Spring Boot中使用Redisson

    微服务项目中,很多资源需要互斥使用,比如一些分布式任务,比如下单的处理,退货的处理等等。这些都需要用到借助分布式锁来保证处理的唯一性。 一开始我们也手工实现了分布式锁,但是随着业务的发展,我们对锁的特性也要求越来越完善,最后选用了Redis官方推荐的Redisson。

一、Spring Boot中使用Redisson

Spring Boot使用Redisson特别简单,只要引入一个依赖就可以,redis的配置跟其他的redis客户端可以兼容,可以不用再额外配置
二、引入依赖

 

    <dependency>
        <groupId>org.redisson</groupId>
        <artifactId>redisson-spring-boot-starter</artifactId>
        <version>3.13.2</version>
    </dependency>

三、属性文件Redis配置

 

    # Redis服务器地址
    spring.redis.host=127.0.0.1
    # Redis服务器连接端口
    spring.redis.port=6379

四、快速入门
4.1 改造RedisDistributedLockApplication启动类

使用锁RedissonClient,并实现业务逻辑在ApplicationRunner#run()方法。

 

    package com.erbadagang.springboot.redisdistributedlock;
     
    import lombok.extern.slf4j.Slf4j;
    import org.redisson.api.RLock;
    import org.redisson.api.RedissonClient;
    import org.springframework.boot.ApplicationArguments;
    import org.springframework.boot.ApplicationRunner;
    import org.springframework.boot.SpringApplication;
    import org.springframework.boot.autoconfigure.SpringBootApplication;
     
    import javax.annotation.Resource;
     
    @Slf4j
    @SpringBootApplication
    public class RedisDistributedLockApplication implements ApplicationRunner {
     
        public static void main(String[] args) {
            SpringApplication.run(RedisDistributedLockApplication.class, args);
        }
     
        /**
         * 直接注入RedissonClient就可以直接使用.
         */
        @Resource
        private RedissonClient redissonClient;
     
        @Override
        public void run(ApplicationArguments args) throws Exception {
            log.info("spring boot run");
     
            //创建锁
            RLock helloLock = redissonClient.getLock("hello");
     
            //加锁
            helloLock.lock();
            try {
                log.info("locked");
                Thread.sleep(1000 * 10);
            } finally {
                //释放锁
                helloLock.unlock();
            }
            log.info("finished");
        }
    }

4.2 测试

启动Redis和RedisDistributedLockApplication,控制台输出:

 

    2020-08-02 22:51:17.169  INFO 8876 --- [main] c.e.s.r.RedisDistributedLockApplication  : spring boot run
    2020-08-02 22:51:36.486  INFO 8876 --- [main] c.e.s.r.RedisDistributedLockApplication  : locked
    2020-08-02 22:51:46.493  INFO 8876 --- [main] c.e.s.r.RedisDistributedLockApplication  : finished

4.3 Rlock 常用的方法

 

    void lock();
    void lock(long leaseTime, TimeUnit unit);
    boolean tryLock(long time, TimeUnit unit) throws InterruptedException;
    boolean tryLock(long waitTime, long leaseTime, TimeUnit unit) throws InterruptedException;

    第一个方法void lock():第一表示lock表示去加锁,加锁成功,没有返回值,继续执行下面代码;但是如果redis已经有这个锁了,它会一直阻塞,直到锁的时间失效(默认30秒),再继续往下执行。这个方法是要保证一定要抢到锁的,它的默认过期时间也是30秒,和tryLock()不同的是,它如果没抢占到锁,会一直自旋。
    第二个方法void lock(long leaseTime, TimeUnit unit):和第一无参数lock逻辑一样,只是可以直接设置锁失效时间。用法:helloLock.lock(5, TimeUnit.SECONDS);。
    第三个方法两个参数的boolean tryLock(long time, TimeUnit unit)表示尝试去加锁(第一个参数表示the maximum time to wait for the lock),加锁成功,返回true,继续执行true下面代码;但是如果redis已经有这个锁了他会等待,还拿不到锁它会返回false,执行false的代码块。为了实现waitTime,使用了redis的订阅发布功能。也就是没有抢到锁的线程订阅消息,直至waitTime过期返回false或者被通知新一轮的开始抢占锁。当然,它如果抢占到锁,锁的过期时间也是30秒,同样也会存在一个定时任务续过期时间,保证业务执行时间不会超过过期时间,抢占失败即返回false。

