Merge r1523750 from trunk to branch-2 for YARN-1027. Implement RMHAProtocolService (Karthik Kambatla via bikas)

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/branches/branch-2@1523753 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Bikas Saha 2013-09-16 18:40:28 +00:00
parent 73c69decf1
commit db17c6ac06
15 changed files with 383 additions and 17 deletions

View File

@ -14,6 +14,7 @@ Release 2.3.0 - UNRELEASED
YARN-905. Add state filters to nodes CLI (Wei Yan via Sandy Ryza)
YARN-1098. Separate out RM services into Always On and Active (Karthik
Kambatla via bikas)
YARN-1027. Implement RMHAProtocolService (Karthik Kambatla via bikas)
OPTIMIZATIONS

View File

@ -270,6 +270,11 @@ public class YarnConfiguration extends Configuration {
public static final String RECOVERY_ENABLED = RM_PREFIX + "recovery.enabled";
public static final boolean DEFAULT_RM_RECOVERY_ENABLED = false;
/** HA related configs */
public static final String RM_HA_PREFIX = RM_PREFIX + "ha.";
public static final String RM_HA_ENABLED = RM_HA_PREFIX + "enabled";
public static final boolean DEFAULT_RM_HA_ENABLED = false;
/** The class to use as the persistent store.*/
public static final String RM_STORE = RM_PREFIX + "store.class";

View File

@ -268,6 +268,14 @@
<!--value>hdfs://localhost:9000/rmstore</value-->
</property>
<property>
<description>Enable RM high-availability. When enabled, the RM starts
in the Standby mode by default, and transitions to the Active mode when
prompted to.</description>
<name>yarn.resourcemanager.ha.enabled</name>
<value>false</value>
</property>
<property>
<description>The maximum number of completed applications RM keeps. </description>
<name>yarn.resourcemanager.max-completed-applications</name>

View File

@ -196,7 +196,7 @@ public class ClientRMService extends AbstractService implements
ApplicationId getNewApplicationId() {
ApplicationId applicationId = org.apache.hadoop.yarn.server.utils.BuilderUtils
.newApplicationId(recordFactory, ResourceManager.clusterTimeStamp,
.newApplicationId(recordFactory, ResourceManager.getClusterTimeStamp(),
applicationCounter.incrementAndGet());
LOG.info("Allocated new applicationId: " + applicationId.getId());
return applicationId;

