big-data/hadoop
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

YARN-966. Fixed ContainerLaunch to not fail quietly when there are no localized resources due to some other failure. Contributed by Zhijie Shen.

Browse Source 
git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1508688 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Vinod Kumar Vavilapalli 2013-07-30 23:58:17 +00:00
parent 48264f1994
commit 813efd25a1
5 changed files with 74 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
hadoop-yarn-project/CHANGES.txt
View File

@ -42,6 +42,9 @@ Release 2.1.1-beta - UNRELEASED
YARN-948. Changed ResourceManager to validate the release container list
before actually releasing them. (Omkar Vinit Joshi via vinodkv)
YARN-966. Fixed ContainerLaunch to not fail quietly when there are no
localized resources due to some other failure. (Zhijie Shen via vinodkv)
Release 2.1.0-beta - 2013-08-06
INCOMPATIBLE CHANGES

7
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/containermanager/container/ContainerImpl.java
View File

@ -335,8 +335,11 @@ public String getUser() {
public Map<Path,List<String>> getLocalizedResources() {
this.readLock.lock();
try {
assert ContainerState.LOCALIZED == getContainerState(); // TODO: FIXME!!
return localizedResources;
if (ContainerState.LOCALIZED == getContainerState()) {
return localizedResources;
} else {
return null;
}
} finally {
this.readLock.unlock();
}

11
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
View File

@ -53,6 +53,7 @@
import org.apache.hadoop.yarn.api.records.ContainerLaunchContext;
import org.apache.hadoop.yarn.conf.YarnConfiguration;
import org.apache.hadoop.yarn.event.Dispatcher;
import org.apache.hadoop.yarn.ipc.RPCUtil;
import org.apache.hadoop.yarn.server.nodemanager.ContainerExecutor;
import org.apache.hadoop.yarn.server.nodemanager.ContainerExecutor.DelayedProcessKiller;
import org.apache.hadoop.yarn.server.nodemanager.ContainerExecutor.ExitCode;
@ -120,14 +121,20 @@ public ContainerLaunch(Context context, Configuration configuration,
@SuppressWarnings("unchecked") // dispatcher not typed
public Integer call() {
final ContainerLaunchContext launchContext = container.getLaunchContext();
final Map<Path,List<String>> localResources =
container.getLocalizedResources();
Map<Path,List<String>> localResources = null;
ContainerId containerID = container.getContainerId();
String containerIdStr = ConverterUtils.toString(containerID);
final List<String> command = launchContext.getCommands();
int ret = -1;
try {
localResources = container.getLocalizedResources();
if (localResources == null) {
RPCUtil.getRemoteException(
"Unable to get local resources when Container " + containerID +
" is at " + container.getContainerState());
}
final String user = container.getUser();
// /////////////////////////// Variable expansion
// Before the container script gets written out.

22
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/test/java/org/apache/hadoop/yarn/server/nodemanager/containermanager/container/TestContainer.java
View File

@ -19,6 +19,8 @@
import static org.junit.Assert.assertEquals;
import static org.junit.Assert.assertTrue;
import static org.junit.Assert.assertNotNull;
import static org.junit.Assert.assertNull;
import static org.mockito.Matchers.argThat;
import static org.mockito.Mockito.mock;
import static org.mockito.Mockito.reset;
@ -124,6 +126,7 @@ public void testLocalizationLaunch() throws Exception {
// all resources should be localized
assertEquals(ContainerState.LOCALIZED, wc.c.getContainerState());
assertNotNull(wc.c.getLocalizedResources());
for (Entry<Path, List<String>> loc : wc.c.getLocalizedResources()
.entrySet()) {
assertEquals(localPaths.remove(loc.getKey()), loc.getValue());
@ -161,6 +164,7 @@ public void testExternalKill() throws Exception {
wc.containerKilledOnRequest();
assertEquals(ContainerState.EXITED_WITH_FAILURE,
wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
verifyCleanupCall(wc);
}
finally {
@ -183,6 +187,7 @@ public void testCleanupOnFailure() throws Exception {
wc.containerFailed(ExitCode.FORCE_KILLED.getExitCode());
assertEquals(ContainerState.EXITED_WITH_FAILURE,
wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
verifyCleanupCall(wc);
}
finally {
@ -205,7 +210,7 @@ public void testCleanupOnSuccess() throws Exception {
wc.containerSuccessful();
assertEquals(ContainerState.EXITED_WITH_SUCCESS,
wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
verifyCleanupCall(wc);
}
finally {
@ -228,10 +233,12 @@ public void testInitWhileDone() throws Exception {
wc.containerSuccessful();
wc.containerResourcesCleanup();
assertEquals(ContainerState.DONE, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
// Now in DONE, issue INIT
wc.initContainer();
// Verify still in DONE
assertEquals(ContainerState.DONE, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
verifyCleanupCall(wc);
}
finally {
@ -255,10 +262,12 @@ public void testLocalizationFailureAtDone() throws Exception {
wc.containerSuccessful();
wc.containerResourcesCleanup();
assertEquals(ContainerState.DONE, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
// Now in DONE, issue RESOURCE_FAILED as done by LocalizeRunner
wc.resourceFailedContainer();
// Verify still in DONE
assertEquals(ContainerState.DONE, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
verifyCleanupCall(wc);
} finally {
if (wc != null) {
@ -279,6 +288,7 @@ public void testCleanupOnKillRequest() throws Exception {
reset(wc.localizerBus);
wc.killContainer();
assertEquals(ContainerState.KILLING, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
wc.containerKilledOnRequest();
verifyCleanupCall(wc);
@ -297,8 +307,10 @@ public void testKillOnLocalizationFailed() throws Exception {
wc.initContainer();
wc.failLocalizeResources(wc.getLocalResourceCount());
assertEquals(ContainerState.LOCALIZATION_FAILED, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
wc.killContainer();
assertEquals(ContainerState.LOCALIZATION_FAILED, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
verifyCleanupCall(wc);
} finally {
if (wc != null) {
@ -319,8 +331,10 @@ public void testResourceLocalizedOnLocalizationFailed() throws Exception {
}
wc.failLocalizeResources(failCount);
assertEquals(ContainerState.LOCALIZATION_FAILED, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
wc.localizeResourcesFromInvalidState(failCount);
assertEquals(ContainerState.LOCALIZATION_FAILED, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
verifyCleanupCall(wc);
Assert.assertTrue(wc.getDiagnostics().contains(FAKE_LOCALIZATION_ERROR));
} finally {
@ -342,8 +356,10 @@ public void testResourceFailedOnLocalizationFailed() throws Exception {
String key2 = lRsrcKeys.next();
wc.failLocalizeSpecificResource(key1);
assertEquals(ContainerState.LOCALIZATION_FAILED, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
wc.failLocalizeSpecificResource(key2);
assertEquals(ContainerState.LOCALIZATION_FAILED, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
verifyCleanupCall(wc);
} finally {
if (wc != null) {
@ -363,8 +379,10 @@ public void testResourceFailedOnKilling() throws Exception {
String key1 = lRsrcKeys.next();
wc.killContainer();
assertEquals(ContainerState.KILLING, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
wc.failLocalizeSpecificResource(key1);
assertEquals(ContainerState.KILLING, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
verifyCleanupCall(wc);
} finally {
if (wc != null) {
@ -425,8 +443,10 @@ public void testLaunchAfterKillRequest() throws Exception {
wc.localizeResources();
wc.killContainer();
assertEquals(ContainerState.KILLING, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
wc.launchContainer();
assertEquals(ContainerState.KILLING, wc.c.getContainerState());
assertNull(wc.c.getLocalizedResources());
wc.containerKilledOnRequest();
verifyCleanupCall(wc);
} finally {

