RabbitMQ学习笔记(三、生产者与消费者)
目录:
- 细说交换器
- 细说队列
- 发送消息
- 消费消息
- 确认与拒绝
细说交换器:
1、方法:
public AMQP.Exchange.DeclareOk exchangeDeclare(String exchange, String type, boolean durable, boolean autoDelete, boolean internal, Map<String, Object> arguments) throws IOException
autoDelete:自动删除;必须有解绑的动作,且需要全部解绑后交换器才会删除。
internal:内置路由器;客户端无法直接发送消息到交换器,只能通过交换器路由到内置路由器。
2、其他方法:
)不等待服务通知创建交换器命令:
public void exchangeDeclareNoWait(String exchange, String type, boolean durable, boolean autoDelete, boolean internal, Map<String, Object> arguments) throws IOException
)检测交换器
public AMQP.Exchange.DeclareOk exchangeDeclarePassive(String name) throws IOException
)删除交换器
public AMQP.Exchange.DeleteOk exchangeDelete(String exchange, boolean ifUnused) throws IOException
public void exchangeDeleteNoWait(String exchange, boolean ifUnused) throws IOException
public AMQP.Exchange.DeleteOk exchangeDelete(String exchange) throws IOException
ifUnused:为true,交换器没有被使用时删除这个交换器;为false,不管三七二十一直接删除这个交换器。
细说队列:
Queue.DeclareOk queueDeclare(String queue, boolean durable, boolean exclusive, boolean autoDelete, Map<String, Object> arguments) throws IOException;
队列主要介绍参数:boolean exclusive、boolean autoDelete、Map<String, Object> arguments。
1、exclusive:排它性,true-排它,false-非排它。
设置为排它队列后,该队列只对首次声明它的连接可见,并在连接断开时自动删除;适用于一个客户端同时读写消息的场景。
1 public class Product { 2 3 private static final String EXCHANGE_NAME = "exclusive.exchange"; 4 private static final String QUEUE_NAME = "exclusive.queue"; 5 private static final String ROUTING_KEY_NAME = "exclusive.routing.key"; 6 7 public static void main(String[] args) throws IOException, TimeoutException, InterruptedException { 8 Connection connection = RabbitMqUtils.getConnection(); 9 Channel channel = connection.createChannel(); 10 channel.exchangeDeclare(EXCHANGE_NAME, BuiltinExchangeType.DIRECT); 11 channel.queueDeclare(QUEUE_NAME, true, true, false, null); 12 channel.queueBind(QUEUE_NAME, EXCHANGE_NAME, ROUTING_KEY_NAME); 13 channel.basicPublish(EXCHANGE_NAME, ROUTING_KEY_NAME, MessageProperties.PERSISTENT_TEXT_PLAIN, 14 "排它队列".getBytes()); 15 16 Connection connection2 = RabbitMqUtils.getConnection(); 17 Channel channel2 = connection2.createChannel(); 18 channel2.basicConsume(QUEUE_NAME, new DefaultConsumer(channel2) { 19 @Override 20 public void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException { 21 System.err.println("收到消息:" + new String(body)); 22 } 23 }); 24 25 // 线程休眠5秒,待消息回调完成 26 TimeUnit.SECONDS.sleep(5); 27 28 RabbitMqUtils.close(connection, channel); 29 } 30 }
代码中18行用的是channel2,也就是connection2创建的信道,此时不满足排它队列的只允许首次创建它的连接使用,所以会抛出异常;将16行注释掉,17行的connection2替换成connection便能正常运行。
2、autoDelete:自动删除,true-自动删除,false-非自动删除。
设置为自动删除的队列后,该队列在其消费者都断开连接后,自动删除,不管队列里是否存在数据。
3、arguments:队列的其它参数。
前面几个参数很简单,我们这里只讲重点的x-dead-letter-exchange、x-dead-letter-routing-key。
