MAPREDUCE-5616. MR Client-AppMaster RPC max retries on socket timeout is too high. Contributed by Chris Nauroth.

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1542001 13f79535-47bb-0310-9956-ffa450edef68

hadoop-mapreduce-project/CHANGES.txt

@@ -217,6 +217,9 @@ Release 2.3.0 - UNRELEASED
 
     MAPREDUCE-5610. TestSleepJob fails in jdk7 (Jonathan Eagles via jlowe)
 
+    MAPREDUCE-5616. MR Client-AppMaster RPC max retries on socket timeout is too
+    high. (cnauroth)
+
 Release 2.2.1 - UNRELEASED
 
   INCOMPATIBLE CHANGES

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/MRJobConfig.java

@@ -357,13 +357,21 @@ public interface MRJobConfig {
 
   public static final String MR_AM_PREFIX = MR_PREFIX + "am.";
 
-  /** The number of client retires to the AM - before reconnecting to the RM
+  /** The number of client retries to the AM - before reconnecting to the RM
    * to fetch Application State. 
    */
   public static final String MR_CLIENT_TO_AM_IPC_MAX_RETRIES = 
     MR_PREFIX + "client-am.ipc.max-retries";
   public static final int DEFAULT_MR_CLIENT_TO_AM_IPC_MAX_RETRIES = 3;
   
+  /** The number of client retries on socket timeouts to the AM - before
+   * reconnecting to the RM to fetch Application Status.
+   */
+  public static final String MR_CLIENT_TO_AM_IPC_MAX_RETRIES_ON_TIMEOUTS =
+    MR_PREFIX + "yarn.app.mapreduce.client-am.ipc.max-retries-on-timeouts";
+  public static final int
+    DEFAULT_MR_CLIENT_TO_AM_IPC_MAX_RETRIES_ON_TIMEOUTS = 3;
+
   /**
    * The number of client retries to the RM/HS before throwing exception.
    */

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/resources/mapred-default.xml

@@ -1010,6 +1010,13 @@
     to the RM to fetch Application Status.</description>
 </property>
 
+<property>
+  <name>yarn.app.mapreduce.client-am.ipc.max-retries-on-timeouts</name>
+  <value>3</value>
+  <description>The number of client retries on socket timeouts to the AM - before
+    reconnecting to the RM to fetch Application Status.</description>
+</property>
+
 <property>
   <name>yarn.app.mapreduce.client.max-retries</name>
   <value>3</value>

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/src/main/java/org/apache/hadoop/mapred/ClientServiceDelegate.java

@@ -107,6 +107,10 @@ public class ClientServiceDelegate {
         CommonConfigurationKeysPublic.IPC_CLIENT_CONNECT_MAX_RETRIES_KEY,
         this.conf.getInt(MRJobConfig.MR_CLIENT_TO_AM_IPC_MAX_RETRIES,
             MRJobConfig.DEFAULT_MR_CLIENT_TO_AM_IPC_MAX_RETRIES));
+    this.conf.setInt(
+        CommonConfigurationKeysPublic.IPC_CLIENT_CONNECT_MAX_RETRIES_ON_SOCKET_TIMEOUTS_KEY,
+        this.conf.getInt(MRJobConfig.MR_CLIENT_TO_AM_IPC_MAX_RETRIES_ON_TIMEOUTS,
+            MRJobConfig.DEFAULT_MR_CLIENT_TO_AM_IPC_MAX_RETRIES_ON_TIMEOUTS));
     this.rm = rm;
     this.jobId = jobId;
     this.historyServerProxy = historyServerProxy;