From 28d99db99236ff2a6e4a605802820e2b512225f9 Mon Sep 17 00:00:00 2001
From: Karthik Kambatla <kasha@apache.org>
Date: Tue, 9 Sep 2014 15:13:07 -0700
Subject: [PATCH] YARN-2526. SLS can deadlock when all the threads are taken by
 AMSimulators. (Wei Yan via kasha)

---
 .../yarn/sls/appmaster/MRAMSimulator.java     | 43 ++++++++++---------
 hadoop-yarn-project/CHANGES.txt               |  3 ++
 2 files changed, 26 insertions(+), 20 deletions(-)

diff --git a/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/MRAMSimulator.java b/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/MRAMSimulator.java
index fb702059ad..da267a1c1b 100644
--- a/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/MRAMSimulator.java
+++ b/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/MRAMSimulator.java
@@ -179,26 +179,8 @@ public AllocateResponse run() throws Exception {
         return rm.getApplicationMasterService().allocate(request);
       }
     });
-
-    // waiting until the AM container is allocated
-    while (true) {
-      if (response != null && ! response.getAllocatedContainers().isEmpty()) {
-        // get AM container
-        Container container = response.getAllocatedContainers().get(0);
-        se.getNmMap().get(container.getNodeId())
-                .addNewContainer(container, -1L);
-        // start AM container
-        amContainer = container;
-        LOG.debug(MessageFormat.format("Application {0} starts its " +
-                "AM container ({1}).", appId, amContainer.getId()));
-        isAMContainerRunning = true;
-        break;
-      }
-      // this sleep time is different from HeartBeat
-      Thread.sleep(1000);
-      // send out empty request
-      sendContainerRequest();
-      response = responseQueue.take();
+    if (response != null) {
+      responseQueue.put(response);
     }
   }
 
@@ -206,6 +188,26 @@ public AllocateResponse run() throws Exception {
   @SuppressWarnings("unchecked")
   protected void processResponseQueue()
           throws InterruptedException, YarnException, IOException {
+    // Check whether receive the am container
+    if (!isAMContainerRunning) {
+      if (!responseQueue.isEmpty()) {
+        AllocateResponse response = responseQueue.take();
+        if (response != null
+            && !response.getAllocatedContainers().isEmpty()) {
+          // Get AM container
+          Container container = response.getAllocatedContainers().get(0);
+          se.getNmMap().get(container.getNodeId())
+              .addNewContainer(container, -1L);
+          // Start AM container
+          amContainer = container;
+          LOG.debug(MessageFormat.format("Application {0} starts its " +
+              "AM container ({1}).", appId, amContainer.getId()));
+          isAMContainerRunning = true;
+        }
+      }
+      return;
+    }
+
     while (! responseQueue.isEmpty()) {
       AllocateResponse response = responseQueue.take();
 
@@ -262,6 +264,7 @@ protected void processResponseQueue()
         LOG.debug(MessageFormat.format("Application {0} sends out event " +
                 "to clean up its AM container.", appId));
         isFinished = true;
+        break;
       }
 
       // check allocated containers
diff --git a/hadoop-yarn-project/CHANGES.txt b/hadoop-yarn-project/CHANGES.txt
index d799c28839..7eaf1c805d 100644
--- a/hadoop-yarn-project/CHANGES.txt
+++ b/hadoop-yarn-project/CHANGES.txt
@@ -302,6 +302,9 @@ Release 2.6.0 - UNRELEASED
     YARN-2519. Credential Provider related unit tests failed on Windows.
     (Xiaoyu Yao via cnauroth)
 
+    YARN-2526. SLS can deadlock when all the threads are taken by AMSimulators. 
+    (Wei Yan via kasha)
+
 Release 2.5.1 - UNRELEASED
 
   INCOMPATIBLE CHANGES