1. 应用场景
消息队列中消息之间有先后的依赖关系，后一条消息的处理依赖于前一条消息的处理结果，那么比较适用于顺序消息消费
2. 分析
 RocketMQ
此图可以看出Rocketmq中一个topic和它的mq之间的关系是一对多的关系，客户端向broker中发送消息是根据topic发送的，而消费方消费时也是按照topic来消费的，那么我们怎么保证消息之间的顺序性呢？首先，需要保证顺序的消息要发送到同一个messagequeue中；其次，一个messagequeue只能被一个消费者消费，这点是由消息队列的分配机制来保证的；最后，一个消费者内部对一个mq的消费要保证是有序的。
我们要做到生产者 - messagequeue - 消费者之间是一对一对一的关系。
3. 具体实现
3.1 消息生产者
Message message = new Message(topic, tags, key, body.getBytes());
boolean result;
try {
SendResult sendResult = getConfigBean().getMqProducer().send(message,queueSelector,args);
LoggerUtil.log(Level.INFO,"rocket message product result : " + sendResult.toString());
result = true;
}
catch (MQClientException e) {
throw new MQException("客户端调用状态异常", e);
}
catch (RemotingException | InterruptedException | MQBrokerException e) {
throw new MQException("远程调用异常", e);
}
相比较普通消息的消费，顺序消费在向broker发送消息的时候要指定MessageQueueSelector，此接口RocketMQ提供了三种实现SelectMessageQueueByRandoom，SelectMessageQueueByHash，SelectMessageQueueByMachineRoom，可由调用方自行根据业务实现。指定将消息发送到对应的队列中去；
TopicPublishInfo topicPublishInfo = this.tryToFindTopicPublishInfo(msg.getTopic());
if (topicPublishInfo != null && topicPublishInfo.ok()) {
MessageQueue mq = null;
try {
mq = selector.select(topicPublishInfo.getMessageQueueList(), msg, arg);
}
catch (Throwable e) {
throw new MQClientException("select message queue throwed exception.", e);
}
if (mq != null) {
return this.sendKernelImpl(msg, mq, communicationMode, sendCallback);
}
else {
throw new MQClientException("select message queue return null.", null);
}
}
而普通消息的发送，客户端调用方无需指定队列，MQ会轮询topic下面的MessageQueue发送消息，代码如下
/**
* 如果lastBrokerName不为null，则寻找与其不同的MessageQueue
*/
public MessageQueue selectOneMessageQueue(final String lastBrokerName) {
if (lastBrokerName != null) {
int index = this.sendWhichQueue.getAndIncrement();
for (int i = 0; i < this.messageQueueList.size(); i++) {
int pos = Math.abs(index++) % this.messageQueueList.size();
MessageQueue mq = this.messageQueueList.get(pos);
if (!mq.getBrokerName().equals(lastBrokerName)) {
return mq;
}
}
return null;
}
else {
int index = this.sendWhichQueue.getAndIncrement();
int pos = Math.abs(index) % this.messageQueueList.size();
return this.messageQueueList.get(pos);
}
}
3.2 消息消费者
消息消费方与普通消息消费只有一个地方不同,在consumer中注册MessageListenerOrderly，而不是MessageListenerConcurrently
this.consumer.registerMessageListener((MessageListenerOrderly) (msgs, context) -> {
msgs.stream().forEach(
messageExt -> {
//TODO consume message
});
return ConsumeOrderlyStatus.SUCCESS;
});
我们进一步看下RocketMQ是怎么处理的ConsumeMessageOrderlyService在启动的时候，如果是集群模式下会启动一个单线程的定时调度任务，延迟一秒，时间间隔为20秒，执行rebalanceImpl的lockAll()方法。
public void start() {
// 启动定时lock队列服务
if (MessageModel.CLUSTERING.equals(ConsumeMessageOrderlyService.this.defaultMQPushConsumerImpl
.messageModel())) {
this.scheduledExecutorService.scheduleAtFixedRate(new Runnable() {
@Override
public void run() {
ConsumeMessageOrderlyService.this.lockMQPeriodically();
}
}, 1000 * 1, ProcessQueue.RebalanceLockInterval, TimeUnit.MILLISECONDS);
}
}
这个方法会锁定相关broker下面的相关的messagequeue，对message对应的processQueue设置是否锁定，这个地方下面会用到
class ConsumeRequest implements Runnable {
private final ProcessQueue processQueue;
private final MessageQueue messageQueue;
public ConsumeRequest(ProcessQueue processQueue, MessageQueue messageQueue) {
this.processQueue = processQueue;
this.messageQueue = messageQueue;
}
@Override
public void run() {
if (this.processQueue.isDroped()) {
log.warn("run, the message queue not be able to consume, because it's dropped. {}",
this.messageQueue);
return;
}
// 保证在当前Consumer内，同一队列串行消费
final Object objLock = messageQueueLock.fetchLockObject(this.messageQueue);
synchronized (objLock) {
// 保证在Consumer集群，同一队列串行消费
if (MessageModel.BROADCASTING
.equals(ConsumeMessageOrderlyService.this.defaultMQPushConsumerImpl.messageModel())
|| (this.processQueue.isLocked() && !this.processQueue.isLockExpired())) {
final long beginTime = System.currentTimeMillis();
for (boolean continueConsume = true; continueConsume;) {
if (this.processQueue.isDroped()) {
log.warn("the message queue not be able to consume, because it's dropped. {}",
