Apache
/
hadoop
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

YARN-8360. Improve YARN service restart policy and node manager auto restart policy. 

Contributed by Suma Shivaprasad

(cherry picked from commit 84d7bf1eef)
This commit is contained in:
Eric Yang 2018-07-23 12:57:01 -04:00
parent 4f2a129f2e
commit e665c0a9dd
7 changed files with 101 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-applications/hadoop-yarn-services/hadoop-yarn-services-core/src/main/java/org/apache/hadoop/yarn/service/component/AlwaysRestartPolicy.java
View File

@ -79,4 +79,9 @@ public final class AlwaysRestartPolicy implements ComponentRestartPolicy {
@Override public boolean shouldTerminate(Component component) {
return false;
}
@Override public boolean allowContainerRetriesForInstance(
ComponentInstance componentInstance) {
return true;
}
}

2
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-applications/hadoop-yarn-services/hadoop-yarn-services-core/src/main/java/org/apache/hadoop/yarn/service/component/ComponentRestartPolicy.java
View File

@ -42,4 +42,6 @@ public interface ComponentRestartPolicy {
boolean shouldTerminate(Component component);
boolean allowContainerRetriesForInstance(ComponentInstance componentInstance);
}

5
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-applications/hadoop-yarn-services/hadoop-yarn-services-core/src/main/java/org/apache/hadoop/yarn/service/component/NeverRestartPolicy.java
View File

@ -79,4 +79,9 @@ public final class NeverRestartPolicy implements ComponentRestartPolicy {
}
return true;
}
@Override public boolean allowContainerRetriesForInstance(
ComponentInstance componentInstance) {
return false;
}
}

5
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-applications/hadoop-yarn-services/hadoop-yarn-services-core/src/main/java/org/apache/hadoop/yarn/service/component/OnFailureRestartPolicy.java
View File

@ -84,4 +84,9 @@ public final class OnFailureRestartPolicy implements ComponentRestartPolicy {
}
return true;
}
@Override public boolean allowContainerRetriesForInstance(
ComponentInstance componentInstance) {
return true;
}
}

25
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-applications/hadoop-yarn-services/hadoop-yarn-services-core/src/main/java/org/apache/hadoop/yarn/service/provider/AbstractProviderService.java
View File

@ -22,6 +22,7 @@ import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.yarn.api.ApplicationConstants;
import org.apache.hadoop.yarn.api.records.Container;
import org.apache.hadoop.yarn.service.api.records.Service;
import org.apache.hadoop.yarn.service.component.ComponentRestartPolicy;
import org.apache.hadoop.yarn.service.conf.YarnServiceConf;
import org.apache.hadoop.yarn.service.conf.YarnServiceConstants;
import org.apache.hadoop.yarn.service.containerlaunch.ContainerLaunchService;
@ -116,19 +117,23 @@ public abstract class AbstractProviderService implements ProviderService,
public void buildContainerRetry(AbstractLauncher launcher,
Configuration yarnConf,
ContainerLaunchService.ComponentLaunchContext compLaunchContext) {
ContainerLaunchService.ComponentLaunchContext compLaunchContext,
ComponentInstance instance) {
// By default retry forever every 30 seconds
launcher.setRetryContext(
YarnServiceConf.getInt(CONTAINER_RETRY_MAX,
DEFAULT_CONTAINER_RETRY_MAX,
compLaunchContext.getConfiguration(), yarnConf),
YarnServiceConf.getInt(CONTAINER_RETRY_INTERVAL,
DEFAULT_CONTAINER_RETRY_INTERVAL,
compLaunchContext.getConfiguration(), yarnConf),
YarnServiceConf.getLong(CONTAINER_FAILURES_VALIDITY_INTERVAL,
ComponentRestartPolicy restartPolicy = instance.getComponent()
.getRestartPolicyHandler();
if (restartPolicy.allowContainerRetriesForInstance(instance)) {
launcher.setRetryContext(YarnServiceConf
.getInt(CONTAINER_RETRY_MAX, DEFAULT_CONTAINER_RETRY_MAX,
compLaunchContext.getConfiguration(), yarnConf), YarnServiceConf
.getInt(CONTAINER_RETRY_INTERVAL, DEFAULT_CONTAINER_RETRY_INTERVAL,
compLaunchContext.getConfiguration(), yarnConf), YarnServiceConf
.getLong(CONTAINER_FAILURES_VALIDITY_INTERVAL,
DEFAULT_CONTAINER_FAILURES_VALIDITY_INTERVAL,
compLaunchContext.getConfiguration(), yarnConf));
}
}
public void buildContainerLaunchContext(AbstractLauncher launcher,
Service service, ComponentInstance instance,
@ -161,6 +166,6 @@ public abstract class AbstractProviderService implements ProviderService,
yarnConf, container, compLaunchContext, tokensForSubstitution);
// Setup container retry settings
buildContainerRetry(launcher, yarnConf, compLaunchContext);
buildContainerRetry(launcher, yarnConf, compLaunchContext, instance);
}
}

