HDFS-11095. BlockManagerSafeMode should respect extension period default config value (30s). Contributed by Mingliang Liu

(cherry picked from commit 9b0c17f852)

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

@@ -35,12 +35,16 @@ import org.apache.hadoop.hdfs.server.namenode.startupprogress.StepType;
 import org.apache.hadoop.net.NetworkTopology;
 import org.apache.hadoop.util.Daemon;
 
+import com.google.common.annotations.VisibleForTesting;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 import java.util.concurrent.atomic.AtomicLong;
 
+import static java.util.concurrent.TimeUnit.MILLISECONDS;
+
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_REPL_QUEUE_THRESHOLD_PCT_KEY;
+import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_SAFEMODE_EXTENSION_DEFAULT;
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_SAFEMODE_EXTENSION_KEY;
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_SAFEMODE_MIN_DATANODES_DEFAULT;
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_SAFEMODE_MIN_DATANODES_KEY;
@@ -96,7 +100,8 @@ class BlockManagerSafeMode {
   private long blockReplQueueThreshold;
 
   /** How long (in ms) is the extension period. */
-  private final int extension;
+  @VisibleForTesting
+  final long extension;
   /** Timestamp of the first time when thresholds are met. */
   private final AtomicLong reachedTime = new AtomicLong();
   /** Timestamp of the safe mode initialized. */
@@ -143,7 +148,9 @@ class BlockManagerSafeMode {
     this.replQueueThreshold =
         conf.getFloat(DFS_NAMENODE_REPL_QUEUE_THRESHOLD_PCT_KEY,
             (float) threshold);
-    this.extension = conf.getInt(DFS_NAMENODE_SAFEMODE_EXTENSION_KEY, 0);
+    this.extension = conf.getTimeDuration(DFS_NAMENODE_SAFEMODE_EXTENSION_KEY,
+        DFS_NAMENODE_SAFEMODE_EXTENSION_DEFAULT,
+        MILLISECONDS);
 
     this.inRollBack = isInRollBackMode(NameNode.getStartupOption(conf));
 

hadoop-hdfs-project/hadoop-hdfs/src/main/resources/hdfs-default.xml

@@ -823,8 +823,9 @@
   <name>dfs.namenode.safemode.extension</name>
   <value>30000</value>
   <description>
-    Determines extension of safe mode in milliseconds 
+    Determines extension of safe mode in milliseconds after the threshold level
-    after the threshold level is reached.
+    is reached.  Support multiple time unit suffix (case insensitive), as
+    described in dfs.heartbeat.interval.
   </description>
 </property>
 

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

@@ -38,6 +38,7 @@ import java.io.IOException;
 import java.io.InterruptedIOException;
 import java.util.concurrent.TimeoutException;
 
+import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_SAFEMODE_EXTENSION_DEFAULT;
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertFalse;
 import static org.junit.Assert.assertTrue;
@@ -386,6 +387,25 @@ public class TestBlockManagerSafeMode {
         "are cleared.", bmSafeMode.leaveSafeMode(false));
   }
 
+  @Test(timeout = 10000)
+  public void testExtensionConfig() {
+    final Configuration conf = new HdfsConfiguration();
+    bmSafeMode = new BlockManagerSafeMode(bm, fsn, false, conf);
+    assertEquals(DFS_NAMENODE_SAFEMODE_EXTENSION_DEFAULT, bmSafeMode.extension);
+
+    conf.set(DFSConfigKeys.DFS_NAMENODE_SAFEMODE_EXTENSION_KEY, "30000");
+    bmSafeMode = new BlockManagerSafeMode(bm, fsn, false, conf);
+    assertEquals(30000, bmSafeMode.extension);
+
+    conf.set(DFSConfigKeys.DFS_NAMENODE_SAFEMODE_EXTENSION_KEY, "20s");
+    bmSafeMode = new BlockManagerSafeMode(bm, fsn, false, conf);
+    assertEquals(20 * 1000, bmSafeMode.extension);
+
+    conf.set(DFSConfigKeys.DFS_NAMENODE_SAFEMODE_EXTENSION_KEY, "7m");
+    bmSafeMode = new BlockManagerSafeMode(bm, fsn, false, conf);
+    assertEquals(7 * 60 * 1000, bmSafeMode.extension);
+  }
+
   /**
    * Test get safe mode tip.
    */