hadoop-common-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From bibinchund...@apache.org
Subject hadoop git commit: YARN-8541. RM startup failure on recovery after user deletion. Contributed by Bibin A Chundatt.
Date Wed, 25 Jul 2018 10:26:41 GMT
Repository: hadoop
Updated Branches:
  refs/heads/branch-3.1.1 23624c924 -> 37f25fbd0


YARN-8541. RM startup failure on recovery after user deletion. Contributed by Bibin A Chundatt.

(cherry picked from commit 8e65057eb10d03db08781da7a5ad8855155883ed)


Project: http://git-wip-us.apache.org/repos/asf/hadoop/repo
Commit: http://git-wip-us.apache.org/repos/asf/hadoop/commit/37f25fbd
Tree: http://git-wip-us.apache.org/repos/asf/hadoop/tree/37f25fbd
Diff: http://git-wip-us.apache.org/repos/asf/hadoop/diff/37f25fbd

Branch: refs/heads/branch-3.1.1
Commit: 37f25fbd084f215da8f3cb7387a29baed33d159c
Parents: 23624c9
Author: bibinchundatt <bibinchundatt@apache.org>
Authored: Wed Jul 25 15:54:32 2018 +0530
Committer: bibinchundatt <bibinchundatt@apache.org>
Committed: Wed Jul 25 15:56:16 2018 +0530

----------------------------------------------------------------------
 .../server/resourcemanager/RMAppManager.java    | 48 ++++++++++----------
 .../placement/PlacementManager.java             |  9 ----
 .../TestWorkPreservingRMRestart.java            | 48 ++++++++++++++++++++
 3 files changed, 72 insertions(+), 33 deletions(-)
----------------------------------------------------------------------


http://git-wip-us.apache.org/repos/asf/hadoop/blob/37f25fbd/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/RMAppManager.java
----------------------------------------------------------------------
diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/RMAppManager.java
b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/RMAppManager.java
index 3e64cfc..7011aaa 100644
--- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/RMAppManager.java
+++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/RMAppManager.java
@@ -364,17 +364,9 @@ public class RMAppManager implements EventHandler<RMAppManagerEvent>,
       ApplicationSubmissionContext submissionContext, long submitTime,
       String user, boolean isRecovery, long startTime) throws YarnException {
 
-    ApplicationPlacementContext placementContext = null;
-    try {
-      placementContext = placeApplication(rmContext, submissionContext, user);
-    } catch (YarnException e) {
-      String msg =
-          "Failed to place application " + submissionContext.getApplicationId()
-              + " to queue and specified " + "queue is invalid : "
-              + submissionContext.getQueue();
-      LOG.error(msg, e);
-      throw e;
-    }
+    ApplicationPlacementContext placementContext =
+        placeApplication(rmContext.getQueuePlacementManager(),
+            submissionContext, user, isRecovery);
 
     // We only replace the queue when it's a new application
     if (!isRecovery) {
@@ -789,23 +781,31 @@ public class RMAppManager implements EventHandler<RMAppManagerEvent>,
   }
 
   @VisibleForTesting
-  ApplicationPlacementContext placeApplication(RMContext rmContext,
-      ApplicationSubmissionContext context, String user) throws YarnException {
+  ApplicationPlacementContext placeApplication(
+      PlacementManager placementManager, ApplicationSubmissionContext context,
+      String user, boolean isRecovery) throws YarnException {
     ApplicationPlacementContext placementContext = null;
-    PlacementManager placementManager = rmContext.getQueuePlacementManager();
-
     if (placementManager != null) {
-      placementContext = placementManager.placeApplication(context, user);
-    } else{
-      if ( context.getQueue() == null || context.getQueue().isEmpty()) {
-        final String msg = "Queue Placement Manager is not set. Cannot place "
-            + "application : " + context.getApplicationId() + " to queue and "
-            + "specified queue is invalid " + context.getQueue();
-        LOG.error(msg);
-        throw new YarnException(msg);
+      try {
+        placementContext = placementManager.placeApplication(context, user);
+      } catch (YarnException e) {
+        // Placement could also fail if the user doesn't exist in system
+        // skip if the user is not found during recovery.
+        if (isRecovery) {
+          LOG.warn("PlaceApplication failed,skipping on recovery of rm");
+          return placementContext;
+        }
+        throw e;
       }
     }
-
+    if (placementContext == null && (context.getQueue() == null) || context
+        .getQueue().isEmpty()) {
+      String msg = "Failed to place application " + context.getApplicationId()
+          + " to queue and specified " + "queue is invalid : " + context
+          .getQueue();
+      LOG.error(msg);
+      throw new YarnException(msg);
+    }
     return placementContext;
   }
 

http://git-wip-us.apache.org/repos/asf/hadoop/blob/37f25fbd/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/placement/PlacementManager.java
----------------------------------------------------------------------
diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/placement/PlacementManager.java
b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/placement/PlacementManager.java
index 5fa7723..74cf7ba 100644
--- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/placement/PlacementManager.java
+++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/placement/PlacementManager.java
@@ -70,15 +70,6 @@ public class PlacementManager {
         }
       }
 