2
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-applications/hadoop-yarn-services/hadoop-yarn-services-core/src/test/java/org/apache/hadoop/yarn/service/ServiceTestUtils.java
View File

@ -115,7 +115,7 @@ public class ServiceTestUtils {
exampleApp.setName(serviceName);
exampleApp.setVersion("v1");
exampleApp.addComponent(
createComponent("terminating-comp1", 2, "sleep " + "1000",
createComponent("terminating-comp1", 2, "sleep 1000",
Component.RestartPolicyEnum.NEVER, null));
exampleApp.addComponent(
createComponent("terminating-comp2", 2, "sleep 1000",

66
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-applications/hadoop-yarn-services/hadoop-yarn-services-core/src/test/java/org/apache/hadoop/yarn/service/containerlaunch/TestAbstractLauncher.java
View File

@ -19,13 +19,33 @@
package org.apache.hadoop.yarn.service.containerlaunch;
import org.apache.hadoop.yarn.service.ServiceContext;
import org.apache.hadoop.yarn.service.api.records.Configuration;
import org.apache.hadoop.yarn.service.component.AlwaysRestartPolicy;
import org.apache.hadoop.yarn.service.component.Component;
import org.apache.hadoop.yarn.service.component.NeverRestartPolicy;
import org.apache.hadoop.yarn.service.component.OnFailureRestartPolicy;
import org.apache.hadoop.yarn.service.component.instance.ComponentInstance;
import org.apache.hadoop.yarn.service.provider.defaultImpl
.DefaultProviderService;
import org.junit.Assert;
import org.junit.Before;
import org.junit.Test;
import java.io.IOException;
import static org.apache.hadoop.fi.FiConfig.getConfig;
import static org.apache.hadoop.yarn.service.conf.YarnServiceConf
.DEFAULT_CONTAINER_FAILURES_VALIDITY_INTERVAL;
import static org.apache.hadoop.yarn.service.conf.YarnServiceConf
.DEFAULT_CONTAINER_RETRY_INTERVAL;
import static org.apache.hadoop.yarn.service.conf.YarnServiceConf
.DEFAULT_CONTAINER_RETRY_MAX;
import static org.mockito.Mockito.mock;
import static org.mockito.Mockito.reset;
import static org.mockito.Mockito.times;
import static org.mockito.Mockito.verify;
import static org.mockito.Mockito.verifyZeroInteractions;
import static org.mockito.Mockito.when;
/**
* Tests for {@link AbstractLauncher}.
@ -51,4 +71,50 @@ public class TestAbstractLauncher {
Assert.assertEquals("s1:t1:ro,s2:t2:ro", dockerContainerMounts);
}
@Test
public void testContainerRetries() throws Exception {
DefaultProviderService providerService = new DefaultProviderService();
AbstractLauncher mockLauncher = mock(AbstractLauncher.class);
ContainerLaunchService.ComponentLaunchContext componentLaunchContext =
mock(ContainerLaunchService.ComponentLaunchContext.class);
ComponentInstance componentInstance = mock(ComponentInstance.class);
//Never Restart Policy
Component component = mock(Component.class);
when(componentInstance.getComponent()).thenReturn(component);
when(component.getRestartPolicyHandler()).thenReturn(NeverRestartPolicy
.getInstance());
providerService.buildContainerRetry(mockLauncher, getConfig(),
componentLaunchContext, componentInstance);
verifyZeroInteractions(mockLauncher);
//OnFailure restart policy
when(component.getRestartPolicyHandler()).thenReturn(OnFailureRestartPolicy
.getInstance());
when(componentLaunchContext.getConfiguration()).thenReturn(new
Configuration());
providerService.buildContainerRetry(mockLauncher, getConfig(),
componentLaunchContext, componentInstance);
verify(mockLauncher).setRetryContext(DEFAULT_CONTAINER_RETRY_MAX,
DEFAULT_CONTAINER_RETRY_INTERVAL,
DEFAULT_CONTAINER_FAILURES_VALIDITY_INTERVAL);
reset(mockLauncher);
//Always restart policy
when(component.getRestartPolicyHandler()).thenReturn(AlwaysRestartPolicy
.getInstance());
providerService.buildContainerRetry(mockLauncher, getConfig(),
componentLaunchContext, componentInstance);
verify(mockLauncher).setRetryContext(DEFAULT_CONTAINER_RETRY_MAX,
DEFAULT_CONTAINER_RETRY_INTERVAL,
DEFAULT_CONTAINER_FAILURES_VALIDITY_INTERVAL);
}
}