心跳服务
Flink对各组件服务状态的监控统一使用心跳服务来管理,如同其他的服务一样,这一部分的代码也是解耦的,被多个地方重复使用。本次重点是学习下Flink是如何封装心跳管理的,不涉及对心跳超时、传递心跳信息的后续处理。先回答如下问题,在看具体代码。
- 如何判定心跳超时?
心跳服务启动后,Flink会启动一个线程来处理心跳超时事件,在设定的心跳超时时间到达后才执行线程。如果接收到组件的心跳消息,会先将该线程取消而后重新开启,重置心跳超时事件的触发。 - 何时调用双方才发起心跳检查?
心跳检查是双向的,一方会主动发起心跳请求,而另一方则是对心跳做出响应,两者通过RPC相互调用,重置对方的超时线程。以JobManager和TaskManager为例,JM在启动时会开启周期调度,向已经注册到JM中的TM发起心跳检查,通过RPC调用TM的requestHeartbeat方法,重置对JM超时线程的调用,表示当前JM状态正常。在TM的requestHeartbeat方法被调用后,通过RPC调用JM的receiveHeartbeat,重置对TM超时线程的调用,表示TM状态正常。 - 如何处理心跳超时?
心跳服务依赖HeartbeatListener,当在timeout时间范围内未接收到心跳响应,则会触发超时处理线程,该线程通过调用HeartbeatListener的notifyHeartbeatTimeout方法做后续重连操作。
心跳服务使用的主要接口和类如下图所示:
HeartbeatTarget:用来发送心跳信息,也用来接收心跳响应。心跳发送者和接收者都是该接口的子类。两者都可以携带Payload信息。
public interface HeartbeatTarget<I> {
// 接收监控目标发送来的心跳请求信息
void receiveHeartbeat(ResourceID heartbeatOrigin, I heartbeatPayload);
// 向监控目标发送心跳请求
void requestHeartbeat(ResourceID requestOrigin, I heartbeatPayload);
}
HeartbeatManager:心跳管理器用来启动或停止监视HeartbeatTarget,并报告该目标心跳超时事件。通过monitorTarget来传递并监控HeartbeatTarget,这个方法可以看做是整个服务的输入,告诉心跳服务去管理哪些目标。
public interface HeartbeatManager<I, O> extends HeartbeatTarget<I> {
// 开始监控心跳目标,当目标心跳超时,会报告给与HeartbeatManager关联的HeartbeatListener
void monitorTarget(ResourceID resourceID, HeartbeatTarget<O> heartbeatTarget);
// 取消监控心跳目标,ResourceID是心跳目标的标识
void unmonitorTarget(ResourceID resourceID);
// 停止当前心跳管理器
void stop();
//返回最近一次心跳时间,如果心跳目标被移除了则返回-1
long getLastHeartbeatFrom(ResourceID resourceId);
}
HeartbeatListener:是和HeartbeatManager密切相关的接口,可以看做服务的输出。主要有以下作用:
- 心跳超时通知
- 接收心跳信息中的Payload
- 检索作为心跳响应输出的Payload
public interface HeartbeatListener<I, O> {
// 心跳超时会调用该方法
void notifyHeartbeatTimeout(ResourceID resourceID);
// 接收到有关心跳的payload就会执行该方法
void reportPayload(ResourceID resourceID, I payload);
// 检索下一个心跳消息的Payload
O retrievePayload(ResourceID resourceID);
}
相关代码
- 心跳服务创建入口
集群启动时会初始化一些服务,在ClusterEntrypoint#initializeServices方法中创建心跳管理服务。
heartbeatServices = createHeartbeatServices(configuration);
protected HeartbeatServices createHeartbeatServices(Configuration configuration) {
return HeartbeatServices.fromConfiguration(configuration);
}
- 从配置文件中提取心跳间隔heartbeat.interval和心跳超时时间heartbeat.timeout配置,并创建HeartbeatServices
public static HeartbeatServices fromConfiguration(Configuration configuration) {
// 心跳间隔,默认10s
long heartbeatInterval = configuration.getLong(HeartbeatManagerOptions.HEARTBEAT_INTERVAL);
// 心跳超时时间,50s
long heartbeatTimeout = configuration.getLong(HeartbeatManagerOptions.HEARTBEAT_TIMEOUT);
return new HeartbeatServices(heartbeatInterval, heartbeatTimeout);
}
-
createHeartbeatManager和createHeartbeatManagerSender核心方法
这两个方法使用的两个类HeartbeatManagerImpl、HeartbeatManagerSenderImpl是整个心跳服务的关键。
HeartbeatManagerImpl由心跳被发起方(例如TM)创建,接收发起方的(JM)心跳发送请求,包含两个重要属性heartbeatListener、heartbeatTargets。heartbeatTargets是一个Map集合,key代表要发送心跳组件(例如:TM)的ID,value则是为当前组件创建的触发心跳超时的线程HeartbeatMonitor,两者一一对应,心跳超会触发 heartbeatListener的notifyHeartbeatTimeout方法。注意:被发起方监控线程的开启是在接收到请求心跳(requestHeartbeat被调用后)以后才触发的,属于被动触发。
