RocketMQ-顺序消费

　　看了https://www.jianshu.com/p/453c6e7ff81c这篇博客，得出顺序消费的结论。“要实现严格的顺序消息，简单且可行的办法就是：保证生产者 - MQServer - 消费者是一对一的关系”。

我们下面通过几个实例来学习RocketMQ的顺序消费。

一、单节点，也就是一个Producer一个Consumer。

操作步骤：

　　1、先启动ConsumerQueue1

　　2、再启动ProducerQueue

Producer端：

package org.hope.lee.consumer.queue;

import com.alibaba.rocketmq.client.exception.MQBrokerException;
import com.alibaba.rocketmq.client.exception.MQClientException;
import com.alibaba.rocketmq.client.producer.DefaultMQProducer;
import com.alibaba.rocketmq.client.producer.MessageQueueSelector;
import com.alibaba.rocketmq.client.producer.SendResult;
import com.alibaba.rocketmq.common.message.Message;
import com.alibaba.rocketmq.common.message.MessageQueue;
import com.alibaba.rocketmq.remoting.exception.RemotingException;

import java.text.SimpleDateFormat;
import java.util.Date;
import java.util.List;

public class ProducerQueue {
    public static void main(String[] args) {
        String group_name = "order_producer";
        DefaultMQProducer producer = new DefaultMQProducer(group_name);
        producer.setNamesrvAddr("192.168.31.176:9876;192.168.31.165:9876");
        try {
            producer.start();
            Date date = new Date();
            SimpleDateFormat sdf = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
            String dateStr = sdf.format(date);
            /**
             * 第一个队列
             * 通过  public SendResult send(Message msg, MessageQueueSelector selector, Object arg)来指定发送消息到哪个队列
             */
            for(int i = 1; i <= 5; i++) {
                String body = dateStr + "body_1_" + i;
                Message message = new Message("TopicTest", "order1", "KEY" + i, body.getBytes());
                SendResult sendResult = producer.send(message, new MessageQueueSelector() {
                    public MessageQueue select(List<MessageQueue> list, Message message, Object o) {
                        Integer id = (Integer) o;
                        return list.get(id);
                    }
                }, 0); //0是队列的下标
                System.out.println(sendResult + ", body:" + body);
            }

            producer.shutdown();

        } catch (MQClientException e) {
            e.printStackTrace();
        } catch (InterruptedException e) {
            e.printStackTrace();
        } catch (RemotingException e) {
            e.printStackTrace();
        } catch (MQBrokerException e) {
            e.printStackTrace();
        }
    }
}

Consumer端：

package org.hope.lee.producer.queue;

import com.alibaba.rocketmq.client.consumer.DefaultMQPushConsumer;
import com.alibaba.rocketmq.client.consumer.listener.ConsumeOrderlyContext;
import com.alibaba.rocketmq.client.consumer.listener.ConsumeOrderlyStatus;
import com.alibaba.rocketmq.client.consumer.listener.MessageListenerOrderly;
import com.alibaba.rocketmq.common.consumer.ConsumeFromWhere;
import com.alibaba.rocketmq.common.message.MessageExt;

import java.util.List;
import java.util.Random;
import java.util.concurrent.TimeUnit;

public class ConsumerQueue1 {

    public ConsumerQueue1() throws Exception {
        String group_name = "order_consumer";
        DefaultMQPushConsumer consumer = new DefaultMQPushConsumer(group_name);
        consumer.setNamesrvAddr("192.168.31.176:9876;192.168.31.165:9876");
        consumer.setConsumeFromWhere(ConsumeFromWhere.CONSUME_FROM_FIRST_OFFSET);
        //订阅的主题， 以及过滤的标签内容
        consumer.subscribe("TopicTest", "*");
        //注册监听
        consumer.registerMessageListener(new Listener());
        consumer.start();
        System.out.println("Consumer Started.....");
    }

    /**
     * 这里实现MessageListenerOrderLy接口就是为了达到顺序消费的目的,
     * 如果是使用MessageListenerConcurrently,则需要把线程池改为单线程模式。
     * 但是也不能保证说一定会顺序消费，因为如果master宕机了，导致写入队列的数量上
     * 出现变化。
     *
     * 从消费端，如果想保证这批消息是M1消费完成再消费M2的话，可以使用MessageListenerOrderly接口，但是这样的话会有以下问题：
     * 1. 遇到消息失败的消息，无法跳过，当前队列消费暂停
     * 2. 目前版本的RocketMQ的MessageListenerOrderly是不能从slave消费消息的。
     */
    class Listener implements MessageListenerOrderly {
        private Random random = new Random();

        public ConsumeOrderlyStatus consumeMessage(List<MessageExt> list, ConsumeOrderlyContext context) {
            context.setAutoCommit(true);
            for(MessageExt msg : list) {
                System.out.println(msg + ", content:" + new String(msg.getBody()));
            }
            try {
                TimeUnit.SECONDS.sleep(random.nextInt(5)); //随机休眠时间，模拟业务处理时间

            } catch (InterruptedException e) {
                e.printStackTrace();
            }
            return ConsumeOrderlyStatus.SUCCESS;
        }
    }
    public static void main(String[] args) throws Exception {
        ConsumerQueue1 c = new ConsumerQueue1();