View File

@ -0,0 +1,153 @@
/**
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.hadoop.yarn.server.resourcemanager;
import com.google.common.annotations.VisibleForTesting;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.classification.InterfaceAudience;
import org.apache.hadoop.classification.InterfaceStability;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.ha.HAServiceProtocol;
import org.apache.hadoop.ha.HAServiceStatus;
import org.apache.hadoop.ha.HealthCheckFailedException;
import org.apache.hadoop.service.AbstractService;
import org.apache.hadoop.yarn.conf.YarnConfiguration;
import org.apache.hadoop.yarn.exceptions.YarnRuntimeException;
import java.io.IOException;
@InterfaceAudience.Private
@InterfaceStability.Unstable
public class RMHAProtocolService extends AbstractService implements
HAServiceProtocol {
private static final Log LOG = LogFactory.getLog(RMHAProtocolService.class);
private Configuration conf;
private ResourceManager rm;
@VisibleForTesting
protected HAServiceState haState = HAServiceState.INITIALIZING;
public RMHAProtocolService(ResourceManager resourceManager) {
super("RMHAProtocolService");
this.rm = resourceManager;
}
@Override
public synchronized void serviceInit(Configuration conf) throws Exception {
this.conf = conf;
rm.createAndInitActiveServices();
super.serviceInit(this.conf);
}
@Override
public synchronized void serviceStart() throws Exception {
boolean haEnabled = this.conf.getBoolean(YarnConfiguration.RM_HA_ENABLED,
YarnConfiguration.DEFAULT_RM_HA_ENABLED);
if (haEnabled) {
transitionToStandby(true);
} else {
transitionToActive();
}
super.serviceStart();
}
@Override
public synchronized void serviceStop() throws Exception {
transitionToStandby(false);
haState = HAServiceState.STOPPING;
super.serviceStop();
}
@Override
public synchronized void monitorHealth() throws HealthCheckFailedException {
if (haState == HAServiceState.ACTIVE && !rm.areActiveServicesRunning()) {
throw new HealthCheckFailedException(
"Active ResourceManager services are not running!");
}
}
private synchronized void transitionToActive() throws Exception {
if (haState == HAServiceState.ACTIVE) {
LOG.info("Already in active state");
return;
}
LOG.info("Transitioning to active");
rm.startActiveServices();
haState = HAServiceState.ACTIVE;
LOG.info("Transitioned to active");
}
@Override
public synchronized void transitionToActive(StateChangeRequestInfo reqInfo) {
// TODO (YARN-1177): When automatic failover is enabled,
// check if transition should be allowed for this request
try {
transitionToActive();
} catch (Exception e) {
LOG.error("Error when transitioning to Active mode", e);
throw new YarnRuntimeException(e);
}
}
private synchronized void transitionToStandby(boolean initialize)
throws Exception {
if (haState == HAServiceState.STANDBY) {
LOG.info("Already in standby state");
return;
}
LOG.info("Transitioning to standby");
if (haState == HAServiceState.ACTIVE) {
rm.stopActiveServices();
if (initialize) {
rm.createAndInitActiveServices();
}
}
haState = HAServiceState.STANDBY;
LOG.info("Transitioned to standby");
}
@Override
public synchronized void transitionToStandby(StateChangeRequestInfo reqInfo) {
// TODO (YARN-1177): When automatic failover is enabled,
// check if transition should be allowed for this request
try {
transitionToStandby(true);
} catch (Exception e) {
LOG.error("Error when transitioning to Standby mode", e);
throw new YarnRuntimeException(e);
}
}
@Override
public synchronized HAServiceStatus getServiceStatus() throws IOException {
HAServiceStatus ret = new HAServiceStatus(haState);
if (haState == HAServiceState.ACTIVE || haState == HAServiceState.STANDBY) {
ret.setReadyToBecomeActive();
} else {
ret.setNotReadyToBecomeActive("State is " + haState);
}
return ret;
}
}

View File

@ -105,7 +105,14 @@ public class ResourceManager extends CompositeService implements Recoverable {
public static final int SHUTDOWN_HOOK_PRIORITY = 30;
private static final Log LOG = LogFactory.getLog(ResourceManager.class);
public static final long clusterTimeStamp = System.currentTimeMillis();
private static long clusterTimeStamp = System.currentTimeMillis();
/**
* "Always On" services. Services that need to run always irrespective of
* the HA state of the RM.
*/
@VisibleForTesting
protected RMHAProtocolService haService;
/**
* "Active" services. Services that need to run only on the Active RM.
@ -155,14 +162,18 @@ public class ResourceManager extends CompositeService implements Recoverable {
public RMContext getRMContext() {
return this.rmContext;
}
public static long getClusterTimeStamp() {
return clusterTimeStamp;
}
@Override
protected void serviceInit(Configuration conf) throws Exception {
validateConfigs(conf);
this.conf = conf;
activeServices = new RMActiveServices();
addService(activeServices);
haService = new RMHAProtocolService(this);
addService(haService);
super.serviceInit(conf);
}
@ -470,6 +481,7 @@ public class ResourceManager extends CompositeService implements Recoverable {
LOG.error("Error closing store.", e);
}
}
super.serviceStop();
}
}
@ -708,6 +720,43 @@ public class ResourceManager extends CompositeService implements Recoverable {
webApp = builder.start(new RMWebApp(this));
}
/**
* Helper method to create and init {@link #activeServices}. This creates an
* instance of {@link RMActiveServices} and initializes it.
* @throws Exception
*/
void createAndInitActiveServices() throws Exception {
activeServices = new RMActiveServices();
activeServices.init(conf);
}
/**
* Helper method to start {@link #activeServices}.
* @throws Exception
*/
void startActiveServices() throws Exception {
if (activeServices != null) {
clusterTimeStamp = System.currentTimeMillis();
activeServices.start();
}
}
/**
* Helper method to stop {@link #activeServices}.
* @throws Exception
*/
void stopActiveServices() throws Exception {
if (activeServices != null) {
activeServices.stop();
activeServices = null;
}
}
@VisibleForTesting
protected boolean areActiveServicesRunning() {
return activeServices != null && activeServices.isInState(STATE.STARTED);
}
@Override
protected void serviceStart() throws Exception {
try {
@ -715,7 +764,6 @@ public class ResourceManager extends CompositeService implements Recoverable {
} catch(IOException ie) {
throw new YarnRuntimeException("Failed to login", ie);
}
super.serviceStart();
}

