HDFS-14858. [SBN read] Allow configurably enable/disable AlignmentContext on NameNode. Contributed by Chen Liang.

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

@@ -1357,6 +1357,10 @@ public class DFSConfigKeys extends CommonConfigurationKeys {
       "dfs.namenode.send.qop.enabled";
   public static final boolean DFS_NAMENODE_SEND_QOP_ENABLED_DEFAULT = false;
 
+  public static final String DFS_NAMENODE_STATE_CONTEXT_ENABLED_KEY =
+      "dfs.namenode.state.context.enabled";
+  public static final boolean DFS_NAMENODE_STATE_CONTEXT_ENABLED_DEFAULT = false;
+
   // dfs.client.retry confs are moved to HdfsClientConfigKeys.Retry
   @Deprecated
   public static final String  DFS_CLIENT_RETRY_POLICY_ENABLED_KEY

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

@@ -27,6 +27,8 @@
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_SERVICE_HANDLER_COUNT_DEFAULT;
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_SERVICE_HANDLER_COUNT_KEY;
 import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_RPC_ADDRESS_AUXILIARY_KEY;
+import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_STATE_CONTEXT_ENABLED_DEFAULT;
+import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_STATE_CONTEXT_ENABLED_KEY;
 import static org.apache.hadoop.hdfs.server.common.HdfsServerConstants.MAX_PATH_DEPTH;
 import static org.apache.hadoop.hdfs.server.common.HdfsServerConstants.MAX_PATH_LENGTH;
 import static org.apache.hadoop.util.Time.now;
@@ -447,6 +449,16 @@ public NameNodeRpcServer(Configuration conf, NameNode nn)
     }
     LOG.info("RPC server is binding to " + bindHost + ":" + rpcAddr.getPort());
 
+    boolean enableStateContext = conf.getBoolean(
+        DFS_NAMENODE_STATE_CONTEXT_ENABLED_KEY,
+        DFS_NAMENODE_STATE_CONTEXT_ENABLED_DEFAULT);
+    LOG.info("Enable NameNode state context:" + enableStateContext);
+
+    GlobalStateIdContext stateIdContext = null;
+    if (enableStateContext) {
+      stateIdContext = new GlobalStateIdContext((namesystem));
+    }
+
     clientRpcServer = new RPC.Builder(conf)
         .setProtocol(
             org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolPB.class)
@@ -456,7 +468,7 @@ public NameNodeRpcServer(Configuration conf, NameNode nn)
         .setNumHandlers(handlerCount)
         .setVerbose(false)
         .setSecretManager(namesystem.getDelegationTokenSecretManager())
-        .setAlignmentContext(new GlobalStateIdContext(namesystem))
+        .setAlignmentContext(stateIdContext)
         .build();
 
     // Add all the RPC protocols that the namenode implements

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

@@ -3302,6 +3302,17 @@
   </description>
 </property>
 
+<property>
+  <name>dfs.namenode.state.context.enabled</name>
+  <value>false</value>
+  <description>
+    Whether enable namenode sending back its current txnid back to client.
+    Setting this to true is required by Consistent Read from Standby feature.
+    But for regular cases, this should be set to false to avoid the overhead
+    of updating and maintaining this state.
+  </description>
+</property>
+
 <property>
   <name>dfs.namenode.ec.system.default.policy</name>
   <value>RS-6-3-1024k</value>

hadoop-hdfs-project/hadoop-hdfs/src/site/markdown/ObserverNameNode.md

@@ -120,6 +120,20 @@ Deployment
 To enable consistent reads from Observer NameNode, you'll need to add a
 few configurations to your **hdfs-site.xml**:
 
+*  **dfs.namenode.state.context.enabled** - to enable NameNode to maintain
+   and update server state and id.
+
+   This will lead to NameNode creating alignment context instance, which
+   keeps track of current server state id. Server state id will be carried
+   back to client. It is disabled by default to optimize performance of
+   Observer read cases. But this is **required to be turned on**
+   for the Observer NameNode feature.
+
+        <property>
+           <name>dfs.namenode.state.context.enabled</name>
+           <value>true</value>
+        </property>
+
 *  **dfs.ha.tail-edits.in-progress** - to enable fast tailing on
    in-progress edit logs.
 

hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestStateAlignmentContextWithHA.java

@@ -94,6 +94,7 @@ public static void startUpCluster() throws IOException {
     CONF.setBoolean(String.format(
         "fs.%s.impl.disable.cache", HdfsConstants.HDFS_URI_SCHEME), true);
     CONF.setInt(DFSConfigKeys.DFS_REPLICATION_KEY, NUMDATANODES);
+    CONF.setBoolean(DFSConfigKeys.DFS_NAMENODE_STATE_CONTEXT_ENABLED_KEY, true);
 
     qjmhaCluster = HATestUtil.setUpObserverCluster(CONF, 1, NUMDATANODES, true);
     cluster = qjmhaCluster.getDfsCluster();

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

@@ -17,6 +17,7 @@
  */
 package org.apache.hadoop.hdfs.server.namenode.ha;
 
+import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_STATE_CONTEXT_ENABLED_KEY;
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertTrue;
 import static org.junit.Assert.fail;
@@ -71,6 +72,7 @@ public class TestConsistentReadsObserver {
   @BeforeClass
   public static void startUpCluster() throws Exception {
     conf = new Configuration();
+    conf.setBoolean(DFS_NAMENODE_STATE_CONTEXT_ENABLED_KEY, true);
     // disable fast tailing here because this test's assertions are based on the
     // timing of explicitly called rollEditLogAndTail. Although this means this
     // test takes some time to run

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

@@ -17,6 +17,7 @@
  */
 package org.apache.hadoop.hdfs.server.namenode.ha;
 
+import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_STATE_CONTEXT_ENABLED_KEY;
 import static org.junit.Assert.assertTrue;
 
 import java.io.IOException;
@@ -46,6 +47,7 @@ public class TestMultiObserverNode {
   @BeforeClass
   public static void startUpCluster() throws Exception {
     conf = new Configuration();
+    conf.setBoolean(DFS_NAMENODE_STATE_CONTEXT_ENABLED_KEY, true);
     qjmhaCluster = HATestUtil.setUpObserverCluster(conf, 2, 0, true);
     dfsCluster = qjmhaCluster.getDfsCluster();
     dfs = HATestUtil.configureObserverReadFs(

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

@@ -17,6 +17,7 @@
  */
 package org.apache.hadoop.hdfs.server.namenode.ha;
 
+import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_STATE_CONTEXT_ENABLED_KEY;
 import static org.apache.hadoop.hdfs.server.namenode.NameNodeAdapter.getServiceState;
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertTrue;
@@ -77,6 +78,7 @@ public class TestObserverNode {
   @BeforeClass
   public static void startUpCluster() throws Exception {
     conf = new Configuration();
+    conf.setBoolean(DFS_NAMENODE_STATE_CONTEXT_ENABLED_KEY, true);
     qjmhaCluster = HATestUtil.setUpObserverCluster(conf, 1, 0, true);
     dfsCluster = qjmhaCluster.getDfsCluster();
   }