/**
* 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.rmnode;
import java.util.ArrayList;
import java.util.EnumSet;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.TreeSet;
import java.util.concurrent.locks.ReentrantReadWriteLock;
import java.util.concurrent.locks.ReentrantReadWriteLock.ReadLock;
import java.util.concurrent.locks.ReentrantReadWriteLock.WriteLock;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.classification.InterfaceAudience.Private;
import org.apache.hadoop.classification.InterfaceStability.Unstable;
import org.apache.hadoop.net.Node;
import org.apache.hadoop.yarn.api.records.ApplicationId;
import org.apache.hadoop.yarn.api.records.ContainerId;
import org.apache.hadoop.yarn.api.records.ContainerState;
import org.apache.hadoop.yarn.api.records.ContainerStatus;
import org.apache.hadoop.yarn.api.records.NodeHealthStatus;
import org.apache.hadoop.yarn.api.records.NodeId;
import org.apache.hadoop.yarn.api.records.Resource;
import org.apache.hadoop.yarn.event.EventHandler;
import org.apache.hadoop.yarn.factories.RecordFactory;
import org.apache.hadoop.yarn.factory.providers.RecordFactoryProvider;
import org.apache.hadoop.yarn.server.api.records.HeartbeatResponse;
import org.apache.hadoop.yarn.server.resourcemanager.ClusterMetrics;
import org.apache.hadoop.yarn.server.resourcemanager.RMContext;
import org.apache.hadoop.yarn.server.resourcemanager.scheduler.event.NodeAddedSchedulerEvent;
import org.apache.hadoop.yarn.server.resourcemanager.scheduler.event.NodeRemovedSchedulerEvent;
import org.apache.hadoop.yarn.server.resourcemanager.scheduler.event.NodeUpdateSchedulerEvent;
import org.apache.hadoop.yarn.state.InvalidStateTransitonException;
import org.apache.hadoop.yarn.state.MultipleArcTransition;
import org.apache.hadoop.yarn.state.SingleArcTransition;
import org.apache.hadoop.yarn.state.StateMachine;
import org.apache.hadoop.yarn.state.StateMachineFactory;
import org.apache.hadoop.yarn.util.BuilderUtils.ContainerIdComparator;
/**
* This class is used to keep track of all the applications/containers
* running on a node.
*
*/
@Private
@Unstable
@SuppressWarnings("unchecked")
public class RMNodeImpl implements RMNode, EventHandler<RMNodeEvent> {
private static final Log LOG = LogFactory.getLog(RMNodeImpl.class);
private static final RecordFactory recordFactory = RecordFactoryProvider
.getRecordFactory(null);
private final ReadLock readLock;
private final WriteLock writeLock;
private final NodeId nodeId;
private final RMContext context;
private final String hostName;
private final int commandPort;
private final int httpPort;
private final String nodeAddress; // The containerManager address
private final String httpAddress;
private final Resource totalCapability;
private final Node node;
private final NodeHealthStatus nodeHealthStatus = recordFactory
.newRecordInstance(NodeHealthStatus.class);
/* set of containers that have just launched */
private final Map<ContainerId, ContainerStatus> justLaunchedContainers =
new HashMap<ContainerId, ContainerStatus>();
/* set of containers that need to be cleaned */
private final Set<ContainerId> containersToClean = new TreeSet<ContainerId>(
new ContainerIdComparator());
/* the list of applications that have finished and need to be purged */
private final List<ApplicationId> finishedApplications = new ArrayList<ApplicationId>();
private HeartbeatResponse latestHeartBeatResponse = recordFactory
.newRecordInstance(HeartbeatResponse.class);
private static final StateMachineFactory<RMNodeImpl,
RMNodeState,
RMNodeEventType,
RMNodeEvent> stateMachineFactory
= new StateMachineFactory<RMNodeImpl,
RMNodeState,
RMNodeEventType,
RMNodeEvent>(RMNodeState.NEW)
//Transitions from NEW state
.addTransition(RMNodeState.NEW, RMNodeState.RUNNING,
RMNodeEventType.STARTED, new AddNodeTransition())
//Transitions from RUNNING state
.addTransition(RMNodeState.RUNNING,
EnumSet.of(RMNodeState.RUNNING, RMNodeState.UNHEALTHY),
RMNodeEventType.STATUS_UPDATE, new StatusUpdateWhenHealthyTransition())
.addTransition(RMNodeState.RUNNING, RMNodeState.DECOMMISSIONED,
RMNodeEventType.DECOMMISSION,
new DeactivateNodeTransition(RMNodeState.DECOMMISSIONED))
.addTransition(RMNodeState.RUNNING, RMNodeState.LOST,