View File

@ -229,7 +229,7 @@ public class ResourceTrackerService extends AbstractService implements
+ ", assigned nodeId " + nodeId;
LOG.info(message);
response.setNodeAction(NodeAction.NORMAL);
response.setRMIdentifier(ResourceManager.clusterTimeStamp);
response.setRMIdentifier(ResourceManager.getClusterTimeStamp());
return response;
}

View File

@ -185,7 +185,7 @@ public class RMContainerTokenSecretManager extends
tokenIdentifier =
new ContainerTokenIdentifier(containerId, nodeId.toString(),
appSubmitter, capability, expiryTimeStamp, this.currentMasterKey
.getMasterKey().getKeyId(), ResourceManager.clusterTimeStamp);
.getMasterKey().getKeyId(), ResourceManager.getClusterTimeStamp());
password = this.createPassword(tokenIdentifier);
} finally {

View File

@ -108,7 +108,7 @@ public class AppInfo {
this.diagnostics = "";
}
this.finalStatus = app.getFinalApplicationStatus();
this.clusterId = ResourceManager.clusterTimeStamp;
this.clusterId = ResourceManager.getClusterTimeStamp();
if (hasAccess) {
this.startedTime = app.getStartTime();
this.finishedTime = app.getFinishTime();

View File

@ -44,7 +44,7 @@ public class ClusterInfo {
} // JAXB needs this
public ClusterInfo(ResourceManager rm) {
long ts = ResourceManager.clusterTimeStamp;
long ts = ResourceManager.getClusterTimeStamp();
this.id = ts;
this.state = rm.getServiceState();

View File

@ -0,0 +1,151 @@
/**
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.hadoop.yarn.server.resourcemanager;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.ha.HAServiceProtocol;
import org.apache.hadoop.ha.HAServiceProtocol.HAServiceState;
import org.apache.hadoop.ha.HAServiceProtocol.StateChangeRequestInfo;
import org.apache.hadoop.ha.HealthCheckFailedException;
import org.apache.hadoop.yarn.conf.YarnConfiguration;
import org.junit.Before;
import org.junit.Test;
import java.io.IOException;
import static org.junit.Assert.assertEquals;
import static org.junit.Assert.assertFalse;
import static org.junit.Assert.assertTrue;
import static org.junit.Assert.fail;
public class TestRMHA {
private Log LOG = LogFactory.getLog(TestRMHA.class);
private MockRM rm = null;
private static final String STATE_ERR =
"ResourceManager is in wrong HA state";
@Before
public void setUp() throws Exception {
Configuration conf = new YarnConfiguration();
conf.setBoolean(YarnConfiguration.RM_HA_ENABLED, true);
rm = new MockRM(conf);
rm.init(conf);
}
private void checkMonitorHealth() {
try {
rm.haService.monitorHealth();
} catch (HealthCheckFailedException e) {
fail("The RM is in bad health: it is Active, but the active services " +
"are not running");
}
}
private void checkStandbyRMFunctionality() throws IOException {
assertEquals(STATE_ERR, HAServiceState.STANDBY,
rm.haService.getServiceStatus().getState());
assertFalse("Active RM services are started",
rm.areActiveServicesRunning());
assertTrue("RM is not ready to become active",
rm.haService.getServiceStatus().isReadyToBecomeActive());
}
private void checkActiveRMFunctionality() throws IOException {
assertEquals(STATE_ERR, HAServiceState.ACTIVE,
rm.haService.getServiceStatus().getState());
assertTrue("Active RM services aren't started",
rm.areActiveServicesRunning());
assertTrue("RM is not ready to become active",
rm.haService.getServiceStatus().isReadyToBecomeActive());
try {
rm.getNewAppId();
rm.registerNode("127.0.0.1:0", 2048);
rm.submitApp(1024);
} catch (Exception e) {
fail("Unable to perform Active RM functions");
LOG.error("ActiveRM check failed", e);
}
}
/**
* Test to verify the following RM HA transitions to the following states.
* 1. Standby: Should be a no-op
* 2. Active: Active services should start
* 3. Active: Should be a no-op.
* While active, submit a couple of jobs
* 4. Standby: Active services should stop
* 5. Active: Active services should start
* 6. Stop the RM: All services should stop and RM should not be ready to
* become Active
*/
@Test (timeout = 30000)
public void testStartAndTransitions() throws IOException {
StateChangeRequestInfo requestInfo = new StateChangeRequestInfo(
HAServiceProtocol.RequestSource.REQUEST_BY_USER);
assertEquals(STATE_ERR, HAServiceState.INITIALIZING,
rm.haService.getServiceStatus().getState());
assertFalse("RM is ready to become active before being started",
rm.haService.getServiceStatus().isReadyToBecomeActive());
checkMonitorHealth();
rm.start();
checkMonitorHealth();
checkStandbyRMFunctionality();
// 1. Transition to Standby - must be a no-op
rm.haService.transitionToStandby(requestInfo);
checkMonitorHealth();
checkStandbyRMFunctionality();
// 2. Transition to active
rm.haService.transitionToActive(requestInfo);
checkMonitorHealth();
checkActiveRMFunctionality();
// 3. Transition to active - no-op
rm.haService.transitionToActive(requestInfo);
checkMonitorHealth();
checkActiveRMFunctionality();
// 4. Transition to standby
rm.haService.transitionToStandby(requestInfo);
checkMonitorHealth();
checkStandbyRMFunctionality();
// 5. Transition to active to check Active->Standby->Active works
rm.haService.transitionToActive(requestInfo);
checkMonitorHealth();
checkActiveRMFunctionality();
// 6. Stop the RM. All services should stop and RM should not be ready to
// become active
rm.stop();
assertEquals(STATE_ERR, HAServiceState.STOPPING,
rm.haService.getServiceStatus().getState());
assertFalse("RM is ready to become active even after it is stopped",
rm.haService.getServiceStatus().isReadyToBecomeActive());
assertFalse("Active RM services are started",
rm.areActiveServicesRunning());
checkMonitorHealth();
}
}