 

            String key ="product:001";
            RLock lock = redisson.getLock(key);
            try {
                boolean res = lock.tryLock(10,TimeUnit.SECONDS);
                if ( res){
                    System.out.println("这里是你的业务代码");
                }else{
                    System.out.println("系统繁忙");
                }
            }catch (Exception e){
                e.printStackTrace();
            }finally {
                lock.unlock();
            }

如果把lock.unlock();注释,第一次执行正常加锁,可以跑到业务逻辑代码,快速第二次执行发现他等待10秒,如果拿不到锁就走else的系统繁忙逻辑。

    三个参数的tryLock(long waitTime, long leaseTime, TimeUnit unit)表示尝试去加锁(第一个参数表示等待时间,第二个参数表示key的失效时间),加锁成功,返回true,继续执行true下面代码;如果返回false,它会等待第一个参数设置的时间,然后去执行false下面的代码。个方法的参数leaseTime如果不是-1的话,是不会有定时任务续过期时间的,也就存在业务处理时间可能超过过期时间的风险。其他的和tryLock(long waitTime, TimeUnit unit)一致。

 

 boolean res = lock.tryLock(5,3, TimeUnit.SECONDS);

这种情况,锁3秒失效,我们配置的是等待5秒,在单机刷的情况下,肯定每次都能拿到锁。
4.4 异步执行分布式锁

 

            /**
             * 异步锁
             */
            lock = redissonClient.getLock("erbadagang-lock");
            Future<Boolean> res = null;
            try {
                // lock.lockAsync();
                // lock.lockAsync(100, TimeUnit.SECONDS);
                res = lock.tryLockAsync(3, 100, TimeUnit.SECONDS);
                if (res.get()) {
                    System.out.println("这里是你的Async业务代码");
                } else {
                    System.out.println("系统繁忙Async");
                }
            } catch (Exception e) {
                e.printStackTrace();
            } finally {
                if (res.get()) {
                    lock.unlock();
                }
            }
            log.info("finished");
        }

4.5 公平锁(Fair Lock)

Redisson分布式可重入公平锁也是实现了java.util.concurrent.locks.Lock接口的一种RLock对象。在提供了自动过期解锁功能的同时,保证了当多个Redisson客户端线程同时请求加锁时,优先分配给先发出请求的线程。

 

        /**
         * 公平锁测试。
         */
        @Test
        public void testFairLock() {
            RLock fairLock = redissonClient.getFairLock("anyLock");
            try {
                // 最常见的使用方法
                fairLock.lock();
                // 支持过期解锁功能, 10秒钟以后自动解锁,无需调用unlock方法手动解锁
                fairLock.lock(10, TimeUnit.SECONDS);
                // 尝试加锁,最多等待100秒,上锁以后10秒自动解锁
                boolean res = fairLock.tryLock(100, 10, TimeUnit.SECONDS);
                if (res) {
                    System.out.println("这里是你的业务代码");
                } else {
                    System.out.println("系统繁忙");
                }
            } catch (InterruptedException e) {
                e.printStackTrace();
            } finally {
                fairLock.unlock();
            }
        }

Redisson同时还为分布式可重入公平锁提供了异步执行的相关方法:

 

    RLock fairLock = redisson.getFairLock("anyLock");
    fairLock.lockAsync();
    fairLock.lockAsync(10, TimeUnit.SECONDS);
    Future<Boolean> res = fairLock.tryLockAsync(100, 10, TimeUnit.SECONDS);

五、扩展实现

Redis单节点配置:

