HDFS-4059. Add number of stale DataNodes to metrics. Contributed by Jing Zhao.

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1398949 13f79535-47bb-0310-9956-ffa450edef68

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

@@ -344,6 +344,8 @@ Release 2.0.3-alpha - Unreleased
     HDFS-3912. Detect and avoid stale datanodes for writes.
     (Jing Zhao via suresh)
 
+    HDFS-4059. Add number of stale DataNodes to metrics. (Jing Zhao via suresh)
+
   IMPROVEMENTS
   
     HDFS-3925. Prettify PipelineAck#toString() for printing to a log

hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/DatanodeManager.java

@@ -885,7 +885,7 @@ public class DatanodeManager {
    * @return Return the current number of stale DataNodes (detected by
    * HeartbeatManager). 
    */
-  int getNumStaleNodes() {
+  public int getNumStaleNodes() {
     return this.numStaleNodes;
   }
 

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

@@ -4678,6 +4678,13 @@ public class FSNamesystem implements Namesystem, FSClusterStats,
     return getBlockManager().getDatanodeManager().getNumDeadDataNodes();
   }
   
+  @Override // FSNamesystemMBean
+  @Metric({"StaleDataNodes", 
+    "Number of datanodes marked stale due to delayed heartbeat"})
+  public int getNumStaleDataNodes() {
+    return getBlockManager().getDatanodeManager().getNumStaleNodes();
+  }
+
   /**
    * Sets the generation stamp for this filesystem
    */

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

@@ -112,4 +112,10 @@ public interface FSNamesystemMBean {
    * @return number of dead data nodes
    */
   public int getNumDeadDataNodes();
+  
+  /**
+   * Number of stale data nodes
+   * @return number of stale data nodes
+   */
+  public int getNumStaleDataNodes();
 }

hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/blockmanagement/BlockManagerTestUtil.java

@@ -191,4 +191,12 @@ public class BlockManagerTestUtil {
         "Must use default policy, got %s", bpp.getClass());
     ((BlockPlacementPolicyDefault)bpp).setPreferLocalNode(prefer);
   }
+  
+  /**
+   * Call heartbeat check function of HeartbeatManager
+   * @param bm the BlockManager to manipulate
+   */
+  public static void checkHeartbeat(BlockManager bm) {
+    bm.getDatanodeManager().getHeartbeatManager().heartbeatCheck();
+  }
 }

hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/namenode/metrics/TestNameNodeMetrics.java

@@ -41,10 +41,14 @@ import org.apache.hadoop.hdfs.MiniDFSCluster;
 import org.apache.hadoop.hdfs.protocol.HdfsConstants.SafeModeAction;
 import org.apache.hadoop.hdfs.protocol.LocatedBlock;
 import org.apache.hadoop.hdfs.server.blockmanagement.BlockManager;
+import org.apache.hadoop.hdfs.server.blockmanagement.BlockManagerTestUtil;
+import org.apache.hadoop.hdfs.server.datanode.DataNode;
+import org.apache.hadoop.hdfs.server.datanode.DataNodeTestUtils;
 import org.apache.hadoop.hdfs.server.namenode.FSNamesystem;
 import org.apache.hadoop.hdfs.server.namenode.NameNodeAdapter;
 import org.apache.hadoop.metrics2.MetricsRecordBuilder;
 import org.apache.hadoop.test.MetricsAsserts;
+import org.apache.hadoop.util.Time;
 import org.apache.log4j.Level;
 import org.junit.After;
 import org.junit.Before;
@@ -77,7 +81,8 @@ public class TestNameNodeMetrics {
         DFS_REPLICATION_INTERVAL);
     CONF.set(DFSConfigKeys.DFS_METRICS_PERCENTILES_INTERVALS_KEY, 
         "" + PERCENTILES_INTERVAL);
-
+    // Enable stale DataNodes checking
+    CONF.setBoolean(DFSConfigKeys.DFS_NAMENODE_CHECK_STALE_DATANODE_KEY, true);
     ((Log4JLogger)LogFactory.getLog(MetricsAsserts.class))
       .getLogger().setLevel(Level.DEBUG);
   }
@@ -119,6 +124,40 @@ public class TestNameNodeMetrics {
     stm.close();
   }
   
+  /** Test metrics indicating the number of stale DataNodes */
+  @Test
+  public void testStaleNodes() throws Exception {
+    // Set two datanodes as stale
+    for (int i = 0; i < 2; i++) {
+      DataNode dn = cluster.getDataNodes().get(i);
+      DataNodeTestUtils.setHeartbeatsDisabledForTests(dn, true);
+      long staleInterval = CONF.getLong(
+          DFSConfigKeys.DFS_NAMENODE_STALE_DATANODE_INTERVAL_KEY,
+          DFSConfigKeys.DFS_NAMENODE_STALE_DATANODE_INTERVAL_DEFAULT);
+      cluster.getNameNode().getNamesystem().getBlockManager()
+          .getDatanodeManager().getDatanode(dn.getDatanodeId())
+          .setLastUpdate(Time.now() - staleInterval - 1);
+    }
+    // Let HeartbeatManager to check heartbeat
+    BlockManagerTestUtil.checkHeartbeat(cluster.getNameNode().getNamesystem()
+        .getBlockManager());
+    assertGauge("StaleDataNodes", 2, getMetrics(NS_METRICS));
+    
+    // Reset stale datanodes
+    for (int i = 0; i < 2; i++) {
+      DataNode dn = cluster.getDataNodes().get(i);
+      DataNodeTestUtils.setHeartbeatsDisabledForTests(dn, false);
+      cluster.getNameNode().getNamesystem().getBlockManager()
+          .getDatanodeManager().getDatanode(dn.getDatanodeId())
+          .setLastUpdate(Time.now());
+    }
+    
+    // Let HeartbeatManager to refresh
+    BlockManagerTestUtil.checkHeartbeat(cluster.getNameNode().getNamesystem()
+        .getBlockManager());
+    assertGauge("StaleDataNodes", 0, getMetrics(NS_METRICS));
+  }
+  
   /** Test metrics associated with addition of a file */
   @Test
   public void testFileAdd() throws Exception {