RMNodeEventType.EXPIRE,
new DeactivateNodeTransition(RMNodeState.LOST))
.addTransition(RMNodeState.RUNNING, RMNodeState.REBOOTED,
RMNodeEventType.REBOOTING,
new DeactivateNodeTransition(RMNodeState.REBOOTED))
.addTransition(RMNodeState.RUNNING, RMNodeState.RUNNING,
RMNodeEventType.CLEANUP_APP, new CleanUpAppTransition())
.addTransition(RMNodeState.RUNNING, RMNodeState.RUNNING,
RMNodeEventType.CLEANUP_CONTAINER, new CleanUpContainerTransition())
.addTransition(RMNodeState.RUNNING, RMNodeState.RUNNING,
RMNodeEventType.RECONNECTED, new ReconnectNodeTransition())
//Transitions from UNHEALTHY state
.addTransition(RMNodeState.UNHEALTHY,
EnumSet.of(RMNodeState.UNHEALTHY, RMNodeState.RUNNING),
RMNodeEventType.STATUS_UPDATE, new StatusUpdateWhenUnHealthyTransition())
.addTransition(RMNodeState.UNHEALTHY, RMNodeState.UNHEALTHY,
RMNodeEventType.RECONNECTED, new ReconnectNodeTransition())
// create the topology tables
.installTopology();
private final StateMachine<RMNodeState, RMNodeEventType,
RMNodeEvent> stateMachine;
public RMNodeImpl(NodeId nodeId, RMContext context, String hostName,
int cmPort, int httpPort, Node node, Resource capability) {
this.nodeId = nodeId;
this.context = context;
this.hostName = hostName;
this.commandPort = cmPort;
this.httpPort = httpPort;
this.totalCapability = capability;
this.nodeAddress = hostName + ":" + cmPort;
this.httpAddress = hostName + ":" + httpPort;
this.node = node;
this.nodeHealthStatus.setIsNodeHealthy(true);
this.nodeHealthStatus.setHealthReport("Healthy");
this.nodeHealthStatus.setLastHealthReportTime(System.currentTimeMillis());
this.latestHeartBeatResponse.setResponseId(0);
ReentrantReadWriteLock lock = new ReentrantReadWriteLock();
this.readLock = lock.readLock();
this.writeLock = lock.writeLock();
this.stateMachine = stateMachineFactory.make(this);
}
@Override
public String toString() {
return this.nodeId.toString();
}
@Override
public String getHostName() {
return hostName;
}
@Override
public int getCommandPort() {
return commandPort;
}
@Override
public int getHttpPort() {
return httpPort;
}
@Override
public NodeId getNodeID() {
return this.nodeId;
}
@Override
public String getNodeAddress() {
return this.nodeAddress;
}
@Override
public String getHttpAddress() {
return this.httpAddress;
}
@Override
public Resource getTotalCapability() {
return this.totalCapability;
}
@Override
public String getRackName() {
return node.getNetworkLocation();
}
@Override
public Node getNode() {
return this.node;
}
@Override
public NodeHealthStatus getNodeHealthStatus() {
this.readLock.lock();
try {
return this.nodeHealthStatus;
} finally {
this.readLock.unlock();
}
}
private void setNodeHealthStatus(NodeHealthStatus status)
{
this.writeLock.lock();
try {
this.nodeHealthStatus.setHealthReport(status.getHealthReport());
this.nodeHealthStatus.setIsNodeHealthy(status.getIsNodeHealthy());
this.nodeHealthStatus.setLastHealthReportTime(status.getLastHealthReportTime());
} finally {
this.writeLock.unlock();
}
}
@Override
public RMNodeState getState() {
this.readLock.lock();
try {
return this.stateMachine.getCurrentState();
} finally {
this.readLock.unlock();
}
}
@Override
public List<ApplicationId> getAppsToCleanup() {
this.readLock.lock();
try {
return new ArrayList<ApplicationId>(this.finishedApplications);
} finally {
this.readLock.unlock();
}
}
@Override
public List<ContainerId> getContainersToCleanUp() {
this.readLock.lock();
try {
return new ArrayList<ContainerId>(this.containersToClean);
} finally {
this.readLock.unlock();
}
};
@Override
public HeartbeatResponse getLastHeartBeatResponse() {
this.readLock.lock();
try {
return this.latestHeartBeatResponse;
} finally {
this.readLock.unlock();
}
}
public void handle(RMNodeEvent event) {
LOG.debug("Processing " + event.getNodeId() + " of type " + event.getType());
try {
writeLock.lock();
RMNodeState oldState = getState();
try {
stateMachine.doTransition(event.getType(), event);
} catch (InvalidStateTransitonException e) {
LOG.error("Can't handle this event at current state", e);
LOG.error("Invalid event " + event.getType() +
" on Node " + this.nodeId);
}
if (oldState != getState()) {
LOG.info(nodeId + " Node Transitioned from " + oldState + " to "
+ getState());
}
}
finally {
writeLock.unlock();