 

    server:
      port: 8080
     
    spring:
      application:
        name: redis-distributed-lock
     
      ################ Redis  ##############
      redis:
        host: 127.0.0.1
        port: 6379
        #password:
        timeout: 3000
        lettuce:
          pool:
            max-active: 8
            max-wait: -1
            max-idle: 8
            min-idle: 0
        redisson:
          config:
            # 单节点配置
            singleServerConfig:
              # 连接空闲超时,单位:毫秒
              idleConnectionTimeout: 10000
              pingTimeout: 1000
              # 连接超时,单位:毫秒
              connectTimeout: 10000
              # 命令等待超时,单位:毫秒
              timeout: 3000
              # 命令失败重试次数,如果尝试达到 retryAttempts(命令失败重试次数) 仍然不能将命令发送至某个指定的节点时,将抛出错误。
              # 如果尝试在此限制之内发送成功,则开始启用 timeout(命令等待超时) 计时。
              retryAttempts: 3
              # 命令重试发送时间间隔,单位:毫秒
              retryInterval: 1500
              # 重新连接时间间隔,单位:毫秒
              reconnectionTimeout: 3000
              # 执行失败最大次数
              failedAttempts: 3
              # 密码
              password: null
              # 单个连接最大订阅数量
              subscriptionsPerConnection: 5
              # 客户端名称
              clientName: null
              # 节点地址
              address: redis://127.0.0.1:6379
              # 发布和订阅连接的最小空闲连接数
              subscriptionConnectionMinimumIdleSize: 1
              # 发布和订阅连接池大小
              subscriptionConnectionPoolSize: 50
              # 最小空闲连接数
              connectionMinimumIdleSize: 32
              # 连接池大小
              connectionPoolSize: 64
              # 数据库编号
              database: 0
              # DNS监测时间间隔,单位:毫秒
              dnsMonitoringInterval: 5000
            # 线程池数量,默认值: 当前处理核数量 * 2
            threads: 0
            # Netty线程池数量,默认值: 当前处理核数量 * 2
            nettyThreads: 0
            # 编码
            #codec: !<org.redisson.codec.JsonJacksonCodec> {}
            # 传输模式
            transportMode: "NIO"

Redis集群配置:

 

    spring:
      redis:
        redisson:
          config:
            clusterServersConfig:
              idleConnectionTimeout: 10000
              connectTimeout: 10000
              timeout: 3000
              retryAttempts: 3
              retryInterval: 1500
              failedSlaveReconnectionInterval: 3000
              failedSlaveCheckInterval: 60000
              password: null
              subscriptionsPerConnection: 5
              clientName: null
              loadBalancer: !<org.redisson.connection.balancer.RoundRobinLoadBalancer> {}
              subscriptionConnectionMinimumIdleSize: 1
              subscriptionConnectionPoolSize: 50
              slaveConnectionMinimumIdleSize: 24
              slaveConnectionPoolSize: 64
              masterConnectionMinimumIdleSize: 24
              masterConnectionPoolSize: 64
              readMode: "SLAVE"
              subscriptionMode: "SLAVE"
              nodeAddresses:
              - "redis://192.168.35.142:7002"
              - "redis://192.168.35.142:7001"
              - "redis://192.168.35.142:7000"
              scanInterval: 1000
              pingConnectionInterval: 0
              keepAlive: false
              tcpNoDelay: false
            threads: 16
            nettyThreads: 32
            #codec: !<org.redisson.codec.FstCodec> {}
            transportMode: "NIO"

多线程测试:

 

    package com.erbadagang.springboot.redisdistributedlock;
     
    import lombok.extern.slf4j.Slf4j;
    import org.junit.jupiter.api.Test;
    import org.redisson.api.RLock;
    import org.redisson.api.RedissonClient;
    import org.springframework.boot.test.context.SpringBootTest;
     
    import javax.annotation.Resource;
    import java.util.concurrent.CountDownLatch;
     
    @SpringBootTest
    @Slf4j
    class RedisDistributedLockApplicationTests {
     
     
        /**
         * 有锁测试共享变量
         */
        private Integer lockCount = 10;
     
        /**
         * 无锁测试共享变量
         */
        private Integer count = 10;
     
        /**
         * 模拟线程数
         */
        private static int threadNum = 10;
     
        /**
         * 直接注入RedissonClient就可以直接使用.
         */
        @Resource
        private RedissonClient redissonClient;
     
        /**
         * 模拟并发测试加锁和不加锁2个方法。
         *
         * @return
         */
        @Test
        public void lock() {
            // 计数器
            final CountDownLatch countDownLatch = new CountDownLatch(1);
            for (int i = 0; i < threadNum; i++) {
                MyRunnable myRunnable = new MyRunnable(countDownLatch);
                Thread myThread = new Thread(myRunnable);
                myThread.start();
            }
            // 释放所有线程
            countDownLatch.countDown();
        }
     