View File

@ -283,7 +283,7 @@ public class TestResourceTrackerService {
RegisterNodeManagerResponse response = nm.registerNode();
// Verify the RMIdentifier is correctly set in RegisterNodeManagerResponse
Assert.assertEquals(ResourceManager.clusterTimeStamp,
Assert.assertEquals(ResourceManager.getClusterTimeStamp(),
response.getRMIdentifier());
}

View File

@ -88,7 +88,7 @@ public class TestApplicationMasterService {
ContainerTokenIdentifier tokenId =
BuilderUtils.newContainerTokenIdentifier(allocatedContainer
.getContainerToken());
Assert.assertEquals(MockRM.clusterTimeStamp, tokenId.getRMIdentifer());
Assert.assertEquals(MockRM.getClusterTimeStamp(), tokenId.getRMIdentifer());
rm.stop();
}

View File

@ -295,10 +295,10 @@ public class TestRMWebServices extends JerseyTest {
String hadoopVersion, String resourceManagerVersionBuiltOn,
String resourceManagerBuildVersion, String resourceManagerVersion) {
assertEquals("clusterId doesn't match: ", ResourceManager.clusterTimeStamp,
clusterid);
assertEquals("startedOn doesn't match: ", ResourceManager.clusterTimeStamp,
startedon);
assertEquals("clusterId doesn't match: ",
ResourceManager.getClusterTimeStamp(), clusterid);
assertEquals("startedOn doesn't match: ",
ResourceManager.getClusterTimeStamp(), startedon);
assertTrue("stated doesn't match: " + state,
state.matches(STATE.INITED.toString()));

View File

@ -1181,8 +1181,8 @@ public class TestRMWebServicesApps extends JerseyTest {
trackingUI);
WebServicesTestUtils.checkStringMatch("diagnostics", app.getDiagnostics()
.toString(), diagnostics);
assertEquals("clusterId doesn't match", ResourceManager.clusterTimeStamp,
clusterId);
assertEquals("clusterId doesn't match",
ResourceManager.getClusterTimeStamp(), clusterId);
assertEquals("startedTime doesn't match", app.getStartTime(), startedTime);
assertEquals("finishedTime doesn't match", app.getFinishTime(),
finishedTime);