HDFS-8716. Introduce a new config specifically for safe mode block count. Contributed by Chang Li.

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

@@ -905,6 +905,9 @@ Release 2.8.0 - UNRELEASED
 
     HDFS-8860. Remove unused Replica copyOnWrite code (Lei (Eddy) Xu via Colin P. McCabe)
 
+    HDFS-8716. Introduce a new config specifically for safe mode block count
+    (Chang Li via kihwal)
+
   OPTIMIZATIONS
 
     HDFS-8026. Trace FSOutputSummer#writeChecksumChunks rather than

hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/DFSConfigKeys.java

@@ -170,6 +170,8 @@ public class DFSConfigKeys extends CommonConfigurationKeys {
   public static final int     DFS_NAMENODE_REPLICATION_INTERVAL_DEFAULT = 3;
   public static final String  DFS_NAMENODE_REPLICATION_MIN_KEY = "dfs.namenode.replication.min";
   public static final int     DFS_NAMENODE_REPLICATION_MIN_DEFAULT = 1;
+  public static final String  DFS_NAMENODE_SAFEMODE_REPLICATION_MIN_KEY =
+      "dfs.namenode.safemode.replication.min";
   public static final String  DFS_NAMENODE_REPLICATION_PENDING_TIMEOUT_SEC_KEY = "dfs.namenode.replication.pending.timeout-sec";
   public static final int     DFS_NAMENODE_REPLICATION_PENDING_TIMEOUT_SEC_DEFAULT = -1;
   public static final String  DFS_NAMENODE_REPLICATION_MAX_STREAMS_KEY = "dfs.namenode.replication.max-streams";

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

@@ -64,8 +64,6 @@ import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_INODE_ATTRIBUTES
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_MAX_OBJECTS_DEFAULT;
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_MAX_OBJECTS_KEY;
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_NAME_DIR_KEY;
-import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_REPLICATION_MIN_DEFAULT;
-import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_REPLICATION_MIN_KEY;
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_REPL_QUEUE_THRESHOLD_PCT_KEY;
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_RESOURCE_CHECK_INTERVAL_DEFAULT;
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_RESOURCE_CHECK_INTERVAL_KEY;
@@ -4151,9 +4149,17 @@ public class FSNamesystem implements Namesystem, FSNamesystemMBean,
         DFS_NAMENODE_SAFEMODE_MIN_DATANODES_KEY,
         DFS_NAMENODE_SAFEMODE_MIN_DATANODES_DEFAULT);
       this.extension = conf.getInt(DFS_NAMENODE_SAFEMODE_EXTENSION_KEY, 0);
-      this.safeReplication = conf.getInt(DFS_NAMENODE_REPLICATION_MIN_KEY, 
-                                         DFS_NAMENODE_REPLICATION_MIN_DEFAULT);
-      
+      int minReplication =
+          conf.getInt(DFSConfigKeys.DFS_NAMENODE_REPLICATION_MIN_KEY,
+              DFSConfigKeys.DFS_NAMENODE_REPLICATION_MIN_DEFAULT);
+      // DFS_NAMENODE_SAFEMODE_REPLICATION_MIN_KEY is an expert level setting,
+      // setting this lower than the min replication is not recommended
+      // and/or dangerous for production setups.
+      // When it's unset, safeReplication will use dfs.namenode.replication.min
+      this.safeReplication =
+          conf.getInt(DFSConfigKeys.DFS_NAMENODE_SAFEMODE_REPLICATION_MIN_KEY,
+              minReplication);
+
       LOG.info(DFS_NAMENODE_SAFEMODE_THRESHOLD_PCT_KEY + " = " + threshold);
       LOG.info(DFS_NAMENODE_SAFEMODE_MIN_DATANODES_KEY + " = " + datanodeThreshold);
       LOG.info(DFS_NAMENODE_SAFEMODE_EXTENSION_KEY + "     = " + extension);

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

@@ -469,6 +469,18 @@
   </description>
 </property>
 
+<property>
+  <name>dfs.namenode.safemode.replication.min</name>
+  <value></value>
+  <description>
+      a separate minimum replication factor for calculating safe block count.
+      This is an expert level setting.
+      Setting this lower than the dfs.namenode.replication.min
+      is not recommend and/or dangerous for production setups.
+      When it's not set it takes value from dfs.namenode.replication.min
+  </description>
+</property>
+
 <property>
   <name>dfs.blocksize</name>
   <value>134217728</value>

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

@@ -29,11 +29,13 @@ import java.util.Collection;
 
 import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.fs.FileUtil;
+import org.apache.hadoop.hdfs.DFSConfigKeys;
 import org.apache.hadoop.hdfs.DFSTestUtil;
 import org.apache.hadoop.hdfs.HdfsConfiguration;
 import org.apache.hadoop.hdfs.MiniDFSCluster;
 import org.apache.hadoop.hdfs.server.blockmanagement.BlockManager;
 import org.apache.hadoop.hdfs.server.common.HdfsServerConstants.NamenodeRole;
+import org.apache.hadoop.hdfs.server.namenode.FSNamesystem.SafeModeInfo;
 import org.apache.hadoop.hdfs.server.namenode.ha.HAContext;
 import org.apache.hadoop.hdfs.server.namenode.ha.HAState;
 import org.apache.hadoop.hdfs.server.namenode.snapshot.Snapshot;
@@ -264,4 +266,16 @@ public class TestFSNamesystem {
     Assert.assertEquals("Expected number of blocked thread not found",
                         threadCount, rwLock.getQueueLength());
   }
+
+  @Test
+  public void testSafemodeReplicationConf() throws IOException {
+    Configuration conf = new Configuration();
+    FSImage fsImage = Mockito.mock(FSImage.class);
+    FSEditLog fsEditLog = Mockito.mock(FSEditLog.class);
+    Mockito.when(fsImage.getEditLog()).thenReturn(fsEditLog);
+    conf.setInt(DFSConfigKeys.DFS_NAMENODE_REPLICATION_MIN_KEY, 2);
+    FSNamesystem fsn = new FSNamesystem(conf, fsImage);
+    SafeModeInfo safemodeInfo = fsn.getSafeModeInfoForTests();
+    assertTrue(safemodeInfo.toString().contains("Minimal replication = 2"));
+  }
 }