1 public class DeadProduct { 2 3 private static final String EXCHANGE_NAME = "normal.exchange"; 4 private static final String DEAD_EXCHANGE_NAME = "dead.exchange"; 5 private static final String QUEUE_NAME = "normal.queue"; 6 private static final String DEAD_QUEUE_NAME = "dead.queue"; 7 private static final String ROUTING_KEY_NAME = "normal.routing.key"; 8 private static final String DEAD_ROUTING_KEY_NAME = "dead.routing.key"; 9 10 public static void main(String[] args) throws IOException, TimeoutException { 11 Connection connection = RabbitMqUtils.getConnection(); 12 Channel channel = connection.createChannel(); 13 14 channel.exchangeDeclare(EXCHANGE_NAME, BuiltinExchangeType.DIRECT); 15 channel.exchangeDeclare(DEAD_EXCHANGE_NAME, BuiltinExchangeType.DIRECT); 16 17 channel.queueDeclare(QUEUE_NAME, false, false, false, getArguments()); 18 channel.queueDeclare(DEAD_QUEUE_NAME, false, false, false, null); 19 20 channel.queueBind(QUEUE_NAME, EXCHANGE_NAME, ROUTING_KEY_NAME); 21 channel.queueBind(DEAD_QUEUE_NAME, DEAD_EXCHANGE_NAME, ROUTING_KEY_NAME); 22 for (int i = 0; i < 10; i++) { 23 String message = "死信交换机" + i; 24 channel.basicPublish(EXCHANGE_NAME, ROUTING_KEY_NAME, MessageProperties.PERSISTENT_TEXT_PLAIN, 25 message.getBytes()); 26 } 27 28 RabbitMqUtils.close(connection, channel); 29 } 30 31 private static Map<String, Object> getArguments() { 32 Map<String, Object> result = new HashMap<String, Object>(5); 33 result.put("x-message-ttl", 5000); 34 result.put("x-dead-letter-exchange", DEAD_EXCHANGE_NAME); 35 result.put("x-dead-letter-routing-key", ROUTING_KEY_NAME); 36 return result; 37 } 38 }
x-dead-letter-exchange参数其实只是定义了一个交换机的名称,实际上你还需要定义一套mq的流程,如代码exchangeDeclare、queueDeclare、queueBind。
x-dead-letter-routing-key参数的routing-key需要与之前的路由key一致,否则数据将会匹配不到,也就不会走到DEAD_EXCHANGE_NAME里去了。
发送消息:
void basicPublish(String exchange, String routingKey, boolean mandatory, boolean immediate, BasicProperties props, byte[] body) throws IOException;
发送消息主要参数:mandatory、immediate。
1、mandatory:当消息无法路由到队列的处理方式。
2、immediate:当路由的队列无消费者时如何处理。
消费消息:
消费消息分为两种,一种是推,一种是主动拉。
1、推:basicConsumer(推,相当于监听,会消费调当前队列中所有的消息)
String basicConsume(String queue, boolean autoAck, String consumerTag, boolean noLocal, boolean exclusive, Map<String, Object> arguments, Consumer callback) throws IOException;
autoAck:是否自动确认,true-自动确认,false-非自动确认;当autoAck=true时,消费完时就会自动删除这条消息。
noLocal:设置为ture时,不能将同一个connection生产的消息在此connection消费,也就是说一个connection不能同时为生产者和消费者。
2、拉:basicGet(拉,只会拉取一条消息)
GetResponse basicGet(String queue, boolean autoAck) throws IOException;
注意:不能用for循环的basicGet代替basicConsumer,因为这样性能非常差。
确认与拒绝:
1、消息确认:
)概念解释:
为了保证消息可靠的到达消费者,RabbitMQ提供了消息确认机制。
首先我们要知道除了手动的确认消息,还可以通过basicConsumer指定的autoAck参数来确认消息的到达。
当autoAck=false时,RabbitMQ会显式的等待消息回复确认后,才会将消息从内存或磁盘中移除(先设置为移除标记,然后在真正的移除);当autoAck=true时,RabbitMQ不管消费者有没有接收到消息,都会直接将消息移除。
注意:当RabbitMQ一直没有收到消费者确认消息的通知,并且消费者的连接断开时,那么此条消息会重新进入队列。
void basicAck(long deliveryTag, boolean multiple) throws IOException;
multiple:是否确认多条消息;true-确认多条(确认该信道上deliveryTag标记之前所有未经确认的消息),false-确认单条。