this.messageQueue);
break;
}
if (MessageModel.CLUSTERING
.equals(ConsumeMessageOrderlyService.this.defaultMQPushConsumerImpl
.messageModel())
&& !this.processQueue.isLocked()) {
log.warn("the message queue not locked, so consume later, {}", this.messageQueue);
ConsumeMessageOrderlyService.this.tryLockLaterAndReconsume(this.messageQueue,
this.processQueue, 10);
break;
}
if (MessageModel.CLUSTERING
.equals(ConsumeMessageOrderlyService.this.defaultMQPushConsumerImpl
.messageModel())
&& this.processQueue.isLockExpired()) {
log.warn("the message queue lock expired, so consume later, {}",
this.messageQueue);
ConsumeMessageOrderlyService.this.tryLockLaterAndReconsume(this.messageQueue,
this.processQueue, 10);
break;
}
// 在线程数小于队列数情况下，防止个别队列被饿死
long interval = System.currentTimeMillis() - beginTime;
if (interval > MaxTimeConsumeContinuously) {
// 过10ms后再消费
ConsumeMessageOrderlyService.this.submitConsumeRequestLater(processQueue,
messageQueue, 10);
break;
}
final int consumeBatchSize =
ConsumeMessageOrderlyService.this.defaultMQPushConsumer
.getConsumeMessageBatchMaxSize();
List<MessageExt> msgs = this.processQueue.takeMessags(consumeBatchSize);
if (!msgs.isEmpty()) {
final ConsumeOrderlyContext context =
new ConsumeOrderlyContext(this.messageQueue);
ConsumeOrderlyStatus status = null;
// 执行Hook
ConsumeMessageContext consumeMessageContext = null;
if (ConsumeMessageOrderlyService.this.defaultMQPushConsumerImpl.hasHook()) {
consumeMessageContext = new ConsumeMessageContext();
consumeMessageContext
.setConsumerGroup(ConsumeMessageOrderlyService.this.defaultMQPushConsumer
.getConsumerGroup());
consumeMessageContext.setMq(messageQueue);
consumeMessageContext.setMsgList(msgs);
consumeMessageContext.setSuccess(false);
ConsumeMessageOrderlyService.this.defaultMQPushConsumerImpl
.executeHookBefore(consumeMessageContext);
}
long beginTimestamp = System.currentTimeMillis();
//处理队列加锁
try {
this.processQueue.getLockConsume().lock();
if (this.processQueue.isDroped()) {
log.warn(
"consumeMessage, the message queue not be able to consume, because it's dropped. {}",
this.messageQueue);
break;
}
status =
messageListener.consumeMessage(Collections.unmodifiableList(msgs),
context);
}
catch (Throwable e) {
log.warn("consumeMessage exception: {} Group: {} Msgs: {} MQ: {}",//
RemotingHelper.exceptionSimpleDesc(e),//
ConsumeMessageOrderlyService.this.consumerGroup,//
msgs,//
messageQueue);
}
finally {
this.processQueue.getLockConsume().unlock();
}
// 针对异常返回代码打印日志
if (null == status //
|| ConsumeOrderlyStatus.ROLLBACK == status//
|| ConsumeOrderlyStatus.SUSPEND_CURRENT_QUEUE_A_MOMENT == status) {
log.warn("consumeMessage Orderly return not OK, Group: {} Msgs: {} MQ: {}",//
ConsumeMessageOrderlyService.this.consumerGroup,//
msgs,//
messageQueue);
}
long consumeRT = System.currentTimeMillis() - beginTimestamp;
// 用户抛出异常或者返回null，都挂起队列
if (null == status) {
status = ConsumeOrderlyStatus.SUSPEND_CURRENT_QUEUE_A_MOMENT;
}
// 执行Hook
if (ConsumeMessageOrderlyService.this.defaultMQPushConsumerImpl.hasHook()) {
consumeMessageContext.setStatus(status.toString());
consumeMessageContext.setSuccess(ConsumeOrderlyStatus.SUCCESS == status
|| ConsumeOrderlyStatus.COMMIT == status);
ConsumeMessageOrderlyService.this.defaultMQPushConsumerImpl
.executeHookAfter(consumeMessageContext);
}
// 记录统计信息
ConsumeMessageOrderlyService.this.getConsumerStatsManager().incConsumeRT(
ConsumeMessageOrderlyService.this.consumerGroup, messageQueue.getTopic(),
consumeRT);
continueConsume =
ConsumeMessageOrderlyService.this.processConsumeResult(msgs, status,
context, this);
}
else {
continueConsume = false;
}
}
}
// 没有拿到当前队列的锁，稍后再消费
else {
if (this.processQueue.isDroped()) {
log.warn("the message queue not be able to consume, because it's dropped. {}",
this.messageQueue);
return;
}
ConsumeMessageOrderlyService.this.tryLockLaterAndReconsume(this.messageQueue,
this.processQueue, 100);
}
}
}
public ProcessQueue getProcessQueue() {
return processQueue;
}
public MessageQueue getMessageQueue() {
return messageQueue;
}
}
4. 问题
4.1 降低了吞吐量
4.2 前一条消息消费出现问题，后续的处理流程会阻塞
|