// 外部调用者传递heartbeatTarget,并为其创建一个HeartbeatMonitor
public void monitorTarget(ResourceID resourceID, HeartbeatTarget<O> heartbeatTarget) {
if (!stopped) {
if (heartbeatTargets.containsKey(resourceID)) {
log.debug("The target with resource ID {} is already been monitored.", resourceID);
} else {
HeartbeatManagerImpl.HeartbeatMonitor<O> heartbeatMonitor = new HeartbeatManagerImpl.HeartbeatMonitor<>(
resourceID,
heartbeatTarget,
mainThreadExecutor,
heartbeatListener,
heartbeatTimeoutIntervalMs);
heartbeatTargets.put(
resourceID,
heartbeatMonitor);
// check if we have stopped in the meantime (concurrent stop operation)
if (stopped) {
heartbeatMonitor.cancel();
heartbeatTargets.remove(resourceID);
}
}
}
}
Heartbeat monitor管理心跳目标,在timeout时间内没有接收到心跳信号,则判定心跳超时,通知给HeartbeatListener,每次接收到心跳信号则重置当前timer。
static class HeartbeatMonitor<O> implements Runnable {
/** Resource ID of the monitored heartbeat target. */
private final ResourceID resourceID;
/** Associated heartbeat target. */
private final HeartbeatTarget<O> heartbeatTarget;
private final ScheduledExecutor scheduledExecutor;
/** Listener which is notified about heartbeat timeouts. */
private final HeartbeatListener<?, ?> heartbeatListener;
/** Maximum heartbeat timeout interval. */
private final long heartbeatTimeoutIntervalMs;
private volatile ScheduledFuture<?> futureTimeout;
// AtomicReference 使用
private final AtomicReference<State> state = new AtomicReference<>(State.RUNNING);
// 最近一次接收到心跳的时间
private volatile long lastHeartbeat;
HeartbeatMonitor(
ResourceID resourceID,
HeartbeatTarget<O> heartbeatTarget,
ScheduledExecutor scheduledExecutor,
HeartbeatListener<?, O> heartbeatListener,
long heartbeatTimeoutIntervalMs) {
// 被监控的机器ID
this.resourceID = Preconditions.checkNotNull(resourceID);
// 心跳目录
this.heartbeatTarget = Preconditions.checkNotNull(heartbeatTarget);
this.scheduledExecutor = Preconditions.checkNotNull(scheduledExecutor);
// 心跳监听器
this.heartbeatListener = Preconditions.checkNotNull(heartbeatListener);
Preconditions.checkArgument(heartbeatTimeoutIntervalMs > 0L, "The heartbeat timeout interval has to be larger than 0.");
this.heartbeatTimeoutIntervalMs = heartbeatTimeoutIntervalMs;
lastHeartbeat = 0L;
resetHeartbeatTimeout(heartbeatTimeoutIntervalMs);
}
HeartbeatTarget<O> getHeartbeatTarget() {
return heartbeatTarget;
}
ResourceID getHeartbeatTargetId() {
return resourceID;
}
public long getLastHeartbeat() {
return lastHeartbeat;
}
// 报告心跳
void reportHeartbeat() {
// 保留最近一次接收心跳时间
lastHeartbeat = System.currentTimeMillis();
// 接收心跳后, 重置timeout线程
resetHeartbeatTimeout(heartbeatTimeoutIntervalMs);
}
// 重置TIMEOUT
void resetHeartbeatTimeout(long heartbeatTimeout) {
if (state.get() == State.RUNNING) {
//先取消线程,在重新开启
cancelTimeout();
// 启动超时线程
futureTimeout = scheduledExecutor.schedule(this, heartbeatTimeout, TimeUnit.MILLISECONDS);
// Double check for concurrent accesses (e.g. a firing of the scheduled future)
if (state.get() != State.RUNNING) {
cancelTimeout();
}
