HDFS-8863. The remaining space check in BlockPlacementPolicyDefault is flawed. (Kihwal Lee via yliu)

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

@@ -8,11 +8,11 @@ Release 2.7.2 - UNRELEASED
 
   IMPROVEMENTS
 
-  HDFS-8659. Block scanner INFO message is spamming logs. (Yongjun Zhang)
+    HDFS-8659. Block scanner INFO message is spamming logs. (Yongjun Zhang)
 
   OPTIMIZATIONS
 
-  HDFS-8722. Optimize datanode writes for small writes and flushes (kihwal)
+    HDFS-8722. Optimize datanode writes for small writes and flushes (kihwal)
 
   BUG FIXES
 
@@ -27,6 +27,9 @@ Release 2.7.2 - UNRELEASED
 
     HDFS-8867. Enable optimized block reports. (Daryn Sharp via jing9)
 
+    HDFS-8863. The remaining space check in BlockPlacementPolicyDefault is
+    flawed. (Kihwal Lee via yliu)
+
 Release 2.7.1 - 2015-07-06
 
   INCOMPATIBLE CHANGES

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

@@ -779,7 +779,8 @@ public class BlockPlacementPolicyDefault extends BlockPlacementPolicy {
     
     final long requiredSize = blockSize * HdfsConstants.MIN_BLOCKS_FOR_WRITE;
     final long scheduledSize = blockSize * node.getBlocksScheduled(storage.getStorageType());
-    final long remaining = node.getRemaining(storage.getStorageType());
+    final long remaining = node.getRemaining(storage.getStorageType(),
+        requiredSize);
     if (requiredSize > remaining - scheduledSize) {
       logNodeIsNotChosen(storage, "the node does not have enough "
           + storage.getStorageType() + " space"

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

@@ -44,6 +44,7 @@ import org.apache.hadoop.hdfs.protocol.DatanodeInfo;
 import org.apache.hadoop.hdfs.server.namenode.CachedBlock;
 import org.apache.hadoop.hdfs.server.protocol.BlockReportContext;
 import org.apache.hadoop.hdfs.server.protocol.DatanodeStorage;
+import org.apache.hadoop.hdfs.server.protocol.DatanodeStorage.State;
 import org.apache.hadoop.hdfs.server.protocol.StorageReport;
 import org.apache.hadoop.hdfs.server.protocol.VolumeFailureSummary;
 import org.apache.hadoop.hdfs.util.EnumCounters;
@@ -657,16 +658,26 @@ public class DatanodeDescriptor extends DatanodeInfo {
   }
 
   /**
-   * @return Approximate number of blocks currently scheduled to be written 
+   * Return the sum of remaining spaces of the specified type. If the remaining
+   * space of a storage is less than minSize, it won't be counted toward the
+   * sum.
+   *
+   * @param t The storage type. If null, the type is ignored.
+   * @param minSize The minimum free space required.
+   * @return the sum of remaining spaces that are bigger than minSize.
    */
-  public long getRemaining(StorageType t) {
+  public long getRemaining(StorageType t, long minSize) {
     long remaining = 0;
-    for(DatanodeStorageInfo s : getStorageInfos()) {
-      if (s.getStorageType() == t) {
-        remaining += s.getRemaining();
+    for (DatanodeStorageInfo s : getStorageInfos()) {
+      if (s.getState() == State.NORMAL &&
+          (t == null || s.getStorageType() == t)) {
+        long r = s.getRemaining();
+        if (r >= minSize) {
+          remaining += r;
+        }
       }
     }
-    return remaining;    
+    return remaining;
   }
 
   /**

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

@@ -101,6 +101,16 @@ public class TestReplicationPolicy {
         dnCacheCapacity, dnCacheUsed, xceiverCount, volFailures, null);
   }
 
