HDFS-3102. Add CLI tool to initialize the shared-edits dir. Contributed by Aaron T. Myers.

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1309580 13f79535-47bb-0310-9956-ffa450edef68

hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt

@@ -184,6 +184,8 @@ Release 2.0.0 - UNRELEASED
 
     HDFS-3000. Add a public API for setting quotas. (atm)
 
+    HDFS-3102. Add CLI tool to initialize the shared-edits dir. (atm)
+
   IMPROVEMENTS
 
     HDFS-2018. Move all journal stream management code into one place.

hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/common/HdfsServerConstants.java

@@ -53,7 +53,8 @@ public final class HdfsServerConstants {
     ROLLBACK("-rollback"),
     FINALIZE("-finalize"),
     IMPORT  ("-importCheckpoint"),
-    BOOTSTRAPSTANDBY("-bootstrapStandby");
+    BOOTSTRAPSTANDBY("-bootstrapStandby"),
+    INITIALIZESHAREDEDITS("-initializeSharedEdits");
     
     private String name = null;
     

hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/FSNamesystem.java

@@ -164,6 +164,7 @@ import org.apache.hadoop.hdfs.server.namenode.ha.HAState;
 import org.apache.hadoop.hdfs.server.namenode.ha.StandbyCheckpointer;
 import org.apache.hadoop.hdfs.server.namenode.ha.StandbyState;
 import org.apache.hadoop.hdfs.server.namenode.metrics.FSNamesystemMBean;
+import org.apache.hadoop.hdfs.server.namenode.metrics.NameNodeMetrics;
 import org.apache.hadoop.hdfs.server.protocol.DatanodeCommand;
 import org.apache.hadoop.hdfs.server.protocol.DatanodeRegistration;
 import org.apache.hadoop.hdfs.server.protocol.HeartbeatResponse;
@@ -346,10 +347,27 @@ public class FSNamesystem implements Namesystem, FSClusterStats,
    * @throws IOException if loading fails
    */
   public static FSNamesystem loadFromDisk(Configuration conf)
-    throws IOException {
+      throws IOException {
     Collection<URI> namespaceDirs = FSNamesystem.getNamespaceDirs(conf);
     List<URI> namespaceEditsDirs = 
       FSNamesystem.getNamespaceEditsDirs(conf);
+    return loadFromDisk(conf, namespaceDirs, namespaceEditsDirs);
+  }
+
+  /**
+   * Instantiates an FSNamesystem loaded from the image and edits
+   * directories passed.
+   * 
+   * @param conf the Configuration which specifies the storage directories
+   *             from which to load
+   * @param namespaceDirs directories to load the fsimages
+   * @param namespaceEditsDirs directories to load the edits from
+   * @return an FSNamesystem which contains the loaded namespace
+   * @throws IOException if loading fails
+   */
+  public static FSNamesystem loadFromDisk(Configuration conf,
+      Collection<URI> namespaceDirs, List<URI> namespaceEditsDirs)
+      throws IOException {
 
     if (namespaceDirs.size() == 1) {
       LOG.warn("Only one " + DFS_NAMENODE_NAME_DIR_KEY
@@ -370,8 +388,10 @@ public class FSNamesystem implements Namesystem, FSClusterStats,
       HAUtil.isHAEnabled(conf, nameserviceId));
     long timeTakenToLoadFSImage = now() - loadStart;
     LOG.info("Finished loading FSImage in " + timeTakenToLoadFSImage + " msecs");
-    NameNode.getNameNodeMetrics().setFsImageLoadTime(
+    NameNodeMetrics nnMetrics = NameNode.getNameNodeMetrics();
-                              (int) timeTakenToLoadFSImage);
+    if (nnMetrics != null) {
+      nnMetrics.setFsImageLoadTime((int) timeTakenToLoadFSImage);
+    }
     return namesystem;
   }
 

hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/NameNode.java

@@ -58,6 +58,7 @@ import org.apache.hadoop.hdfs.server.protocol.JournalProtocol;
 import org.apache.hadoop.hdfs.server.protocol.NamenodeProtocol;
 import org.apache.hadoop.hdfs.server.protocol.NamenodeProtocols;
 import org.apache.hadoop.hdfs.server.protocol.NamenodeRegistration;
+import org.apache.hadoop.hdfs.server.protocol.NamespaceInfo;
 import org.apache.hadoop.ipc.StandbyException;
 import org.apache.hadoop.metrics2.lib.DefaultMetricsSystem;
 import org.apache.hadoop.net.NetUtils;
@@ -70,6 +71,9 @@ import org.apache.hadoop.tools.GetUserMappingsProtocol;
 import org.apache.hadoop.util.ServicePlugin;
 import org.apache.hadoop.util.StringUtils;
 
