YARN-3692. Allow REST API to set a user generated message when killing an application. Contributed by Rohith Sharma K S

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/src/main/java/org/apache/hadoop/mapred/ResourceMgrDelegate.java

@@ -511,4 +511,10 @@ public void signalToContainer(ContainerId containerId,
       throws YarnException, IOException {
     client.signalToContainer(containerId, command);
   }
+
+  @Override
+  public void killApplication(ApplicationId appId, String diagnostics)
+      throws YarnException, IOException {
+    client.killApplication(appId, diagnostics);
+  }
 }

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/api/protocolrecords/KillApplicationRequest.java

@@ -20,6 +20,7 @@
 
 import org.apache.hadoop.classification.InterfaceAudience.Public;
 import org.apache.hadoop.classification.InterfaceStability.Stable;
+import org.apache.hadoop.classification.InterfaceStability.Unstable;
 import org.apache.hadoop.yarn.api.ApplicationClientProtocol;
 import org.apache.hadoop.yarn.api.records.ApplicationId;
 import org.apache.hadoop.yarn.util.Records;
@@ -57,4 +58,21 @@ public static KillApplicationRequest newInstance(ApplicationId applicationId) {
   @Public
   @Stable
   public abstract void setApplicationId(ApplicationId applicationId);
+
+  /**
+   * Get the <em>diagnostics</em> to which the application is being killed.
+   * @return <em>diagnostics</em> to which the application is being killed
+   */
+  @Public
+  @Unstable
+  public abstract String getDiagnostics();
+
+  /**
+   * Set the <em>diagnostics</em> to which the application is being killed.
+   * @param diagnostics <em>diagnostics</em> to which the application is being
+   *          killed
+   */
+  @Public
+  @Unstable
+  public abstract void setDiagnostics(String diagnostics);
 }

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/proto/yarn_service_protos.proto

@@ -156,6 +156,7 @@ message FailApplicationAttemptResponseProto {
 
 message KillApplicationRequestProto {
   optional ApplicationIdProto application_id = 1;
+  optional string diagnostics = 2;
 }
 
 message KillApplicationResponseProto {

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-client/src/main/java/org/apache/hadoop/yarn/client/api/YarnClient.java

@@ -172,6 +172,20 @@ public abstract void failApplicationAttempt(
   public abstract void killApplication(ApplicationId applicationId) throws YarnException,
       IOException;
 
+  /**
+   * <p>
+   * Kill an application identified by given ID.
+   * </p>
+   * @param applicationId {@link ApplicationId} of the application that needs to
+   *          be killed
+   * @param diagnostics for killing an application.
+   * @throws YarnException in case of errors or if YARN rejects the request due
+   *           to access-control restrictions.
+   * @throws IOException
+   */
+  public abstract void killApplication(ApplicationId applicationId,
+      String diagnostics) throws YarnException, IOException;
+
   /**
    * <p>
    * Get a report of the given Application.

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-client/src/main/java/org/apache/hadoop/yarn/client/api/impl/YarnClientImpl.java

@@ -405,10 +405,21 @@ public void failApplicationAttempt(ApplicationAttemptId attemptId)
   @Override
   public void killApplication(ApplicationId applicationId)
       throws YarnException, IOException {
+    killApplication(applicationId, null);
+  }
+
+  @Override
+  public void killApplication(ApplicationId applicationId, String diagnostics)
+      throws YarnException, IOException {
+
     KillApplicationRequest request =
         Records.newRecord(KillApplicationRequest.class);
     request.setApplicationId(applicationId);
 
+    if (diagnostics != null) {
+      request.setDiagnostics(diagnostics);
+    }
+
     try {
       int pollCount = 0;
       long startTime = System.currentTimeMillis();
@@ -422,14 +433,15 @@ public void killApplication(ApplicationId applicationId)
         }
 
         long elapsedMillis = System.currentTimeMillis() - startTime;
-        if (enforceAsyncAPITimeout() &&
-            elapsedMillis >= this.asyncApiPollTimeoutMillis) {
-          throw new YarnException("Timed out while waiting for application " +
-            applicationId + " to be killed.");
+        if (enforceAsyncAPITimeout()
+            && elapsedMillis >= this.asyncApiPollTimeoutMillis) {
+          throw new YarnException("Timed out while waiting for application "
+              + applicationId + " to be killed.");
         }
 
         if (++pollCount % 10 == 0) {
-          LOG.info("Waiting for application " + applicationId + " to be killed.");
+          LOG.info(
+              "Waiting for application " + applicationId + " to be killed.");
         }
         Thread.sleep(asyncApiPollIntervalMillis);
       }

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/java/org/apache/hadoop/yarn/api/protocolrecords/impl/pb/KillApplicationRequestPBImpl.java

@@ -127,6 +127,24 @@ private ApplicationIdProto convertToProtoFormat(ApplicationId t) {
     return ((ApplicationIdPBImpl)t).getProto();
   }
 
+  @Override
+  public String getDiagnostics() {
+    KillApplicationRequestProtoOrBuilder p = viaProto ? proto : builder;
+    if (!p.hasDiagnostics()) {
+      return null;
+    }
+    return (p.getDiagnostics());
+  }
+
+  @Override
+  public void setDiagnostics(String diagnostics) {
+    maybeInitBuilder();
+    if (diagnostics == null) {
+      builder.clearDiagnostics();
+      return;
+    }
+    builder.setDiagnostics(diagnostics);
+  }
 
