redis实现分布式锁

清理邮件的时候发现以前看的一篇关于redis分布式锁实现的文章有人回复- -当时随意扫了眼文章，为了防止发生死锁，思路是使用setnx设置value为本地时间，而后获取锁失败时读取value进行时间比对。。而后我回复了下。。多台应用服务器存在时间不一样步的问题。。

其实使用setnx时设置下redis过时时间简单方便，只是一般在应用程序内经过sdk作这项操做时因为赋值+设置过时不在同一原子性操做中。。因此不少人以为不可行了。。那就直接使用lua脚本呗，简单方便，原子性操做，性能也OK。

加锁：aquire_lock_with_timeout.lua

if redis.call("exists",KEYS[1]) == 0 then
    local lockSrc = redis.call("setex",KEYS[1],unpack(ARGV))
    if lockSrc then
        return "1"
    end
        return "0"
end
return "-1"

释放锁：release_lock.lua

if redis.call("get",KEYS[1]) == ARGV[1] then
    local lockRelease = redis.call("del",KEYS[1])
        if lockRelease then
            return "1"
        end
            return "0"
end
return "-1"

java为例，加载两个lua脚本,而后简单加锁，释放锁

  @Bean
    @Qualifier("lockScript")
    public RedisScript<Integer> acquireLockWithTimeout() {
        DefaultRedisScript redisScript = new DefaultRedisScript();
        redisScript.setLocation(new ClassPathResource("redis/acquire_lock_with_timeout.lua"));
        redisScript.setResultType(Integer.class);
        return redisScript;
    }


    @Bean
    @Qualifier("unLockScript")
    public RedisScript<Integer> releaseLock() {
        DefaultRedisScript redisScript = new DefaultRedisScript();
        redisScript.setLocation(new ClassPathResource("redis/release_lock.lua"));
        redisScript.setResultType(Integer.class);
        return redisScript;
    }

    private Integer acquireTimeout;//资源占有锁的时间 秒s
    private Integer acquireInterval;//尝试获取锁的时限 ms
    private Integer lockTime;//尝试获取锁的时间间隔 ms
    
    @Autowired
    private RedisTemplate redisTemplate;
    @Autowired
    @Qualifier("lockScript")
    private RedisScript<Integer> acquireLockWithTimeout;
    @Autowired
    @Qualifier("unLockScript")
    private RedisScript<Integer> releaseLock;

     
    public String tryLock(String lockKey) {
        String lockValue = UUID.randomUUID().toString();
        Long endTime = System.currentTimeMillis() + acquireTimeout;
        while (System.currentTimeMillis() < endTime) {
            Integer lockResult = (Integer) redisTemplate.execute(acquireLockWithTimeout, Collections.singletonList(lockKey), lockTime, lockValue);
            if (lockResult.equals(1)) {
                return lockValue;
            } else {
                try {
                    Thread.sleep(acquireInterval);
                } catch (InterruptedException ex) {
                    continue;
                }
            }
        }
        return "";
    }
 
    public boolean releaseLock(String lockKey, String lockValue) {
        Integer releaseResult = (Integer) redisTemplate.execute(releaseLock, Collections.singletonList(lockKey), lockValue);
        if (releaseResult.equals(1)) {
            return true;
        }
        return false;
    }

固然java也有更复杂更丰富的组件 redisson