    }
}

Consuerm端输出结果:（横向拖到最后看蓝色字体）

Consumer Started.....
MessageExt [queueId=0, storeSize=159, queueOffset=0, sysFlag=0, bornTimestamp=1515420522468, bornHost=/192.168.31.38:9357, storeTimestamp=1515420551480, storeHost=/192.168.31.165:10911, msgId=C0A81FA500002A9F00000000000001EF, commitLogOffset=495, bodyCRC=829956747, reconsumeTimes=0, preparedTransactionOffset=0, toString()=Message [topic=TopicTest, flag=0, properties={TAGS=order1, KEYS=KEY1, WAIT=true, MAX_OFFSET=5, MIN_OFFSET=0}, body=27]], content:2018-01-08 22:08:42body_1_1
MessageExt [queueId=0, storeSize=159, queueOffset=1, sysFlag=0, bornTimestamp=1515420522668, bornHost=/192.168.31.38:9357, storeTimestamp=1515420551683, storeHost=/192.168.31.165:10911, msgId=C0A81FA500002A9F000000000000028E, commitLogOffset=654, bodyCRC=678523697, reconsumeTimes=0, preparedTransactionOffset=0, toString()=Message [topic=TopicTest, flag=0, properties={TAGS=order1, KEYS=KEY2, WAIT=true, MAX_OFFSET=5, MIN_OFFSET=0}, body=27]], content:2018-01-08 22:08:42body_1_2
MessageExt [queueId=0, storeSize=159, queueOffset=2, sysFlag=0, bornTimestamp=1515420522781, bornHost=/192.168.31.38:9357, storeTimestamp=1515420551716, storeHost=/192.168.31.165:10911, msgId=C0A81FA500002A9F000000000000032D, commitLogOffset=813, bodyCRC=1601586087, reconsumeTimes=0, preparedTransactionOffset=0, toString()=Message [topic=TopicTest, flag=0, properties={TAGS=order1, KEYS=KEY3, WAIT=true, MAX_OFFSET=5, MIN_OFFSET=0}, body=27]], content:2018-01-08 22:08:42body_1_3
MessageExt [queueId=0, storeSize=159, queueOffset=3, sysFlag=0, bornTimestamp=1515420522792, bornHost=/192.168.31.38:9357, storeTimestamp=1515420551753, storeHost=/192.168.31.165:10911, msgId=C0A81FA500002A9F00000000000003CC, commitLogOffset=972, bodyCRC=1091753476, reconsumeTimes=0, preparedTransactionOffset=0, toString()=Message [topic=TopicTest, flag=0, properties={TAGS=order1, KEYS=KEY4, WAIT=true, MAX_OFFSET=5, MIN_OFFSET=0}, body=27]], content:2018-01-08 22:08:42body_1_4
MessageExt [queueId=0, storeSize=159, queueOffset=4, sysFlag=0, bornTimestamp=1515420522833, bornHost=/192.168.31.38:9357, storeTimestamp=1515420551768, storeHost=/192.168.31.165:10911, msgId=C0A81FA500002A9F000000000000046B, commitLogOffset=1131, bodyCRC=907404946, reconsumeTimes=0, preparedTransactionOffset=0, toString()=Message [topic=TopicTest, flag=0, properties={TAGS=order1, KEYS=KEY5, WAIT=true, MAX_OFFSET=5, MIN_OFFSET=0}, body=27]], content:2018-01-08 22:08:42body_1_5

一、Consumer端集群消费的顺序消费，也就是一个Producer多个Consumer。

步骤一、我们先改造一下ProducerQueue，在第一个队列下面再加两个队列

/**
 * 第二个队列
 */
for(int i = 1; i <= 5; i++) {
    //时间戳
    String body = dateStr + "order_2" + i;
    Message message = new Message("TopicTest", "body_2_", "KEY" + i, body.getBytes());
    SendResult sendResult = producer.send(message, new MessageQueueSelector() {
        public MessageQueue select(List<MessageQueue> list, Message message, Object o) {
            Integer id = (Integer) o;
            return list.get(id);
        }
    }, 1); //1是队列的下标
    System.out.println(sendResult + ", body:" + body);
}
/**
 * 第三个队列
 */
for(int i = 1; i <= 5; i++) {
    //时间戳
    String body = dateStr + "order_3" + i;
    Message message = new Message("TopicTest", "body_3_", "KEY" + i, body.getBytes());
    SendResult sendResult = producer.send(message, new MessageQueueSelector() {
        public MessageQueue select(List<MessageQueue> list, Message message, Object o) {
            Integer id = (Integer) o;
            return list.get(id);
        }
    }, 2); //2是队列的下标
    System.out.println(sendResult + ", body:" + body);
}

步骤二、再创建一个消费端ConsumerQueue2，代码跟ConsumerQueue1一样，这里就不重复了。

步骤三、启动ConsumerQueue1和ConsumerQueue3

步骤四、启动ProducerQueue

结果：

从结果中看到，两个消费端都是按照队列顺序消费的，并且负载均衡，ConsumerQueue1消费了第三个队列，ConsumerQueue2消费了第一个队列和第二个队列。

需要注意的一点是，对于顺序消费，我们是不能再Consumer端再使用多线程去消费的。这样就破坏了顺序消费的生态环境。