From 17625e40f61ac31fc87f4a74d66388dbe508d33d Mon Sep 17 00:00:00 2001
From: Eric Yang <eyang@apache.org>
Date: Fri, 27 Jul 2018 19:33:58 -0400
Subject: [PATCH] YARN-8508.  Release GPU resource for killed container.       
      Contributed by Chandni Singh

(cherry picked from commit ed9d60e888d0acfd748fda7f66249f5b79a3ed6d)
(cherry picked from commit c2c3eee69c8b389525fbde800e057dbcb2fc643e)
---
 .../nodemanager/LinuxContainerExecutor.java   | 34 +++++++++----------
 .../TestLinuxContainerExecutor.java           |  9 ++++-
 2 files changed, 25 insertions(+), 18 deletions(-)

diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/LinuxContainerExecutor.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/LinuxContainerExecutor.java
index 03b88a44997..4253f2f0d54 100644
--- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/LinuxContainerExecutor.java
+++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/LinuxContainerExecutor.java
@@ -573,15 +573,7 @@ public class LinuxContainerExecutor extends ContainerExecutor {
       return handleExitCode(e, container, containerId);
     } finally {
       resourcesHandler.postExecute(containerId);
-
-      try {
-        if (resourceHandlerChain != null) {
-          resourceHandlerChain.postComplete(containerId);
-        }
-      } catch (ResourceHandlerException e) {
-        LOG.warn("ResourceHandlerChain.postComplete failed for " +
-            "containerId: " + containerId + ". Exception: " + e);
-      }
+      postComplete(containerId);
     }
 
     return 0;
@@ -721,14 +713,7 @@ public class LinuxContainerExecutor extends ContainerExecutor {
       return super.reacquireContainer(ctx);
     } finally {
       resourcesHandler.postExecute(containerId);
-      if (resourceHandlerChain != null) {
-        try {
-          resourceHandlerChain.postComplete(containerId);
-        } catch (ResourceHandlerException e) {
-          LOG.warn("ResourceHandlerChain.postComplete failed for " +
-              "containerId: " + containerId + " Exception: " + e);
-        }
-      }
+      postComplete(containerId);
     }
   }
 
@@ -798,6 +783,8 @@ public class LinuxContainerExecutor extends ContainerExecutor {
       logOutput(e.getOutput());
       throw new IOException("Error in reaping container "
           + container.getContainerId().toString() + " exit = " + retCode, e);
+    } finally {
+      postComplete(container.getContainerId());
     }
     return true;
   }
@@ -968,4 +955,17 @@ public class LinuxContainerExecutor extends ContainerExecutor {
       LOG.warn("Unable to remove docker container: " + containerId);
     }
   }
+
+  @VisibleForTesting
+  void postComplete(final ContainerId containerId) {
+    try {
+      if (resourceHandlerChain != null) {
+        LOG.debug("{} post complete", containerId);
+        resourceHandlerChain.postComplete(containerId);
+      }
+    } catch (ResourceHandlerException e) {
+      LOG.warn("ResourceHandlerChain.postComplete failed for " +
+          "containerId: {}. Exception: ", containerId, e);
+    }
+  }
 }
diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/test/java/org/apache/hadoop/yarn/server/nodemanager/TestLinuxContainerExecutor.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/test/java/org/apache/hadoop/yarn/server/nodemanager/TestLinuxContainerExecutor.java
index ddbf3b90568..6d77fc488da 100644
--- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/test/java/org/apache/hadoop/yarn/server/nodemanager/TestLinuxContainerExecutor.java
+++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/test/java/org/apache/hadoop/yarn/server/nodemanager/TestLinuxContainerExecutor.java
@@ -25,11 +25,14 @@ import static org.junit.Assert.assertFalse;
 import static org.junit.Assert.assertNotNull;
 import static org.junit.Assert.assertTrue;
 import static org.junit.Assert.fail;
+import static org.mockito.Matchers.anyObject;
 import static org.mockito.Mockito.mock;
+import static org.mockito.Mockito.spy;
 import static org.mockito.Mockito.times;
 import static org.mockito.Mockito.verify;
 import static org.mockito.Mockito.when;
 
+import org.apache.hadoop.yarn.server.nodemanager.containermanager.linux.runtime.LinuxContainerRuntime;
 import org.apache.hadoop.yarn.server.nodemanager.executor.ContainerReapContext;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -40,6 +43,7 @@ import java.io.IOException;
 import java.io.PrintWriter;
 import java.net.InetSocketAddress;
 import java.util.ArrayList;
+import java.util.Arrays;
 import java.util.EnumSet;
 import java.util.HashMap;
 import java.util.HashSet;
@@ -667,12 +671,15 @@ public class TestLinuxContainerExecutor {
   @Test
   public void testReapContainer() throws Exception {
     Container container = mock(Container.class);
-    LinuxContainerExecutor lce = mock(LinuxContainerExecutor.class);
+    LinuxContainerRuntime containerRuntime = mock(LinuxContainerRuntime.class);
+    LinuxContainerExecutor lce = spy(new LinuxContainerExecutor(
+        containerRuntime));
     ContainerReapContext.Builder builder =  new ContainerReapContext.Builder();
     builder.setContainer(container).setUser("foo");
     ContainerReapContext ctx = builder.build();
     lce.reapContainer(ctx);
     verify(lce, times(1)).reapContainer(ctx);
+    verify(lce, times(1)).postComplete(anyObject());
   }
 
   @Test