}
}
private void updateMetricsForRejoinedNode(RMNodeState previousNodeState) {
ClusterMetrics metrics = ClusterMetrics.getMetrics();
metrics.incrNumActiveNodes();
switch (previousNodeState) {
case LOST:
metrics.decrNumLostNMs();
break;
case REBOOTED:
metrics.decrNumRebootedNMs();
break;
case DECOMMISSIONED:
metrics.decrDecommisionedNMs();
break;
case UNHEALTHY:
metrics.decrNumUnhealthyNMs();
break;
}
}
private void updateMetricsForDeactivatedNode(RMNodeState finalState) {
ClusterMetrics metrics = ClusterMetrics.getMetrics();
metrics.decrNumActiveNodes();
switch (finalState) {
case DECOMMISSIONED:
metrics.incrDecommisionedNMs();
break;
case LOST:
metrics.incrNumLostNMs();
break;
case REBOOTED:
metrics.incrNumRebootedNMs();
break;
case UNHEALTHY:
metrics.incrNumUnhealthyNMs();
break;
}
}
public static class AddNodeTransition implements
SingleArcTransition<RMNodeImpl, RMNodeEvent> {
@Override
public void transition(RMNodeImpl rmNode, RMNodeEvent event) {
// Inform the scheduler
rmNode.context.getDispatcher().getEventHandler().handle(
new NodeAddedSchedulerEvent(rmNode));
String host = rmNode.nodeId.getHost();
if (rmNode.context.getInactiveRMNodes().containsKey(host)) {
// Old node rejoining
RMNode previouRMNode = rmNode.context.getInactiveRMNodes().get(host);
rmNode.context.getInactiveRMNodes().remove(host);
rmNode.updateMetricsForRejoinedNode(previouRMNode.getState());
} else {
// Increment activeNodes explicitly because this is a new node.
ClusterMetrics.getMetrics().incrNumActiveNodes();
}
}
}
public static class ReconnectNodeTransition implements
SingleArcTransition<RMNodeImpl, RMNodeEvent> {
@Override
public void transition(RMNodeImpl rmNode, RMNodeEvent event) {
// Kill containers since node is rejoining.
rmNode.context.getDispatcher().getEventHandler().handle(
new NodeRemovedSchedulerEvent(rmNode));
RMNode newNode = ((RMNodeReconnectEvent)event).getReconnectedNode();
if (rmNode.getTotalCapability().equals(newNode.getTotalCapability())
&& rmNode.getHttpPort() == newNode.getHttpPort()) {
// Reset heartbeat ID since node just restarted.
rmNode.getLastHeartBeatResponse().setResponseId(0);
rmNode.context.getDispatcher().getEventHandler().handle(
new NodeAddedSchedulerEvent(rmNode));
} else {
// Reconnected node differs, so replace old node and start new node
switch (rmNode.getState()) {
case RUNNING:
ClusterMetrics.getMetrics().decrNumActiveNodes();
break;
case UNHEALTHY:
ClusterMetrics.getMetrics().decrNumUnhealthyNMs();
break;
}
rmNode.context.getRMNodes().put(newNode.getNodeID(), newNode);
rmNode.context.getDispatcher().getEventHandler().handle(
new RMNodeEvent(newNode.getNodeID(), RMNodeEventType.STARTED));
}
}
}
public static class CleanUpAppTransition
implements SingleArcTransition<RMNodeImpl, RMNodeEvent> {
@Override
public void transition(RMNodeImpl rmNode, RMNodeEvent event) {
rmNode.finishedApplications.add(((
RMNodeCleanAppEvent) event).getAppId());
}
}
public static class CleanUpContainerTransition implements
SingleArcTransition<RMNodeImpl, RMNodeEvent> {
@Override
public void transition(RMNodeImpl rmNode, RMNodeEvent event) {
rmNode.containersToClean.add(((
RMNodeCleanContainerEvent) event).getContainerId());
}
}
public static class DeactivateNodeTransition
implements SingleArcTransition<RMNodeImpl, RMNodeEvent> {
private final RMNodeState finalState;
public DeactivateNodeTransition(RMNodeState finalState) {
this.finalState = finalState;
}
@Override
public void transition(RMNodeImpl rmNode, RMNodeEvent event) {
// Inform the scheduler
rmNode.context.getDispatcher().getEventHandler().handle(
new NodeRemovedSchedulerEvent(rmNode));
// Deactivate the node
rmNode.context.getRMNodes().remove(rmNode.nodeId);
LOG.info("Deactivating Node " + rmNode.nodeId + " as it is now "
+ finalState);
rmNode.context.getInactiveRMNodes().put(rmNode.nodeId.getHost(), rmNode);
//Update the metrics
rmNode.updateMetricsForDeactivatedNode(finalState);
}
}
public static class StatusUpdateWhenHealthyTransition implements
MultipleArcTransition<RMNodeImpl, RMNodeEvent, RMNodeState> {
@Override
public RMNodeState transition(RMNodeImpl rmNode, RMNodeEvent event) {
RMNodeStatusEvent statusEvent = (RMNodeStatusEvent) event;
// Switch the last heartbeatresponse.