}
}
void cancel() {
// we can only cancel if we are in state running
if (state.compareAndSet(State.RUNNING, State.CANCELED)) {
cancelTimeout();
}
}
private void cancelTimeout() {
if (futureTimeout != null) {
futureTimeout.cancel(true);
}
}
public boolean isCanceled() {
return state.get() == State.CANCELED;
}
// 心跳超时,触发lister的notifyHeartbeatTimeout
@Override
public void run() {
// The heartbeat has timed out if we're in state running
if (state.compareAndSet(State.RUNNING, State.TIMEOUT)) {
heartbeatListener.notifyHeartbeatTimeout(resourceID);
}
}
private enum State {
RUNNING,
TIMEOUT,
CANCELED
}
}
HeartbeatManagerSenderImpl是HeartbeatManagerImpl的子类,由心跳管理的一方(例如JM)创建,创建后立即开启周期调度线程,每次遍历自己管理的heartbeatTarget,触发heartbeatTarget.requestHeartbeat,属于主动触发。
this.heartbeatPeriod = heartbeatPeriod;
mainThreadExecutor.schedule(this, 0L, TimeUnit.MILLISECONDS);
public void run() {
if (!stopped) {
log.debug("Trigger heartbeat request.");
for (HeartbeatMonitor<O> heartbeatMonitor : getHeartbeatTargets()) {
requestHeartbeat(heartbeatMonitor);
}
// 周期调度
getMainThreadExecutor().schedule(this, heartbeatPeriod, TimeUnit.MILLISECONDS);
}
}
// 主动发起心跳检查
private void requestHeartbeat(HeartbeatMonitor<O> heartbeatMonitor) {
O payload = getHeartbeatListener().retrievePayload(heartbeatMonitor.getHeartbeatTargetId());
final HeartbeatTarget<O> heartbeatTarget = heartbeatMonitor.getHeartbeatTarget();
heartbeatTarget.requestHeartbeat(getOwnResourceID(), payload);
}
- TM中HeartbeatManagerImpl的使用
- TM启动后会和JM建立连接,连接成功后为JM创建HeartbeatTarget,并重写receiveHeartbeat方法。此时,HeartbeatManagerImpl中已经创建好对应monitor线程,只有在JM执行requestHeartbeat后,才会触发该线程的执行。
TaskExecutor#establishJobManagerConnection
private void establishJobManagerConnection(JobID jobId, final JobMasterGateway jobMasterGateway, JMTMRegistrationSuccess registrationSuccess) {
ResourceID jobManagerResourceID = registrationSuccess.getResourceID();
// monitor the job manager as heartbeat target
jobManagerHeartbeatManager.monitorTarget(jobManagerResourceID, new HeartbeatTarget<AccumulatorReport>() {
// tm只接收心跳请求
@Override
public void receiveHeartbeat(ResourceID resourceID, AccumulatorReport payload) {
jobMasterGateway.heartbeatFromTaskManager(resourceID, payload);
}
@Override
public void requestHeartbeat(ResourceID resourceID, AccumulatorReport payload) {
// request heartbeat will never be called on the task manager side
}
});
}
- 在receiveHeartbeat方法内部,直接通过RPC调用JM的heartbeatFromTaskManager方法,最终进入HeartbeatManagerImpl#receiveHeartbeat中,在reportHeartbeat重置JM monitor线程的触发,代表TM正常执行。
## jobMaster
public void heartbeatFromTaskManager(final ResourceID resourceID, AccumulatorReport accumulatorReport) {
taskManagerHeartbeatManager.receiveHeartbeat(resourceID, accumulatorReport);
}
## taskManagerHeartbeatManager的创建
taskManagerHeartbeatManager = heartbeatServices.createHeartbeatManagerSender(
resourceId,
new TaskManagerHeartbeatListener(),
getMainThreadExecutor(),
log);
## JM接收到心跳
public void receiveHeartbeat(ResourceID heartbeatOrigin, I heartbeatPayload) {