+import com.google.common.annotations.VisibleForTesting;
+import com.google.common.collect.Lists;
+
 /**********************************************************
  * NameNode serves as both directory namespace manager and
  * "inode table" for the Hadoop DFS.  There is a single NameNode
@@ -729,6 +733,67 @@ public class NameNode {
                 + "to true in order to format this filesystem");
     }
   }
+  
+  @VisibleForTesting
+  public static boolean initializeSharedEdits(Configuration conf) {
+    return initializeSharedEdits(conf, true);
+  }
+  
+  @VisibleForTesting
+  public static boolean initializeSharedEdits(Configuration conf,
+      boolean force) {
+    return initializeSharedEdits(conf, force, false);
+  }
+  
+  /**
+   * Format a new shared edits dir.
+   * 
+   * @param conf configuration
+   * @param force format regardless of whether or not the shared edits dir exists
+   * @param interactive prompt the user when a dir exists
+   * @return true if the command aborts, false otherwise
+   */
+  private static boolean initializeSharedEdits(Configuration conf,
+      boolean force, boolean interactive) {
+    NNStorage existingStorage = null;
+    try {
+      FSNamesystem fsns = FSNamesystem.loadFromDisk(conf,
+          FSNamesystem.getNamespaceDirs(conf),
+          FSNamesystem.getNamespaceEditsDirs(conf, false));
+      
+      existingStorage = fsns.getFSImage().getStorage();
+      
+      Collection<URI> sharedEditsDirs = FSNamesystem.getSharedEditsDirs(conf);
+      if (!confirmFormat(sharedEditsDirs, force, interactive)) {
+        return true; // aborted
+      }
+      NNStorage newSharedStorage = new NNStorage(conf,
+          Lists.<URI>newArrayList(),
+          sharedEditsDirs);
+      
+      newSharedStorage.format(new NamespaceInfo(
+          existingStorage.getNamespaceID(),
+          existingStorage.getClusterID(),
+          existingStorage.getBlockPoolID(),
+          existingStorage.getCTime(),
+          existingStorage.getDistributedUpgradeVersion()));
+    } catch (Exception e) {
+      LOG.error("Could not format shared edits dir", e);
+      return true; // aborted
+    } finally {
+      // Have to unlock storage explicitly for the case when we're running in a
+      // unit test, which runs in the same JVM as NNs.
+      if (existingStorage != null) {
+        try {
+          existingStorage.unlockAll();
+        } catch (IOException ioe) {
+          LOG.warn("Could not unlock storage directories", ioe);
+          return true; // aborted
+        }
+      }
+    }
+    return false; // did not abort
+  }
 
   private static boolean finalize(Configuration conf,
                                boolean isConfirmationNeeded
@@ -763,7 +828,8 @@ public class NameNode {
       StartupOption.ROLLBACK.getName() + "] | [" +
       StartupOption.FINALIZE.getName() + "] | [" +
       StartupOption.IMPORT.getName() + "] | [" +
-      StartupOption.BOOTSTRAPSTANDBY.getName() + "]");
+      StartupOption.BOOTSTRAPSTANDBY.getName() + "] | [" +
+      StartupOption.INITIALIZESHAREDEDITS.getName() + "]");
   }
 
   private static StartupOption parseArguments(String args[]) {
@@ -804,6 +870,9 @@ public class NameNode {
       } else if (StartupOption.BOOTSTRAPSTANDBY.getName().equalsIgnoreCase(cmd)) {
         startOpt = StartupOption.BOOTSTRAPSTANDBY;
         return startOpt;
+      } else if (StartupOption.INITIALIZESHAREDEDITS.getName().equalsIgnoreCase(cmd)) {
+        startOpt = StartupOption.INITIALIZESHAREDEDITS;
+        return startOpt;
       } else {
         return null;
       }
@@ -868,29 +937,39 @@ public class NameNode {
     }
 
     switch (startOpt) {
-      case FORMAT:
+      case FORMAT: {
         boolean aborted = format(conf, false);
         System.exit(aborted ? 1 : 0);
         return null; // avoid javac warning
-      case GENCLUSTERID:
+      }
+      case GENCLUSTERID: {
         System.err.println("Generating new cluster id:");
         System.out.println(NNStorage.newClusterID());
         System.exit(0);
         return null;
-      case FINALIZE:
+      }
-        aborted = finalize(conf, true);
+      case FINALIZE: {
+        boolean aborted = finalize(conf, true);
         System.exit(aborted ? 1 : 0);
         return null; // avoid javac warning
-      case BOOTSTRAPSTANDBY:
+      }
+      case BOOTSTRAPSTANDBY: {
         String toolArgs[] = Arrays.copyOfRange(argv, 1, argv.length);
         int rc = BootstrapStandby.run(toolArgs, conf);
         System.exit(rc);
         return null; // avoid warning
+      }
+      case INITIALIZESHAREDEDITS: {
+        boolean aborted = initializeSharedEdits(conf, false, true);
+        System.exit(aborted ? 1 : 0);
+        return null; // avoid warning
+      }
       case BACKUP:
-      case CHECKPOINT:
+      case CHECKPOINT: {
         NamenodeRole role = startOpt.toNodeRole();
         DefaultMetricsSystem.initialize(role.toString().replace(" ", ""));
         return new BackupNode(conf, role);
+      }
       default:
         DefaultMetricsSystem.initialize("NameNode");
         return new NameNode(conf);

hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/namenode/ha/TestBootstrapStandby.java