rmNode.latestHeartBeatResponse = statusEvent.getLatestResponse();
NodeHealthStatus remoteNodeHealthStatus =
statusEvent.getNodeHealthStatus();
rmNode.setNodeHealthStatus(remoteNodeHealthStatus);
if (!remoteNodeHealthStatus.getIsNodeHealthy()) {
// Inform the scheduler
rmNode.context.getDispatcher().getEventHandler().handle(
new NodeRemovedSchedulerEvent(rmNode));
// Update metrics
rmNode.updateMetricsForDeactivatedNode(RMNodeState.UNHEALTHY);
return RMNodeState.UNHEALTHY;
}
// Filter the map to only obtain just launched containers and finished
// containers.
List<ContainerStatus> newlyLaunchedContainers =
new ArrayList<ContainerStatus>();
List<ContainerStatus> completedContainers =
new ArrayList<ContainerStatus>();
for (ContainerStatus remoteContainer : statusEvent.getContainers()) {
ContainerId containerId = remoteContainer.getContainerId();
// Don't bother with containers already scheduled for cleanup, or for
// applications already killed. The scheduler doens't need to know any
// more about this container
if (rmNode.containersToClean.contains(containerId)) {
LOG.info("Container " + containerId + " already scheduled for " +
"cleanup, no further processing");
continue;
}
if (rmNode.finishedApplications.contains(containerId
.getApplicationAttemptId().getApplicationId())) {
LOG.info("Container " + containerId
+ " belongs to an application that is already killed,"
+ " no further processing");
continue;
}
// Process running containers
if (remoteContainer.getState() == ContainerState.RUNNING) {
if (!rmNode.justLaunchedContainers.containsKey(containerId)) {
// Just launched container. RM knows about it the first time.
rmNode.justLaunchedContainers.put(containerId, remoteContainer);
newlyLaunchedContainers.add(remoteContainer);
}
} else {
// A finished container
rmNode.justLaunchedContainers.remove(containerId);
completedContainers.add(remoteContainer);
}
}
rmNode.context.getDispatcher().getEventHandler().handle(
new NodeUpdateSchedulerEvent(rmNode, newlyLaunchedContainers,
completedContainers));
rmNode.context.getDelegationTokenRenewer().updateKeepAliveApplications(
statusEvent.getKeepAliveAppIds());
// HeartBeat processing from our end is done, as node pulls the following
// lists before sending status-updates. Clear data-structures
rmNode.containersToClean.clear();
rmNode.finishedApplications.clear();
return RMNodeState.RUNNING;
}
}
public static class StatusUpdateWhenUnHealthyTransition implements
MultipleArcTransition<RMNodeImpl, RMNodeEvent, RMNodeState> {
@Override
public RMNodeState transition(RMNodeImpl rmNode, RMNodeEvent event) {
RMNodeStatusEvent statusEvent = (RMNodeStatusEvent) event;
// Switch the last heartbeatresponse.
rmNode.latestHeartBeatResponse = statusEvent.getLatestResponse();
NodeHealthStatus remoteNodeHealthStatus = statusEvent.getNodeHealthStatus();
rmNode.setNodeHealthStatus(remoteNodeHealthStatus);
if (remoteNodeHealthStatus.getIsNodeHealthy()) {
rmNode.context.getDispatcher().getEventHandler().handle(
new NodeAddedSchedulerEvent(rmNode));
// Update metrics
rmNode.updateMetricsForRejoinedNode(RMNodeState.UNHEALTHY);
return RMNodeState.RUNNING;
}
return RMNodeState.UNHEALTHY;
}
}
}