-      // Failed to get where to place application
-      if (null == placement && null == asc.getQueue()) {
-        String msg = "Failed to place application " +
-            asc.getApplicationId() + " to queue and specified "
-            + "queue is invalid : " + asc.getQueue();
-        LOG.error(msg);
-        throw new YarnException(msg);
-      }
-
       return placement;
     } finally {
       readLock.unlock();

http://git-wip-us.apache.org/repos/asf/hadoop/blob/37f25fbd/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/TestWorkPreservingRMRestart.java
----------------------------------------------------------------------
diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/TestWorkPreservingRMRestart.java
b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/TestWorkPreservingRMRestart.java
index 88c19a1..a821b0a 100644
--- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/TestWorkPreservingRMRestart.java
+++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/TestWorkPreservingRMRestart.java
@@ -39,8 +39,12 @@ import org.apache.hadoop.yarn.api.records.Resource;
 import org.apache.hadoop.yarn.api.records.ResourceRequest;
 import org.apache.hadoop.yarn.api.records.YarnApplicationState;
 import org.apache.hadoop.yarn.conf.YarnConfiguration;
+import org.apache.hadoop.yarn.exceptions.YarnException;
 import org.apache.hadoop.yarn.server.api.protocolrecords.NMContainerStatus;
 import org.apache.hadoop.yarn.server.resourcemanager.TestRMRestart.TestSecurityMockRM;
+import org.apache.hadoop.yarn.server.resourcemanager.placement
+    .ApplicationPlacementContext;
+import org.apache.hadoop.yarn.server.resourcemanager.placement.PlacementManager;
 import org.apache.hadoop.yarn.server.resourcemanager.recovery.MemoryRMStateStore;
 import org.apache.hadoop.yarn.server.resourcemanager.recovery.RMStateStore.RMState;
 import org.apache.hadoop.yarn.server.resourcemanager.recovery.records.ApplicationAttemptStateData;
@@ -105,6 +109,8 @@ import static org.apache.hadoop.yarn.server.resourcemanager.scheduler
 import static org.apache.hadoop.yarn.server.resourcemanager.webapp
     .RMWebServices.DEFAULT_QUEUE;
 import static org.junit.Assert.*;
+import static org.mockito.Matchers.any;
+import static org.mockito.Mockito.doThrow;
 import static org.mockito.Mockito.mock;
 import static org.mockito.Mockito.when;
 
@@ -1555,6 +1561,48 @@ public class TestWorkPreservingRMRestart extends ParameterizedSchedulerTestBase
   }
 
   @Test(timeout = 30000)
+  public void testUnknownUserOnRecovery() throws Exception {
+
+    MockRM rm1 = new MockRM(conf);
+    rm1.start();
+    MockMemoryRMStateStore memStore =
+        (MockMemoryRMStateStore) rm1.getRMStateStore();
+    MockNM nm1 =
+        new MockNM("127.0.0.1:1234", 15120, rm1.getResourceTrackerService());
+    nm1.registerNode();
+
+    // create app and launch the UAM
+    RMApp app0 = rm1.submitApp(200, true);
+    MockAM am0 = MockRM.launchUAM(app0, rm1, nm1);
+    am0.registerAppAttempt();
+    rm1.killApp(app0.getApplicationId());
+    PlacementManager placementMgr = mock(PlacementManager.class);
+    doThrow(new YarnException("No groups for user")).when(placementMgr)
+        .placeApplication(any(ApplicationSubmissionContext.class),
+            any(String.class));
+    MockRM rm2 = new MockRM(conf, memStore) {
+      @Override
+      protected RMAppManager createRMAppManager() {
+        return new RMAppManager(this.rmContext, this.scheduler,
+            this.masterService, this.applicationACLsManager, conf) {
+          @Override
+          ApplicationPlacementContext placeApplication(
+              PlacementManager placementManager,
+              ApplicationSubmissionContext context, String user,
+              boolean isRecovery) throws YarnException {
+            return super
+                .placeApplication(placementMgr, context, user, isRecovery);
+          }
+        };
+      }
+    };
+    rm2.start();
+    RMApp recoveredApp =
+        rm2.getRMContext().getRMApps().get(app0.getApplicationId());
+    Assert.assertEquals(RMAppState.KILLED, recoveredApp.getState());
+  }
+
+  @Test(timeout = 30000)
   public void testDynamicAutoCreatedQueueRecoveryWithDefaultQueue()
       throws Exception {
     //if queue name is not specified, it should submit to 'default' queue


---------------------------------------------------------------------
To unsubscribe, e-mail: common-commits-unsubscribe@hadoop.apache.org
For additional commands, e-mail: common-commits-help@hadoop.apache.org


Mime
View raw message