+        // which shows up in the scheduler UI as a broken link.  Direct the
+        // user to the app page on the RM so they can see the status and logs.
+        appAttempt.origTrackingUrl = pjoin(
+            YarnConfiguration.getRMWebAppHostAndPort(appAttempt.conf),
+            "cluster", "app", appAttempt.getAppAttemptId().getApplicationId());
+        appAttempt.proxiedTrackingUrl = appAttempt.origTrackingUrl;
 
         new FinalTransition(RMAppAttemptState.FAILED).transition(
             appAttempt, containerFinishedEvent);

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/rmapp/attempt/TestRMAppAttemptTransitions.java

@@ -17,6 +17,7 @@
 */
 package org.apache.hadoop.yarn.server.resourcemanager.rmapp.attempt;
 
+import static org.apache.hadoop.yarn.util.StringHelper.pjoin;
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertFalse;
 import static org.junit.Assert.assertNotNull;
@@ -43,6 +44,7 @@ import org.apache.hadoop.yarn.api.records.ContainerState;
 import org.apache.hadoop.yarn.api.records.ContainerStatus;
 import org.apache.hadoop.yarn.api.records.FinalApplicationStatus;
 import org.apache.hadoop.yarn.api.records.Resource;
+import org.apache.hadoop.yarn.conf.YarnConfiguration;
 import org.apache.hadoop.yarn.event.AsyncDispatcher;
 import org.apache.hadoop.yarn.event.EventHandler;
 import org.apache.hadoop.yarn.server.resourcemanager.ApplicationMasterService;
@@ -85,6 +87,8 @@ public class TestRMAppAttemptTransitions {
       LogFactory.getLog(TestRMAppAttemptTransitions.class);
   
   private static final String EMPTY_DIAGNOSTICS = "";
+  private static final String RM_WEBAPP_ADDR =
+      YarnConfiguration.getRMWebAppHostAndPort(new Configuration());
   
   private RMContext rmContext;
   private YarnScheduler scheduler;
@@ -203,7 +207,7 @@ public class TestRMAppAttemptTransitions {
     application = mock(RMApp.class);
     applicationAttempt = 
         new RMAppAttemptImpl(applicationAttemptId, null, rmContext, scheduler, 
-            masterService, submissionContext, null);
+            masterService, submissionContext, new Configuration());
     when(application.getCurrentAppAttempt()).thenReturn(applicationAttempt);
     when(application.getApplicationId()).thenReturn(applicationId);
     
@@ -216,6 +220,11 @@ public class TestRMAppAttemptTransitions {
   }
   
 
+  private String getProxyUrl(RMAppAttempt appAttempt) {
+    return pjoin(RM_WEBAPP_ADDR, "proxy",
+        appAttempt.getAppAttemptId().getApplicationId(), "");
+  }
+
   /**
    * {@link RMAppAttemptState#NEW}
    */
@@ -373,8 +382,8 @@ public class TestRMAppAttemptTransitions {
     assertEquals(host, applicationAttempt.getHost());
     assertEquals(rpcPort, applicationAttempt.getRpcPort());
     assertEquals(trackingUrl, applicationAttempt.getOriginalTrackingUrl());
-    assertEquals("null/proxy/"+applicationAttempt.getAppAttemptId().
-        getApplicationId()+"/", applicationAttempt.getTrackingUrl());
+    assertEquals(getProxyUrl(applicationAttempt),
+        applicationAttempt.getTrackingUrl());
     
     // TODO - need to add more checks relevant to this state
   }
@@ -390,8 +399,8 @@ public class TestRMAppAttemptTransitions {
         applicationAttempt.getAppAttemptState());
     assertEquals(diagnostics, applicationAttempt.getDiagnostics());
     assertEquals(trackingUrl, applicationAttempt.getOriginalTrackingUrl());
-    assertEquals("null/proxy/"+applicationAttempt.getAppAttemptId().
-        getApplicationId()+"/", applicationAttempt.getTrackingUrl());
+    assertEquals(getProxyUrl(applicationAttempt),
+        applicationAttempt.getTrackingUrl());
     assertEquals(container, applicationAttempt.getMasterContainer());
     assertEquals(finalStatus, applicationAttempt.getFinalApplicationStatus());
   }
@@ -408,8 +417,8 @@ public class TestRMAppAttemptTransitions {
         applicationAttempt.getAppAttemptState());
     assertEquals(diagnostics, applicationAttempt.getDiagnostics());
     assertEquals(trackingUrl, applicationAttempt.getOriginalTrackingUrl());
-    assertEquals("null/proxy/"+applicationAttempt.getAppAttemptId().
-        getApplicationId()+"/", applicationAttempt.getTrackingUrl());
+    assertEquals(getProxyUrl(applicationAttempt),
+        applicationAttempt.getTrackingUrl());
     assertEquals(finishedContainerCount, applicationAttempt
         .getJustFinishedContainers().size());
     assertEquals(container, applicationAttempt.getMasterContainer());
@@ -596,7 +605,30 @@ public class TestRMAppAttemptTransitions {
             diagnostics));
     testAppAttemptFailedState(amContainer, diagnostics);
   }
-  
+
+  @Test
+  public void testRunningToFailed() {
+    Container amContainer = allocateApplicationAttempt();
+    launchApplicationAttempt(amContainer);
+    runApplicationAttempt(amContainer, "host", 8042, "oldtrackingurl");
+    String containerDiagMsg = "some error";
+    int exitCode = 123;
+    ContainerStatus cs = BuilderUtils.newContainerStatus(amContainer.getId(),
+        ContainerState.COMPLETE, containerDiagMsg, exitCode);
+    ApplicationAttemptId appAttemptId = applicationAttempt.getAppAttemptId();
+    applicationAttempt.handle(new RMAppAttemptContainerFinishedEvent(
+        appAttemptId, cs));
+    assertEquals(RMAppAttemptState.FAILED,
+        applicationAttempt.getAppAttemptState());
+    assertEquals(0,applicationAttempt.getJustFinishedContainers().size());
+    assertEquals(amContainer, applicationAttempt.getMasterContainer());
+    assertEquals(0, applicationAttempt.getRanNodes().size());
+    String rmAppPageUrl = pjoin(RM_WEBAPP_ADDR, "cluster", "app",
+        applicationAttempt.getAppAttemptId().getApplicationId());
+    assertEquals(rmAppPageUrl, applicationAttempt.getOriginalTrackingUrl());
+    assertEquals(rmAppPageUrl, applicationAttempt.getTrackingUrl());
+  }
+
   @Test 
   public void testUnregisterToKilledFinishing() {
     Container amContainer = allocateApplicationAttempt();