36
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
View File

@ -19,6 +19,8 @@
package org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher;
import static org.junit.Assert.assertEquals;
import static org.mockito.Mockito.mock;
import static org.mockito.Mockito.when;
import java.io.BufferedReader;
import java.io.File;
@ -28,12 +30,14 @@
import java.io.PrintWriter;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collections;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import junit.framework.Assert;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileUtil;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.fs.UnsupportedFileSystemException;
@ -59,10 +63,16 @@
import org.apache.hadoop.yarn.api.records.Token;
import org.apache.hadoop.yarn.api.records.URL;
import org.apache.hadoop.yarn.conf.YarnConfiguration;
import org.apache.hadoop.yarn.event.Dispatcher;
import org.apache.hadoop.yarn.event.Event;
import org.apache.hadoop.yarn.event.EventHandler;
import org.apache.hadoop.yarn.security.ContainerTokenIdentifier;
import org.apache.hadoop.yarn.server.nodemanager.ContainerExecutor.ExitCode;
import org.apache.hadoop.yarn.server.nodemanager.DefaultContainerExecutor;
import org.apache.hadoop.yarn.server.nodemanager.containermanager.BaseContainerManagerTest;
import org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container;
import org.apache.hadoop.yarn.server.nodemanager.containermanager.container.ContainerEventType;
import org.apache.hadoop.yarn.server.nodemanager.containermanager.container.ContainerExitEvent;
import org.apache.hadoop.yarn.server.nodemanager.containermanager.localizer.ContainerLocalizer;
import org.apache.hadoop.yarn.server.utils.BuilderUtils;
import org.apache.hadoop.yarn.util.ConverterUtils;
@ -617,6 +627,32 @@ public void testDelayedKill() throws Exception {
}
}
@SuppressWarnings("rawtypes")
@Test
public void testCallFailureWithNullLocalizedResources() {
Container container = mock(Container.class);
when(container.getContainerId()).thenReturn(ContainerId.newInstance(
ApplicationAttemptId.newInstance(ApplicationId.newInstance(
System.currentTimeMillis(), 1), 1), 1));
ContainerLaunchContext clc = mock(ContainerLaunchContext.class);
when(clc.getCommands()).thenReturn(Collections.<String>emptyList());
when(container.getLaunchContext()).thenReturn(clc);
when(container.getLocalizedResources()).thenReturn(null);
Dispatcher dispatcher = mock(Dispatcher.class);
EventHandler eventHandler = new EventHandler() {
public void handle(Event event) {
Assert.assertTrue(event instanceof ContainerExitEvent);
ContainerExitEvent exitEvent = (ContainerExitEvent) event;
Assert.assertEquals(ContainerEventType.CONTAINER_EXITED_WITH_FAILURE,
exitEvent.getType());
}
};
when(dispatcher.getEventHandler()).thenReturn(eventHandler);
ContainerLaunch launch = new ContainerLaunch(context, new Configuration(),
dispatcher, exec, null, container, dirsHandler);
launch.call();
}
protected Token createContainerToken(ContainerId cId) throws InvalidToken {
Resource r = BuilderUtils.newResource(1024, 1);
ContainerTokenIdentifier containerTokenIdentifier =