rabbitmq工作队列

工作队列的主要任务是:避免立刻执行资源密集型任务,然后必须等待其完成。相反地,我们进行任务调度:我们把任务封装为消息发送给队列。工作进行在后台运行并不断的从队列中取出任务然后执行。当你运行了多个工作进程时,任务队列中的任务将会被工作进程共享执行。这样的概念在web应用中极其有用,当在很短的HTTP请求间需要执行复杂的任务。

发送端发送消息:我们使用Thread.sleep来模拟耗时的任务。我们在发送到队列的消息的末尾添加一定数量的点,每个点代表在工作线程中需要耗时1秒,例如hello…将会需要等待3秒。

代码如下:

@Test
	public void sendMessage() throws IOException, TimeoutException{
		Connection connection = factory.newConnection();
		Channel channel = connection.createChannel();
		//声明队列
		channel.queueDeclare(QUEUE_NAME, false, false, false, null);
		//发送10条消息,依次在消息后面附加1-10个点
		for (int i = 0; i < 10; i++)
		{
			String dots = "";
			for (int j = 0; j <= i; j++)
			{
				dots += ".";
			}
			String message = "helloworld" + dots+dots.length();
			channel.basicPublish("", QUEUE_NAME, null, message.getBytes());
			System.out.println(" [x] Sent '" + message + "'");
		}
		//关闭频道和资源
		channel.close();
		connection.close();
	}

控制台输出:

 [x] Sent 'helloworld.1'
 [x] Sent 'helloworld..2'
 [x] Sent 'helloworld...3'
 [x] Sent 'helloworld....4'
 [x] Sent 'helloworld.....5'
 [x] Sent 'helloworld......6'
 [x] Sent 'helloworld.......7'
 [x] Sent 'helloworld........8'
 [x] Sent 'helloworld.........9'

 [x] Sent 'helloworld..........10'

接收端代码如下:

@Test
	public void receiveMessage() throws IOException, TimeoutException, ShutdownSignalException, ConsumerCancelledException, InterruptedException{
		//区分不同工作进程的输出
		int hashCode = Work.class.hashCode();
		Connection connection = factory.newConnection();
		Channel channel = connection.createChannel();
		//声明队列
		channel.queueDeclare(QUEUE_NAME, false, false, false, null);
		System.out.println(hashCode
				+ " [*] Waiting for messages. To exit press CTRL+C");
	
		QueueingConsumer consumer = new QueueingConsumer(channel);
		// 指定消费队列
		channel.basicConsume(QUEUE_NAME, true, consumer);
		while (true){
			QueueingConsumer.Delivery delivery = consumer.nextDelivery();
			String message = new String(delivery.getBody());
 
			System.out.println(hashCode + " [x] Received '" + message + "'");
			doWork(message);
			System.out.println(hashCode + " [x] Done");
		}
	}
		 
	/**
	 * 每个点耗时1s
	 * @param task
	 * @throws InterruptedException
	 */
	private static void doWork(String task) throws InterruptedException{
		for (char ch : task.toCharArray()){
			if (ch == '.')
				Thread.sleep(1000);
		}
	}

默认的,RabbitMQ会一个一个的发送信息给下一个消费者(consumer),而不考虑每个任务的时长等等,且是一次性分配,并非一个一个分配。平均的每个消费者将会获得相等数量的消息。这样分发消息的方式叫做round-robin,控制台打印消息如下:

消费者线程1:

1128920667 [*] Waiting for messages. To exit press CTRL+C
1128920667 [x] Received 'helloworld.1'
1128920667 [x] Done
1128920667 [x] Received 'helloworld....4'
1128920667 [x] Done
1128920667 [x] Received 'helloworld.......7'
1128920667 [x] Done
1128920667 [x] Received 'helloworld..........10'

1128920667 [x] Done

消费者线程2:

751629024 [*] Waiting for messages. To exit press CTRL+C
751629024 [x] Received 'helloworld..2'
751629024 [x] Done
751629024 [x] Received 'helloworld.....5'
751629024 [x] Done
751629024 [x] Received 'helloworld........8'

751629024 [x] Done

消费者线程3:

1130194714 [*] Waiting for messages. To exit press CTRL+C
1130194714 [x] Received 'helloworld...3'
1130194714 [x] Done
1130194714 [x] Received 'helloworld......6'
1130194714 [x] Done
1130194714 [x] Received 'helloworld.........9'

1130194714 [x] Done

Rabbitmq的消息应答(message acknowledgments)

