YARN-1401. With zero sleep-delay-before-sigkill.ms, no signal is ever sent (Gera Shegalov via Sandy Ryza)

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1542038 13f79535-47bb-0310-9956-ffa450edef68

hadoop-yarn-project/CHANGES.txt

@@ -130,6 +130,9 @@ Release 2.3.0 - UNRELEASED
     YARN-1400. yarn.cmd uses HADOOP_RESOURCEMANAGER_OPTS. Should be
     YARN_RESOURCEMANAGER_OPTS. (Raja Aluri via cnauroth)
 
+    YARN-1401. With zero sleep-delay-before-sigkill.ms, no signal is ever sent
+    (Gera Shegalov via Sandy Ryza)
+
 Release 2.2.1 - UNRELEASED
 
   INCOMPATIBLE CHANGES

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/containermanager/launcher/ContainerLaunch.java

@@ -375,13 +375,19 @@ public class ContainerLaunch implements Callable<Integer> {
         LOG.debug("Sending signal to pid " + processId
             + " as user " + user
             + " for container " + containerIdStr);
+
+        final Signal signal = sleepDelayBeforeSigKill > 0
+          ? Signal.TERM
+          : Signal.KILL;
+
+        boolean result = exec.signalContainer(user, processId, signal);
+
+        LOG.debug("Sent signal " + signal + " to pid " + processId
+          + " as user " + user
+          + " for container " + containerIdStr
+          + ", result=" + (result? "success" : "failed"));
+
         if (sleepDelayBeforeSigKill > 0) {
-          boolean result = exec.signalContainer(user,
-              processId, Signal.TERM);
-          LOG.debug("Sent signal to pid " + processId
-              + " as user " + user
-              + " for container " + containerIdStr
-              + ", result=" + (result? "success" : "failed"));
           new DelayedProcessKiller(container, user,
               processId, sleepDelayBeforeSigKill, Signal.KILL, exec).start();
         }

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/test/java/org/apache/hadoop/yarn/server/nodemanager/containermanager/launcher/TestContainerLaunch.java

@@ -97,7 +97,6 @@ public class TestContainerLaunch extends BaseContainerManagerTest {
     conf.setClass(
         YarnConfiguration.NM_CONTAINER_MON_RESOURCE_CALCULATOR,
         LinuxResourceCalculatorPlugin.class, ResourceCalculatorPlugin.class);
-    conf.setLong(YarnConfiguration.NM_SLEEP_DELAY_BEFORE_SIGKILL_MS, 1000);
     super.setup();
   }
 
@@ -590,8 +589,9 @@ public class TestContainerLaunch extends BaseContainerManagerTest {
         AuxiliaryServiceHelper.getServiceDataFromEnv(serviceName, env));
   }
 
-  @Test
+  private void internalKillTest(boolean delayed) throws Exception {
-  public void testDelayedKill() throws Exception {
+    conf.setLong(YarnConfiguration.NM_SLEEP_DELAY_BEFORE_SIGKILL_MS,
+      delayed ? 1000 : 0);
     containerManager.start();
 
     // ////// Construct the Container-id
@@ -675,7 +675,8 @@ public class TestContainerLaunch extends BaseContainerManagerTest {
     BaseContainerManagerTest.waitForContainerState(containerManager, cId,
         ContainerState.COMPLETE);
 
-    // container stop sends a sigterm followed by a sigkill
+    // if delayed container stop sends a sigterm followed by a sigkill
+    // otherwise sigkill is sent immediately 
     GetContainerStatusesRequest gcsRequest =
         GetContainerStatusesRequest.newInstance(containerIds);
     
@@ -690,7 +691,7 @@ public class TestContainerLaunch extends BaseContainerManagerTest {
     // Windows, because the process is not notified when killed by winutils.
     // There is no way for the process to trap and respond.  Instead, we can
     // verify that the job object with ID matching container ID no longer exists.
-    if (Shell.WINDOWS) {
+    if (Shell.WINDOWS || !delayed) {
       Assert.assertFalse("Process is still alive!",
         DefaultContainerExecutor.containerIsAlive(cId.toString()));
     } else {
@@ -713,6 +714,16 @@ public class TestContainerLaunch extends BaseContainerManagerTest {
     }
   }
 
+  @Test
+  public void testDelayedKill() throws Exception {
+    internalKillTest(true);
+  }
+
+  @Test
+  public void testImmediateKill() throws Exception {
+    internalKillTest(false);
+  }
+
   @SuppressWarnings("rawtypes")
   @Test
   public void testCallFailureWithNullLocalizedResources() {