        /**
         * 加锁测试
         */
        private void testLockCount() {
            String lockKey = "lock-test";
            //创建锁
            RLock helloLock = redissonClient.getLock(lockKey);
     
            try {
                //加锁
                helloLock.lock();
                lockCount--;
                log.info("lockCount值:" + lockCount);
            } catch (Exception e) {
                log.error(e.getMessage(), e);
            } finally {
                // 释放锁
                helloLock.unlock();
            }
        }
     
        /**
         * 无锁测试
         */
        private void testCount() {
            count--;
            log.info("count值:" + count);
        }
     
     
        public class MyRunnable implements Runnable {
            /**
             * 计数器
             */
            final CountDownLatch countDownLatch;
     
            public MyRunnable(CountDownLatch countDownLatch) {
                this.countDownLatch = countDownLatch;
            }
     
            @Override
            public void run() {
                try {
                    // 阻塞当前线程,直到计时器的值为0
                    countDownLatch.await();
                } catch (InterruptedException e) {
                    log.error(e.getMessage(), e);
                }
                // 无锁操作
                testCount();
                // 加锁操作
                testLockCount();
            }
     
        }
     
    }

控制台输出:

 

    2020-08-02 23:55:39.832  INFO 4144 --- [     Thread-281] s.r.RedisDistributedLockApplicationTests : count值:3
    2020-08-02 23:55:39.832  INFO 4144 --- [     Thread-283] s.r.RedisDistributedLockApplicationTests : count值:2
    2020-08-02 23:55:39.832  INFO 4144 --- [     Thread-279] s.r.RedisDistributedLockApplicationTests : count值:5
    2020-08-02 23:55:39.832  INFO 4144 --- [     Thread-282] s.r.RedisDistributedLockApplicationTests : count值:2
    2020-08-02 23:55:39.832  INFO 4144 --- [     Thread-284] s.r.RedisDistributedLockApplicationTests : count值:2
    2020-08-02 23:55:39.832  INFO 4144 --- [     Thread-278] s.r.RedisDistributedLockApplicationTests : count值:5
    2020-08-02 23:55:39.832  INFO 4144 --- [     Thread-280] s.r.RedisDistributedLockApplicationTests : count值:4
    2020-08-02 23:55:39.832  INFO 4144 --- [     Thread-275] s.r.RedisDistributedLockApplicationTests : count值:2
    2020-08-02 23:55:39.832  INFO 4144 --- [     Thread-276] s.r.RedisDistributedLockApplicationTests : count值:2
    2020-08-02 23:55:39.832  INFO 4144 --- [     Thread-277] s.r.RedisDistributedLockApplicationTests : count值:2
    2020-08-02 23:55:39.848  INFO 4144 --- [     Thread-280] s.r.RedisDistributedLockApplicationTests : lockCount值:9
    2020-08-02 23:55:39.848  INFO 4144 --- [     Thread-275] s.r.RedisDistributedLockApplicationTests : lockCount值:8
     
    2020-08-02 23:55:39.883  INFO 4144 --- [     Thread-281] s.r.RedisDistributedLockApplicationTests : lockCount值:7
    2020-08-02 23:55:39.885  INFO 4144 --- [extShutdownHook] o.s.s.concurrent.ThreadPoolTaskExecutor  : Shutting down ExecutorService 'applicationTaskExecutor'
    2020-08-02 23:55:39.885  INFO 4144 --- [     Thread-279] s.r.RedisDistributedLockApplicationTests : lockCount值:6
    2020-08-02 23:55:39.885  INFO 4144 --- [     Thread-277] s.r.RedisDistributedLockApplicationTests : lockCount值:5
    2020-08-02 23:55:39.885  INFO 4144 --- [     Thread-284] s.r.RedisDistributedLockApplicationTests : lockCount值:4
    2020-08-02 23:55:39.903  INFO 4144 --- [     Thread-282] s.r.RedisDistributedLockApplicationTests : lockCount值:3

根据打印结果可以明显看到,未加锁的count--后值是乱序的,而加锁后的结果和我们预期的一样。由于条件问题没办法测试分布式的并发。只能模拟单服务的这种并发,但是原理是一样

posted @ 2023-02-24 15:31  甜菜波波  阅读(2944)  评论(0编辑  收藏  举报