执行一个任务需要花费几秒钟。当一个工作者在执行任务时发生中断。按照上面的代码,一旦RabbItMQ交付了一个信息给消费者,会马上从内存中移除这个信息。在这种情况下,如果杀死正在执行任务的某个工作者,我们会丢失它正在处理的信息。我们也会丢失已经转发给这个工作者且它还未执行的消息。上面的例子,我们首先开启两个任务,然后执行发送任务的代码,然后立即关闭第二个任务,结果为:

消费者线程1:

1128920667 [x] Received 'helloworld..2'
1128920667 [x] Done
1128920667 [x] Received 'helloworld....4'
1128920667 [x] Done
1128920667 [x] Received 'helloworld......6'
1128920667 [x] Done
1128920667 [x] Received 'helloworld........8'
1128920667 [x] Done
1128920667 [x] Received 'helloworld..........10'

1128920667 [x] Done

消费者线程2:

751629024 [x] Received 'helloworld.1'
751629024 [x] Done
751629024 [x] Received 'helloworld...3'
751629024 [x] Done

751629024 [x] Received 'helloworld.....5'

可以看到,第二个工作者至少丢失了7,9号任务,且5号任务未完成。

但是,我们不希望丢失任何任务(信息)。当某个工作者(接收者)被杀死时,我们希望将任务传递给另一个工作者。
为了保证消息永远不会丢失,RabbitMQ支持消息应答(message acknowledgments)。消费者发送应答给RabbitMQ,告诉它信息已经被接收和处理,然后RabbitMQ可以自由的进行信息删除。如果消费者被杀死而没有发送应答,RabbitMQ会认为该信息没有被完全的处理,然后将会重新转发给别的消费者。通过这种方式,可以确认信息不会被丢失,即使消者偶尔被杀死。这种机制并没有超时时间这么一说,RabbitMQ只有在消费者连接断开是重新转发此信息。如果消费者处理一个信息需要耗费特别特别长的时间是允许的。

消息应答默认是打开的。上面的代码中我们通过显示的设置autoAsk=true关闭了这种机制,修改后代码如下:注意标红部分,在开启消息应答之后,消息处理完成之后需要向服务器发送应答

@Test
	public void receiveMessage() throws IOException, TimeoutException, ShutdownSignalException, ConsumerCancelledException, InterruptedException{
		//区分不同工作进程的输出
		int hashCode = Work.class.hashCode();
		Connection connection = factory.newConnection();
		Channel channel = connection.createChannel();
		//声明队列
		channel.queueDeclare(QUEUE_NAME, false, false, false, null);
		System.out.println(hashCode
				+ " [*] Waiting for messages. To exit press CTRL+C");
	
		QueueingConsumer consumer = new QueueingConsumer(channel);
		boolean ack = false ; //打开应答机制
		// 指定消费队列
		channel.basicConsume(QUEUE_NAME, ack, consumer);
		while (true){
			QueueingConsumer.Delivery delivery = consumer.nextDelivery();
			String message = new String(delivery.getBody());
 
			System.out.println(hashCode + " [x] Received '" + message + "'");
			doWork(message);
			System.out.println(hashCode + " [x] Done");
			//发送应答
			channel.basicAck(delivery.getEnvelope().getDeliveryTag(), false);
		}
	}
此时, 打开两个消费者,然后发送任务,立即关闭一个消费者,观察输出:
消费者线程1:
1826774922 [*] Waiting for messages. To exit press CTRL+C
1826774922 [x] Received 'helloworld.1'
1826774922 [x] Done
1826774922 [x] Received 'helloworld...3'
1826774922 [x] Done
1826774922 [x] Received 'helloworld.....5'
1826774922 [x] Done
1826774922 [x] Received 'helloworld.......7'
1826774922 [x] Done
1826774922 [x] Received 'helloworld.........9'
1826774922 [x] Done
1826774922 [x] Received 'helloworld......6'
1826774922 [x] Done
1826774922 [x] Received 'helloworld........8'
1826774922 [x] Done
1826774922 [x] Received 'helloworld..........10'

1826774922 [x] Done

消费者线程2:

2082519650 [*] Waiting for messages. To exit press CTRL+C
2082519650 [x] Received 'helloworld..2'
2082519650 [x] Done
2082519650 [x] Received 'helloworld....4'
2082519650 [x] Done

2082519650 [x] Received 'helloworld......6'