if (!stopped) {
log.debug("Received heartbeat from {}.", heartbeatOrigin);
//接收到心跳后的操作
reportHeartbeat(heartbeatOrigin);
if (heartbeatPayload != null) {
heartbeatListener.reportPayload(heartbeatOrigin, heartbeatPayload);
}
}
}
- JM中HeartbeatManagerSenderImpl使用
- 接收TM的注册后,加入到心跳目标的集合中,在下一个周期会触发TM的requestHeartbeat。
public CompletableFuture<RegistrationResponse> registerTaskManager(
final String taskManagerRpcAddress,
final TaskManagerLocation taskManagerLocation,
final Time timeout) {
final ResourceID taskManagerId = taskManagerLocation.getResourceID();
if (registeredTaskManagers.containsKey(taskManagerId)) {
final RegistrationResponse response = new JMTMRegistrationSuccess(resourceId);
return CompletableFuture.completedFuture(response);
} else {
return getRpcService()
.connect(taskManagerRpcAddress, TaskExecutorGateway.class)
.handleAsync(
(TaskExecutorGateway taskExecutorGateway, Throwable throwable) -> {
if (throwable != null) {
return new RegistrationResponse.Decline(throwable.getMessage());
}
slotPool.registerTaskManager(taskManagerId);
registeredTaskManagers.put(taskManagerId, Tuple2.of(taskManagerLocation, taskExecutorGateway));
// 加入心跳目标
// monitor the task manager as heartbeat target
taskManagerHeartbeatManager.monitorTarget(taskManagerId, new HeartbeatTarget<AllocatedSlotReport>() {
@Override
public void receiveHeartbeat(ResourceID resourceID, AllocatedSlotReport payload) {
// the task manager will not request heartbeat, so this method will never be called currently
}
// JM要求TM发送心跳请求
@Override
public void requestHeartbeat(ResourceID resourceID, AllocatedSlotReport allocatedSlotReport) {
taskExecutorGateway.heartbeatFromJobManager(resourceID, allocatedSlotReport);
}
});
return new JMTMRegistrationSuccess(resourceId);
},
getMainThreadExecutor());
}
}
- 在requestHeartbeat中RPC调用taskExecutor#heartbeatFromJobManager,最终调用HeartbeatManagerImpl中的requestHeartbeat,启动或重置超时线程,表示JM状态正常。在该方法中又通过RPC调用JM的receiveHeartbeat。
public void requestHeartbeat(final ResourceID requestOrigin, I heartbeatPayload) {
if (!stopped) {
log.debug("Received heartbeat request from {}.", requestOrigin);
//启动超时线程 ,并获取heartbeatTarget,此时的目标是JM
final HeartbeatTarget<O> heartbeatTarget = reportHeartbeat(requestOrigin);
if (heartbeatTarget != null) {
if (heartbeatPayload != null) {
heartbeatListener.reportPayload(requestOrigin, heartbeatPayload);
}
// RPC调用JM的receiveHeartbeat
heartbeatTarget.receiveHeartbeat(getOwnResourceID(), heartbeatListener.retrievePayload(requestOrigin));
}
}
}
涉及的类不是很多,麻烦的是RPC过程中确定方法的调用方。
项目中的实现
我们项目中使用的心跳检查机制是通过ZK进行消息传递实现的。Slave心跳服务运行时,会将节点心跳信息以数字形式定期同步到Zookeeper中。Master节点会在设定的调度周期内从Zookeeper中拉取节点心跳信息,初次获取时将节点心跳信息缓存到本地内存,再次获取时判断是否与本地内存缓存中的心跳信息相等,如果不等则代表工作节点正常工作,新的心跳信息覆盖本地缓存心跳信息。如果相等则表示上一个心跳检测周期内,节点未将心跳信息同步到Zookeeper中,此时心跳异常次数递增。如果心跳异常次数达到设定的阀值,则Master判定该Slave节点宕机并禁用该节点,同时进行任务迁移。
Flink则是通过RPC相互调用的方式,并重置对方超时线程的调度。相较于我们那种方式,Flink把心跳管理封装成一个单独的服务来使用,做到了解耦,扩展起来也比较方便,也确实在很多地方都使用了这部分代码,不过要依赖RPC之间的通信。
网友评论