@@ -36,7 +36,6 @@ import org.apache.hadoop.hdfs.server.namenode.NameNode;
 import org.apache.hadoop.hdfs.server.namenode.NameNodeAdapter;
 import org.apache.hadoop.test.GenericTestUtils;
 import org.apache.hadoop.test.GenericTestUtils.LogCapturer;
-import org.codehaus.jackson.sym.NameN;
 import org.junit.After;
 import org.junit.Before;
 import org.junit.Test;

hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/namenode/ha/TestInitializeSharedEdits.java

@@ -0,0 +1,120 @@
+/**
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.hadoop.hdfs.server.namenode.ha;
+
+import java.io.File;
+import java.io.IOException;
+import org.apache.commons.logging.Log;
+import org.apache.commons.logging.LogFactory;
+import org.apache.hadoop.conf.Configuration;
+import org.apache.hadoop.fs.FileSystem;
+import org.apache.hadoop.fs.FileUtil;
+import org.apache.hadoop.fs.Path;
+import org.apache.hadoop.hdfs.MiniDFSCluster;
+import org.apache.hadoop.hdfs.MiniDFSNNTopology;
+import org.apache.hadoop.hdfs.server.namenode.NameNode;
+import org.apache.hadoop.test.GenericTestUtils;
+import org.junit.After;
+import org.junit.Before;
+import org.junit.Test;
+
+import static org.junit.Assert.*;
+
+public class TestInitializeSharedEdits {
+
+  private static final Log LOG = LogFactory.getLog(TestInitializeSharedEdits.class);
+  
+  private static final Path TEST_PATH = new Path("/test");
+  private Configuration conf;
+  private MiniDFSCluster cluster;
+  
+  @Before
+  public void setupCluster() throws IOException {
+    conf = new Configuration();
+
+    MiniDFSNNTopology topology = MiniDFSNNTopology.simpleHATopology();
+    
+    cluster = new MiniDFSCluster.Builder(conf)
+      .nnTopology(topology)
+      .numDataNodes(0)
+      .build();
+    cluster.waitActive();
+  
+    cluster.shutdownNameNode(0);
+    cluster.shutdownNameNode(1);
+    File sharedEditsDir = new File(cluster.getSharedEditsDir(0, 1));
+    assertTrue(FileUtil.fullyDelete(sharedEditsDir));
+  }
+  
+  @After
+  public void shutdownCluster() throws IOException {
+    if (cluster != null) {
+      cluster.shutdown();
+    }
+  }
+  
+  @Test
+  public void testInitializeSharedEdits() throws Exception {
+    // Make sure we can't currently start either NN.
+    try {
+      cluster.restartNameNode(0, false);
+      fail("Should not have been able to start NN1 without shared dir");
+    } catch (IOException ioe) {
+      LOG.info("Got expected exception", ioe);
+      GenericTestUtils.assertExceptionContains(
+          "Cannot start an HA namenode with name dirs that need recovery", ioe);
+    }
+    try {
+      cluster.restartNameNode(1, false);
+      fail("Should not have been able to start NN2 without shared dir");
+    } catch (IOException ioe) {
+      LOG.info("Got expected exception", ioe);
+      GenericTestUtils.assertExceptionContains(
+          "Cannot start an HA namenode with name dirs that need recovery", ioe);
+    }
+    
+    // Initialize the shared edits dir.
+    assertFalse(NameNode.initializeSharedEdits(conf));
+    
+    // Now should be able to start both NNs. Pass "false" here so that we don't
+    // try to waitActive on all NNs, since the second NN doesn't exist yet.
+    cluster.restartNameNode(0, false);
+    cluster.restartNameNode(1, true);
+    
+    // Make sure HA is working.
+    cluster.transitionToActive(0);
+    FileSystem fs = null;
+    try {
+      fs = HATestUtil.configureFailoverFs(cluster, conf);
+      assertTrue(fs.mkdirs(TEST_PATH));
+      cluster.transitionToStandby(0);
+      cluster.transitionToActive(1);
+      assertTrue(fs.isDirectory(TEST_PATH));
+    } finally {
+      if (fs != null) {
+        fs.close();
+      }
+    }
+  }
+  
+  @Test
+  public void testDontOverWriteExistingDir() {
+    assertFalse(NameNode.initializeSharedEdits(conf, false));
+    assertTrue(NameNode.initializeSharedEdits(conf, false));
+  }
+}