)代码示例:
1 public class AckProduct { 2 3 private static final String EXCHANGE_NAME = "ack.exchange"; 4 private static final String QUEUE_NAME = "ack.queue"; 5 private static final String ROUTING_KEY = "ack.routing-key"; 6 7 public static void main(String[] args) throws IOException, TimeoutException { 8 Connection connection = RabbitMqUtils.getConnection(); 9 Channel channel = connection.createChannel(); 10 channel.exchangeDeclare(EXCHANGE_NAME, BuiltinExchangeType.DIRECT); 11 channel.queueDeclare(QUEUE_NAME, false, false, false, null); 12 channel.queueBind(QUEUE_NAME, EXCHANGE_NAME, ROUTING_KEY); 13 14 for (int i = 0; i < 20; i++) { 15 channel.basicPublish(EXCHANGE_NAME, ROUTING_KEY, null, ("ack" + i).getBytes()); 16 } 17 18 RabbitMqUtils.close(connection, channel); 19 } 20 }
1 public class AckConsumer { 2 3 private static final String QUEUE_NAME = "ack.queue"; 4 5 public static void main(String[] args) throws IOException, TimeoutException, InterruptedException { 6 Connection connection = RabbitMqUtils.getConnection(); 7 final Channel channel = connection.createChannel(); 8 9 // 因为basicConsumer会消费队列中所有的消息,这样不方便演示basicAck中multiple参数的效果,故用basicGet来消费消息 10 GetResponse getResponse = channel.basicGet(QUEUE_NAME, false); 11 System.err.println(getResponse.getEnvelope().getDeliveryTag() + ": " + new String(getResponse.getBody())); 12 System.err.println("-------------------------"); 13 14 GetResponse getResponse2 = channel.basicGet(QUEUE_NAME, false); 15 System.err.println(getResponse2.getEnvelope().getDeliveryTag() + ": " + new String(getResponse2.getBody())); 16 System.err.println("-------------------------"); 17 18 TimeUnit.SECONDS.sleep(10); 19 20 GetResponse getResponse3 = channel.basicGet(QUEUE_NAME, false); 21 System.err.println(getResponse3.getEnvelope().getDeliveryTag() + ": " + new String(getResponse3.getBody())); 22 channel.basicAck(getResponse3.getEnvelope().getDeliveryTag(), false); 23 System.err.println("-------------------------"); 24 25 // 休眠10秒,方便看到multiple参数的效果 26 TimeUnit.SECONDS.sleep(10); 27 28 GetResponse getResponse4 = channel.basicGet(QUEUE_NAME, false); 29 System.err.println(getResponse4.getEnvelope().getDeliveryTag() + ": " + new String(getResponse4.getBody())); 30 channel.basicAck(getResponse4.getEnvelope().getDeliveryTag(), true); 31 32 RabbitMqUtils.close(connection, channel); 33 } 34 }
首先AckProduct生产20条消息,AckConsumer先消费两条消息,再休眠10秒,此时可以看到ready为18,unacked为2,这两消息未确认(10 - 18行);
然后我们确认一条消息后,发现ready为17,unacked为2(20 - 23行);
此时便是关键时刻,若multiple真是如其定义一般,那么在执行28 - 30行后,ready应该是16,而未确认消息unacked应该是0才对,10秒后发现果然如此。
2、消息拒绝:
)拒绝单条消息:
void basicReject(long deliveryTag, boolean requeue) throws IOException;
)拒绝多条消息:
void basicNack(long deliveryTag, boolean multiple, boolean requeue) throws IOException;
)参数说明:
multiple:是否拒绝多条消息;true-拒绝多条(拒绝该信道上deliveryTag标记之前所有未经确认的消息),false-拒绝单条(与basicAck类似)。
requeue:是否将拒绝的消息重新放入队列中;true-重新放入,false-丢弃。