可以看到消费者线程2没有完成的任务6,重新转发给消费者线程1进行完成了。

消息持久化(Message durability)

我们已经学习了即使消费者被杀死,消息也不会被丢失。但是如果此时RabbitMQ服务被停止,我们的消息仍然会丢失。

当RabbitMQ退出或者异常退出,将会丢失所有的队列和信息,除非你告诉它不要丢失。我们需要做两件事来确保信息不会被丢失:我们需要给所有的队列和消息设置持久化的标志。
第一, 我们需要确认RabbitMQ永远不会丢失我们的队列。为了这样,我们需要声明它为持久化的。
boolean durable = true;
channel.queueDeclare("task_queue", durable, false, false, null);
注:RabbitMQ不允许使用不同的参数重新定义一个队列,所以已经存在的队列,我们无法修改其属性。
第二, 我们需要标识我们的信息为持久化的。通过设置MessageProperties(implements BasicProperties)值为PERSISTENT_TEXT_PLAIN。
channel.basicPublish("", "task_queue",MessageProperties.PERSISTENT_TEXT_PLAIN,message.getBytes());

现在你可以执行一个发送消息的程序,然后关闭服务,再重新启动服务,运行消费者程序做下实验

公平转发(Fair dispatch)
或许会发现,目前的消息转发机制(Round-robin)并非是我们想要的。例如,这样一种情况,对于两个消费者,有一系列的任务,奇数任务特别耗时,而偶数任务却很轻松,这样造成一个消费者一直繁忙,另一个消费者却很快执行完任务后等待。
造成这样的原因是因为RabbitMQ仅仅是当消息到达队列进行转发消息。并不在乎有多少任务消费者并未传递一个应答给RabbitMQ。仅仅盲目转发所有的奇数给一个消费者,偶数给另一个消费者。

为了解决这样的问题,我们可以使用basicQos方法,传递参数为prefetchCount = 1。这样告诉RabbitMQ不要在同一时间给一个消费者超过一条消息。换句话说,只有在消费者空闲的时候会发送下一条信息。

//设置最大服务转发消息数量
int prefetchCount = 1;
channel.basicQos(prefetchCount);

完整代码如下:

发送端:

@Test
	public void sendMessage() throws IOException, TimeoutException{
		Connection connection = factory.newConnection();
		Channel channel = connection.createChannel();
		//声明队列
		boolean durable = true;// 1、设置队列持久化
		channel.queueDeclare(QUEUE_NAME, durable, false, false, null);
		//发送10条消息,依次在消息后面附加1-10个点
		for (int i = 0; i < 10; i++)
		{
			String dots = "";
			for (int j = 0; j <= i; j++)
			{
				dots += ".";
			}
			String message = "helloworld" + dots+dots.length();
			// MessageProperties 2、设置消息持久化
			channel.basicPublish("", QUEUE_NAME,MessageProperties.PERSISTENT_TEXT_PLAIN, message.getBytes());
			System.out.println(" [x] Sent '" + message + "'");
		}
		//关闭频道和资源
		channel.close();
		connection.close();
	}

接收端代码如下:

@Test
	public void receiveMessage() throws IOException, TimeoutException, ShutdownSignalException, ConsumerCancelledException, InterruptedException{
		//区分不同工作进程的输出
		int hashCode = Work.class.hashCode();
		Connection connection = factory.newConnection();
		Channel channel = connection.createChannel();
		//声明队列
		channel.queueDeclare(QUEUE_NAME, false, false, false, null);
		System.out.println(hashCode
				+ " [*] Waiting for messages. To exit press CTRL+C");
		//设置最大服务转发消息数量
		int prefetchCount = 1;
		channel.basicQos(prefetchCount);
		QueueingConsumer consumer = new QueueingConsumer(channel);
		// 指定消费队列
		boolean ack = false ; //打开应答机制
		// 指定消费队列
		channel.basicConsume(QUEUE_NAME, ack, consumer);
		while (true){
			QueueingConsumer.Delivery delivery = consumer.nextDelivery();
			String message = new String(delivery.getBody());
 
			System.out.println(hashCode + " [x] Received '" + message + "'");
			doWork(message);
			System.out.println(hashCode + " [x] Done");
			//发送应答
			channel.basicAck(delivery.getEnvelope().getDeliveryTag(), false);
		}
	}

转自http://blog.csdn.net/lmj623565791/article/details/37620057

猜你喜欢

转载自blog.csdn.net/BingoXing/article/details/81011997