+  private static void updateHeartbeatForExtraStorage(long capacity,
+      long dfsUsed, long remaining, long blockPoolUsed) {
+    DatanodeDescriptor dn = dataNodes[5];
+    dn.getStorageInfos()[1].setUtilizationForTesting(
+        capacity, dfsUsed, remaining, blockPoolUsed);
+    dn.updateHeartbeat(
+        BlockManagerTestUtil.getStorageReportsForDatanode(dn),
+        0L, 0L, 0, 0, null);
+  }
+
   @BeforeClass
   public static void setupCluster() throws Exception {
     Configuration conf = new HdfsConfiguration();
@@ -114,6 +124,16 @@ public class TestReplicationPolicy {
     storages = DFSTestUtil.createDatanodeStorageInfos(racks);
     dataNodes = DFSTestUtil.toDatanodeDescriptor(storages);
 
+    // create an extra storage for dn5.
+    DatanodeStorage extraStorage = new DatanodeStorage(
+        storages[5].getStorageID() + "-extra", DatanodeStorage.State.NORMAL,
+        StorageType.DEFAULT);
+/*    DatanodeStorageInfo si = new DatanodeStorageInfo(
+        storages[5].getDatanodeDescriptor(), extraStorage);
+*/
+    BlockManagerTestUtil.updateStorage(storages[5].getDatanodeDescriptor(),
+        extraStorage);
+
     FileSystem.setDefaultUri(conf, "hdfs://localhost:0");
     conf.set(DFSConfigKeys.DFS_NAMENODE_HTTP_ADDRESS_KEY, "0.0.0.0:0");
     File baseDir = PathUtils.getTestDir(TestReplicationPolicy.class);
@@ -136,11 +156,17 @@ public class TestReplicationPolicy {
       bm.getDatanodeManager().getHeartbeatManager().addDatanode(
           dataNodes[i]);
     }
+    resetHeartbeatForStorages();
+  }
+
+  private static void resetHeartbeatForStorages() {
     for (int i=0; i < NUM_OF_DATANODES; i++) {
       updateHeartbeatWithUsage(dataNodes[i],
           2*HdfsConstants.MIN_BLOCKS_FOR_WRITE*BLOCK_SIZE, 0L,
-          2*HdfsConstants.MIN_BLOCKS_FOR_WRITE*BLOCK_SIZE, 0L, 0L, 0L, 0, 0);
+          HdfsConstants.MIN_BLOCKS_FOR_WRITE*BLOCK_SIZE, 0L, 0L, 0L, 0, 0);
     }    
+    // No available space in the extra storage of dn0
+    updateHeartbeatForExtraStorage(0L, 0L, 0L, 0L);
   }
 
   private static boolean isOnSameRack(DatanodeStorageInfo left, DatanodeStorageInfo right) {
@@ -150,6 +176,31 @@ public class TestReplicationPolicy {
   private static boolean isOnSameRack(DatanodeStorageInfo left, DatanodeDescriptor right) {
     return cluster.isOnSameRack(left.getDatanodeDescriptor(), right);
   }
+
+  /**
+   * Test whether the remaining space per storage is individually
+   * considered.
+   */
+  @Test
+  public void testChooseNodeWithMultipleStorages() throws Exception {
+    updateHeartbeatWithUsage(dataNodes[5],
+        2* HdfsConstants.MIN_BLOCKS_FOR_WRITE*BLOCK_SIZE, 0L,
+        (2*HdfsConstants.MIN_BLOCKS_FOR_WRITE*BLOCK_SIZE)/3, 0L,
+        0L, 0L, 0, 0);
+
+    updateHeartbeatForExtraStorage(
+        2* HdfsConstants.MIN_BLOCKS_FOR_WRITE*BLOCK_SIZE, 0L,
+        (2*HdfsConstants.MIN_BLOCKS_FOR_WRITE*BLOCK_SIZE)/3, 0L);
+
+    DatanodeStorageInfo[] targets;
+    targets = chooseTarget (1, dataNodes[5],
+        new ArrayList<DatanodeStorageInfo>(), null);
+    assertEquals(1, targets.length);
+    assertEquals(storages[4], targets[0]);
+
+    resetHeartbeatForStorages();
+  }
+
   /**
    * In this testcase, client is dataNodes[0]. So the 1st replica should be
    * placed on dataNodes[0], the 2nd replica should be placed on