YARN-5718. TimelineClient (and other places in YARN) shouldn't over-write HDFS client retry settings which could cause unexpected behavior. Contributed by Junping Du.

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java

@@ -695,10 +695,6 @@ public static boolean isAclEnabled(Configuration conf) {
   /** URI for FileSystemRMStateStore */
   public static final String FS_RM_STATE_STORE_URI = RM_PREFIX
       + "fs.state-store.uri";
-  public static final String FS_RM_STATE_STORE_RETRY_POLICY_SPEC = RM_PREFIX
-      + "fs.state-store.retry-policy-spec";
-  public static final String DEFAULT_FS_RM_STATE_STORE_RETRY_POLICY_SPEC =
-      "2000, 500";
 
   public static final String FS_RM_STATE_STORE_NUM_RETRIES =
       RM_PREFIX + "fs.state-store.num-retries";
@@ -1974,13 +1970,6 @@ public static boolean isAclEnabled(Configuration conf) {
       TIMELINE_SERVICE_ENTITYGROUP_FS_STORE_UNKNOWN_ACTIVE_SECONDS_DEFAULT
       = 24 * 60 * 60;
 
-  public static final String
-      TIMELINE_SERVICE_ENTITYGROUP_FS_STORE_RETRY_POLICY_SPEC =
-      TIMELINE_SERVICE_ENTITYGROUP_FS_STORE_PREFIX + "retry-policy-spec";
-  public static final String
-      DEFAULT_TIMELINE_SERVICE_ENTITYGROUP_FS_STORE_RETRY_POLICY_SPEC =
-      "2000, 500";
-
   public static final String TIMELINE_SERVICE_LEVELDB_CACHE_READ_CACHE_SIZE =
       TIMELINE_SERVICE_ENTITYGROUP_FS_STORE_PREFIX
           + "leveldb-cache-read-cache-size";
@@ -2600,11 +2589,7 @@ public static boolean isAclEnabled(Configuration conf) {
   /** URI for NodeLabelManager */
   public static final String FS_NODE_LABELS_STORE_ROOT_DIR = NODE_LABELS_PREFIX
       + "fs-store.root-dir";
-  public static final String FS_NODE_LABELS_STORE_RETRY_POLICY_SPEC =
-      NODE_LABELS_PREFIX + "fs-store.retry-policy-spec";
-  public static final String DEFAULT_FS_NODE_LABELS_STORE_RETRY_POLICY_SPEC =
-      "2000, 500";
-  
+
   /**
    * Flag to indicate if the node labels feature enabled, by default it's
    * disabled

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/java/org/apache/hadoop/yarn/client/api/impl/FileSystemTimelineWriter.java

@@ -106,13 +106,6 @@ public FileSystemTimelineWriter(Configuration conf,
     super(authUgi, client, resURI);
 
     Configuration fsConf = new Configuration(conf);
-    fsConf.setBoolean("dfs.client.retry.policy.enabled", true);
-    String retryPolicy =
-        fsConf.get(YarnConfiguration.
-            TIMELINE_SERVICE_ENTITYGROUP_FS_STORE_RETRY_POLICY_SPEC,
-          YarnConfiguration.
-              DEFAULT_TIMELINE_SERVICE_ENTITYGROUP_FS_STORE_RETRY_POLICY_SPEC);
-    fsConf.set("dfs.client.retry.policy.spec", retryPolicy);
 
     activePath = new Path(fsConf.get(
       YarnConfiguration

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/java/org/apache/hadoop/yarn/nodelabels/FileSystemNodeLabelsStore.java

@@ -95,13 +95,8 @@ public void close() throws IOException {
 
   void setFileSystem(Configuration conf) throws IOException {
     Configuration confCopy = new Configuration(conf);
-    confCopy.setBoolean("dfs.client.retry.policy.enabled", true);
-    String retryPolicy =
-        confCopy.get(YarnConfiguration.FS_NODE_LABELS_STORE_RETRY_POLICY_SPEC,
-            YarnConfiguration.DEFAULT_FS_NODE_LABELS_STORE_RETRY_POLICY_SPEC);
-    confCopy.set("dfs.client.retry.policy.spec", retryPolicy);
     fs = fsWorkingPath.getFileSystem(confCopy);
-    
+
     // if it's local file system, use RawLocalFileSystem instead of
     // LocalFileSystem, the latter one doesn't support append.
     if (fs.getScheme().equals("file")) {

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/resources/yarn-default.xml

@@ -521,16 +521,6 @@
     <!--value>hdfs://localhost:9000/rmstore</value-->
   </property>
 
-  <property>
-    <description>hdfs client retry policy specification. hdfs client retry
-    is always enabled. Specified in pairs of sleep-time and number-of-retries
-    and (t0, n0), (t1, n1), ..., the first n0 retries sleep t0 milliseconds on
-    average, the following n1 retries sleep t1 milliseconds on average, and so on.
-    </description>
-    <name>yarn.resourcemanager.fs.state-store.retry-policy-spec</name>
-    <value>2000, 500</value>
-  </property>
-
   <property>
     <description>the number of retries to recover from IOException in
     FileSystemRMStateStore.
@@ -2481,16 +2471,6 @@
     <value>false</value>
   </property>
 
-  <property>
-    <description>
-    Retry policy used for FileSystem node label store. The policy is
-    specified by N pairs of sleep-time in milliseconds and number-of-retries
-    "s1,n1,s2,n2,...".
-    </description>
-    <name>yarn.node-labels.fs-store.retry-policy-spec</name>
-    <value>2000, 500</value>
-  </property>
-
   <property>
     <description>
     URI for NodeLabelManager.  The default value is

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/FileSystemRMStateStore.java

@@ -142,11 +142,6 @@ protected synchronized void startInternal() throws Exception {
     // authenticated with kerberos so we are good to create a file-system
     // handle.
     fsConf = new Configuration(getConfig());
-    fsConf.setBoolean("dfs.client.retry.policy.enabled", true);
-    String retryPolicy =
-        fsConf.get(YarnConfiguration.FS_RM_STATE_STORE_RETRY_POLICY_SPEC,
-          YarnConfiguration.DEFAULT_FS_RM_STATE_STORE_RETRY_POLICY_SPEC);
-    fsConf.set("dfs.client.retry.policy.spec", retryPolicy);
 
     String scheme = fsWorkingPath.toUri().getScheme();
     if (scheme == null) {

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestFSRMStateStore.java

@@ -113,8 +113,6 @@ public RMStateStore getRMStateStore() throws Exception {
       YarnConfiguration conf = new YarnConfiguration();
       conf.set(YarnConfiguration.FS_RM_STATE_STORE_URI,
           workingDirPathURI.toString());
-      conf.set(YarnConfiguration.FS_RM_STATE_STORE_RETRY_POLICY_SPEC,
-              "100,6000");
       conf.setInt(YarnConfiguration.FS_RM_STATE_STORE_NUM_RETRIES, 8);
       conf.setLong(YarnConfiguration.FS_RM_STATE_STORE_RETRY_INTERVAL_MS,
               900L);
@@ -325,8 +323,6 @@ public RMStateStore getRMStateStore() throws Exception {
           YarnConfiguration conf = new YarnConfiguration();
           conf.set(YarnConfiguration.FS_RM_STATE_STORE_URI,
               workingDirPathURI.toString());
-          conf.set(YarnConfiguration.FS_RM_STATE_STORE_RETRY_POLICY_SPEC,
-              "100,6000");
           this.store = new TestFileSystemRMStore(conf) {
             Version storedVersion = null;