Redis做分布式无锁CAS的问题

最新推荐文章于 2023-03-30 10:17:12 发布

weixin_33950035

最新推荐文章于 2023-03-30 10:17:12 发布

阅读量218

点赞数

文章标签：数据库 python

原文链接：https://my.oschina.net/u/3768341/blog/1854578

版权

2019独角兽企业重金招聘Python工程师标准>>>

因为Redis本身是单线程的，具备原子性，所以可以用来做分布式无锁的操作，但会有一点小问题。

public interface OrderService {
    public String getOrderNo();
}

public class OrderRedisServiceImpl implements OrderService {
    static JedisPool jedisPool;
    static {
        JedisPoolConfig jedisPoolConfig = new JedisPoolConfig();
        jedisPool = new JedisPool(jedisPoolConfig,"XXX.XXX.XXX.XXX",6379,0);
    }
    @Override
    public String getOrderNo() {
        try {
            Jedis jedis = jedisPool.getResource();
            SimpleDateFormat date = new SimpleDateFormat("YYYYMMDDHHMMSS");
            return date.format(new Date()) + jedis.incr("order_keys");
        }finally {
            jedisPool.close();
        }
    }
}

redis实现的接口

public class OrderTask implements Runnable {
    private CountDownLatch latch;
    private OrderService orderService;
    public OrderTask(CountDownLatch latch,OrderService orderService) {
        this.latch = latch;
        this.orderService = orderService;
    }

    @Override
    public void run() {
        try {
            latch.await();
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
        System.out.printf("线程名%s订单号:%s\n",Thread.currentThread().getName(), orderService.getOrderNo());
    }
}

线程任务类

public static void main(String[] args) {
   ExecutorService service = Executors.newCachedThreadPool();
   final CountDownLatch latch = new CountDownLatch(1);
   OrderService orderService = new OrderRedisServiceImpl();
   for (int i = 0;i < 10;i++) {
      service.submit(new OrderTask(latch,orderService));
   }
   latch.countDown();
   service.shutdown();
}

运行结果如下

线程名pool-2-thread-5订单号:20180719623079251
线程名pool-2-thread-3订单号:20180719623079252
线程名pool-2-thread-4订单号:201807196230710755
线程名pool-2-thread-7订单号:201807196230710754
线程名pool-2-thread-9订单号:201807196230710756
线程名pool-2-thread-2订单号:201807196230710757
线程名pool-2-thread-8订单号:201807196230710753
线程名pool-2-thread-1订单号:201807196230730458

他只跑出了8个结果，我们的确是运行了10次，由此我们记得jedis的默认最大连接数为8，所以只要我们的运行的线程数大于8，也只会跑出8个结果。

所以我们要修改如下

public class OrderRedisServiceImpl implements OrderService {
    static JedisPool jedisPool;
    static {
        JedisPoolConfig jedisPoolConfig = new JedisPoolConfig();
        jedisPoolConfig.setMaxIdle(-1);
        jedisPoolConfig.setMaxTotal(-1);
        jedisPool = new JedisPool(jedisPoolConfig,"XXX.XXX.XXX.XXX",6379,0);
    }
    @Override
    public String getOrderNo() {
        try {
            Jedis jedis = jedisPool.getResource();
            SimpleDateFormat date = new SimpleDateFormat("YYYYMMDDHHMMSS");
            return date.format(new Date()) + jedis.incr("order_keys");
        }finally {
            jedisPool.close();
        }
    }
}

把redis连接池的最大连接数增大，运行结果如下

线程名pool-2-thread-4订单号:201807196230758759
线程名pool-2-thread-5订单号:201807196230758860
线程名pool-2-thread-2订单号:201807196230758961
线程名pool-2-thread-3订单号:201807196230758862
线程名pool-2-thread-7订单号:201807196230758763
线程名pool-2-thread-8订单号:201807196230758764
线程名pool-2-thread-10订单号:201807196230758765
线程名pool-2-thread-6订单号:201807196230758866
线程名pool-2-thread-1订单号:201807196230758768
线程名pool-2-thread-9订单号:201807196230759067

这一次就是10个结果了。

转载于:https://my.oschina.net/u/3768341/blog/1854578