YARN-3212. RMNode State Transition Update with DECOMMISSIONING state.…

… (Junping Du via wangda)
apache · Sep 18, 2015 · 9bc913a · 9bc913a
1 parent 3f42753
commit 9bc913a
Show file tree

Hide file tree

Showing 7 changed files with 513 additions and 169 deletions.
diff --git a/hadoop-yarn-project/CHANGES.txt b/hadoop-yarn-project/CHANGES.txt
@@ -197,6 +197,9 @@ Release 2.8.0 - UNRELEASED
     YARN-4034. Render cluster Max Priority in scheduler metrics in RM web
     UI. (Rohith Sharma K S via jianhe)
 
+    YARN-3212. RMNode State Transition Update with DECOMMISSIONING state.
+    (Junping Du via wangda)
+
   IMPROVEMENTS
 
     YARN-644. Basic null check is not performed on passed in arguments before

diff --git a/...manager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/NodesListManager.java b/...manager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/NodesListManager.java
@@ -399,7 +399,7 @@ public void refreshNodesGracefully(Configuration conf) throws IOException,
       NodeId nodeId = entry.getKey();
       if (!isValidNode(nodeId.getHost())) {
         this.rmContext.getDispatcher().getEventHandler().handle(
-            new RMNodeEvent(nodeId, RMNodeEventType.DECOMMISSION_WITH_TIMEOUT));
+            new RMNodeEvent(nodeId, RMNodeEventType.GRACEFUL_DECOMMISSION));
       } else {
         // Recommissioning the nodes
         if (entry.getValue().getState() == NodeState.DECOMMISSIONING

diff --git a/...r/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/ResourceTrackerService.java b/...r/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/ResourceTrackerService.java
@@ -44,6 +44,7 @@
 import org.apache.hadoop.yarn.api.records.ContainerStatus;
 import org.apache.hadoop.yarn.api.records.NodeId;
 import org.apache.hadoop.yarn.api.records.NodeLabel;
+import org.apache.hadoop.yarn.api.records.NodeState;
 import org.apache.hadoop.yarn.api.records.Resource;
 import org.apache.hadoop.yarn.conf.YarnConfiguration;
 import org.apache.hadoop.yarn.exceptions.YarnException;
@@ -399,8 +400,10 @@ public NodeHeartbeatResponse nodeHeartbeat(NodeHeartbeatRequest request)
 
     NodeId nodeId = remoteNodeStatus.getNodeId();
 
-    // 1. Check if it's a valid (i.e. not excluded) node
+    // 1. Check if it's a valid (i.e. not excluded) node, if not, see if it is
-    if (!this.nodesListManager.isValidNode(nodeId.getHost())) {
+    // in decommissioning.
+    if (!this.nodesListManager.isValidNode(nodeId.getHost())
+        && !isNodeInDecommissioning(nodeId)) {
       String message =
           "Disallowed NodeManager nodeId: " + nodeId + " hostname: "
               + nodeId.getHost();
@@ -486,6 +489,19 @@ public NodeHeartbeatResponse nodeHeartbeat(NodeHeartbeatRequest request)
     return nodeHeartBeatResponse;
   }
 
+  /**
+   * Check if node in decommissioning state.
+   * @param nodeId
+   */
+  private boolean isNodeInDecommissioning(NodeId nodeId) {
+    RMNode rmNode = this.rmContext.getRMNodes().get(nodeId);
+    if (rmNode != null &&
+        rmNode.getState().equals(NodeState.DECOMMISSIONING)) {
+      return true;
+    }
+    return false;
+  }
+
   @SuppressWarnings("unchecked")
   @Override
   public UnRegisterNodeManagerResponse unRegisterNodeManager(

diff --git a/...r/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/rmnode/RMNodeEventType.java b/...r/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/rmnode/RMNodeEventType.java
@@ -24,7 +24,7 @@ public enum RMNodeEventType {
 
   // Source: AdminService
   DECOMMISSION,
-  DECOMMISSION_WITH_TIMEOUT,
+  GRACEFUL_DECOMMISSION,
   RECOMMISSION,
 
   // Source: AdminService, ResourceTrackerService