 
 }  

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

@@ -775,15 +775,25 @@ public KillApplicationResponse forceKillApplication(
       return KillApplicationResponse.newInstance(true);
     }
 
-    String message = "Kill application " + applicationId + " received from "
-        + callerUGI;
+    StringBuilder message = new StringBuilder();
+    message.append("Application ").append(applicationId)
+        .append(" was killed by user ").append(callerUGI.getShortUserName());
+
     InetAddress remoteAddress = Server.getRemoteIp();
     if (null != remoteAddress) {
-      message += " at " + remoteAddress.getHostAddress();
+      message.append(" at ").append(remoteAddress.getHostAddress());
     }
+
+    String diagnostics = org.apache.commons.lang.StringUtils
+        .trimToNull(request.getDiagnostics());
+    if (diagnostics != null) {
+      message.append(" with diagnostic message: ");
+      message.append(diagnostics);
+    }
+
     this.rmContext.getDispatcher().getEventHandler()
-        .handle(new RMAppKillByClientEvent(applicationId, message, callerUGI,
-            remoteAddress));
+        .handle(new RMAppKillByClientEvent(applicationId, message.toString(),
+            callerUGI, remoteAddress));
 
     // For UnmanagedAMs, return true so they don't retry
     return KillApplicationResponse.newInstance(

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

@@ -925,7 +925,7 @@ public Response updateAppState(AppState targetState,
       // allow users to kill the app
 
       if (targetState.getState().equals(YarnApplicationState.KILLED.toString())) {
-        return killApp(app, callerUGI, hsr);
+        return killApp(app, callerUGI, hsr, targetState.getDiagnostics());
       }
       throw new BadRequestException("Only '"
           + YarnApplicationState.KILLED.toString()
@@ -1141,7 +1141,8 @@ public NodeLabelsInfo getLabelsOnNode(@Context HttpServletRequest hsr,
   }
 
   protected Response killApp(RMApp app, UserGroupInformation callerUGI,
-      HttpServletRequest hsr) throws IOException, InterruptedException {
+      HttpServletRequest hsr, String diagnostic)
+      throws IOException, InterruptedException {
 
     if (app == null) {
       throw new IllegalArgumentException("app cannot be null");
@@ -1158,6 +1159,9 @@ public KillApplicationResponse run() throws IOException,
                   YarnException {
                 KillApplicationRequest req =
                     KillApplicationRequest.newInstance(appid);
+                  if (diagnostic != null) {
+                    req.setDiagnostics(diagnostic);
+                  }
                 return rm.getClientRMService().forceKillApplication(req);
               }
             });

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/webapp/dao/AppState.java

@@ -27,6 +27,7 @@
 public class AppState {
 
   String state;
+  private String diagnostics;
 
   public AppState() {
   }
@@ -43,4 +44,11 @@ public String getState() {
     return this.state;
   }
 
+  public String getDiagnostics() {
+    return diagnostics;
+  }
+
+  public void setDiagnostics(String diagnostics) {
+    this.diagnostics = diagnostics;
+  }
 }

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

@@ -506,7 +506,8 @@ public void testForceKillNonExistingApplication() throws YarnException {
   @Test
   public void testForceKillApplication() throws Exception {
     YarnConfiguration conf = new YarnConfiguration();
-    MockRM rm = new MockRM();
+    conf.setBoolean(MockRM.ENABLE_WEBAPP, true);
+    MockRM rm = new MockRM(conf);
     rm.init(conf);
     rm.start();
 
@@ -522,6 +523,8 @@ public void testForceKillApplication() throws Exception {
 
     KillApplicationRequest killRequest1 =
         KillApplicationRequest.newInstance(app1.getApplicationId());
+    String diagnostic = "message1";
+    killRequest1.setDiagnostics(diagnostic);
     KillApplicationRequest killRequest2 =
         KillApplicationRequest.newInstance(app2.getApplicationId());
 
@@ -539,6 +542,8 @@ public void testForceKillApplication() throws Exception {
         killAttemptCount > 1);
     assertEquals("Incorrect number of apps in the RM", 1,
         rmService.getApplications(getRequest).getApplicationList().size());
+    assertTrue("Diagnostic message is incorrect",
+        app1.getDiagnostics().toString().contains(diagnostic));
 
     KillApplicationResponse killResponse2 =
         rmService.forceKillApplication(killRequest2);

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

@@ -359,6 +359,7 @@ public void testSingleAppKill() throws Exception {
         { MediaType.APPLICATION_JSON, MediaType.APPLICATION_XML };
     MediaType[] contentTypes =
         { MediaType.APPLICATION_JSON_TYPE, MediaType.APPLICATION_XML_TYPE };
+    String diagnostic = "message1";
     for (String mediaType : mediaTypes) {
       for (MediaType contentType : contentTypes) {
         RMApp app = rm.submitApp(CONTAINER_MB, "", webserviceUserName);
@@ -366,6 +367,7 @@ public void testSingleAppKill() throws Exception {
 
         AppState targetState =
             new AppState(YarnApplicationState.KILLED.toString());
+        targetState.setDiagnostics(diagnostic);
 
         Object entity;
         if (contentType.equals(MediaType.APPLICATION_JSON_TYPE)) {
@@ -425,6 +427,8 @@ public void testSingleAppKill() throws Exception {
             } else {
               verifyAppStateXML(response, RMAppState.KILLED);
             }
+            assertTrue("Diagnostic message is incorrect",
+                app.getDiagnostics().toString().contains(diagnostic));
             break;
           }
         }