redis用list做消息队列

leftPush消息入队，rightPop对应，消息出队。
rightPop(RedisConstant.MQ_LIST, 0L, TimeUnit.SECONDS)阻塞出队，0表示永久阻塞

生产消息服务

@Service
public class RedisService {
    @Autowired
    private RedisTemplate<String, String> redisTemplate;


    public Object publish() {
        OrderDTO dto = new OrderDTO();
        dto.setId(1);
        dto.setCreateTime(new Date());
        dto.setMoney("12.34");
        dto.setOrderNo("orderNo1");
        String s = JSON.toJSONString(dto);

        ListOperations<String, String> listOperations = redisTemplate.opsForList();
        //leftPush和rightPop对应，左边入队，右边出队
        listOperations.leftPush(RedisConstant.MQ_LIST, s);

        //因为出队是阻塞读取的，所以上一步入队后，数据立刻就被驱走了，下一步size=0
        Long size = listOperations.size(RedisConstant.MQ_LIST);
        List<String> list = new ArrayList<>();
        if (size != null && size > 0) {
             list = listOperations.range(RedisConstant.MQ_LIST, 0, size - 1);
        }
        return list;

    }
}

测试

@RestController
@RequestMapping("redisList")
public class RedisListController {

    @Autowired
    private RedisService redisService;

    @GetMapping("publish")
    public Object publish() {
        return redisService.publish();
    }
}

消费消息服务，定时任务

@Component
public class RedisConsumeTask {
    @Autowired
    private RedisService redisService;

    @TaskLock(RedisConstant.CONSUME_REDIS_LIST)
    @Scheduled(cron = "0/10 * * * * ?")
    public void consumeMqList() {
        redisService.consumeMqList();
    }
}

@Service
@Slf4j
public class RedisService {

    @Autowired
    private RedisTemplate<String, String> redisTemplate;

    public void consumeMqList() {
        ListOperations<String, String> listOperations = redisTemplate.opsForList();
        //0时间，表示阻塞永久
        //待机一小时后，再次发消息，消费不了了，阻塞有问题啊。还得轮寻啊
        //String s = listOperations.rightPop(RedisConstant.MQ_LIST, 0L, TimeUnit.SECONDS);
        String s = listOperations.rightPop(RedisConstant.MQ_LIST);
        if (s == null) {
            return;
        }

        log.info("{} = {}", RedisConstant.MQ_LIST, s);

        OrderDTO dto = JSON.parseObject(s, OrderDTO.class);
        log.info("dto = {}", dto);
    }
}

日志

@Component
@Aspect
public class TaskLockAop {

    @Autowired
    private RedisLockRegistry redisLockRegistry;

    @Around("execution(@TaskLock * * (..))")
    public Object taskAround(ProceedingJoinPoint pjp) throws Throwable {

        TaskLock taskAnnotation = ((MethodSignature)pjp.getSignature()).getMethod().getAnnotation(TaskLock.class);

        String lockKey = taskAnnotation.value();
        Lock lock = redisLockRegistry.obtain(lockKey);
        try {
            lock.tryLock(30L, TimeUnit.SECONDS);
            System.out.println("任务开始, " + lockKey + ", " + new Date());

            return pjp.proceed();

        } finally {
            lock.unlock();
            System.out.println("任务结束, " + lockKey + ", " + new Date());
        }
    }


}

测试

http://localhost:9040/redisList/publish

["{“createTime”:1574394538430,“id”:1,“money”:“12.34”,“orderNo”:“orderNo1”}"]

下面一直阻塞，任务开始了，不收到消息，永远不会结束。
阻塞有问题，改用轮询了。

先启动发送消息服务，发送消息。后启动消费消息服务，可以消费消息。这一点，比发布订阅要稳定。

关联项目https://github.com/mingwulipo/cloud-demo.git