Redis分布式实现原理

一、使用

1、pom.xml导入依赖

<dependency>
	<groupId>org.springframework.bootgroupId>
	<artifactId>spring-boot-starter-integrationartifactId>
dependency>
<dependency>
	<groupId>org.springframework.integrationgroupId>
	<artifactId>spring-integration-redisartifactId>
dependency>
<dependency>
	<groupId>org.springframework.bootgroupId>
	<artifactId>spring-boot-starter-data-redisartifactId>
dependency>

2、配置文件

@Configuration
public class RedissonConfig {

	// 自定义在yml或properties文件中
	@Value("${spring.redis.host}")
	private String host;

	@Value("${spring.redis.port}")
	private String port;

	@Value("${spring.redis.password}")
	private String password;

	@Bean
	public RedissonClient getRedisson() {
		Config config = new Config();
		if (StringUtils.isBlank(password)) {
			config.useSingleServer().setAddress("redis://" + host + ":" + port);
		} else {
			config.useSingleServer().setAddress("redis://" + host + ":" + port).setPassword(password);
		}
		return Redisson.create(config);
	}

}

3、使用类

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.integration.redis.util.RedisLockRegistry;

import java.util.concurrent.TimeUnit;
import java.util.concurrent.locks.Lock;

/**
 * @author Forlan
 * @since 2022-04-03
 */
public class ForlanTest {

	private static final Logger logger = LoggerFactory.getLogger(ForlanTest.class);

	@Autowired
	private RedisLockRegistry redisLockRegistry;

	public void test() {
		String lockKey = "key_id";
		Lock loginLock = redisLockRegistry.obtain(lockKey.intern());
		boolean getLock = false;
		try {
			// tryLock()底层调用this.tryLock(0L, TimeUnit.MILLISECONDS)
			getLock = loginLock.tryLock(5, TimeUnit.SECONDS);
			if (getLock) {
				//获得锁执行业务
			}
		} catch (Exception e) {
			logger.error("异常信息...", e);
		} finally {
			if (getLock) {
				//释放锁
				loginLock.unlock();
			}
		}
	}
}

二、原理

obtain方法

private final Map locks;

private final class RedisLock implements Lock {
	private final String lockKey;
	private final ReentrantLock localLock;
	private volatile long lockedAt;

	private RedisLock(String path) {
		this.localLock = new ReentrantLock();
		this.lockKey = this.constructLockKey(path);
	}
}

public Lock obtain(Object lockKey) {
    Assert.isInstanceOf(String.class, lockKey);
    String path = (String)lockKey;
    return (Lock)this.locks.computeIfAbsent(path, (x$0) -> {
        return new RedisLockRegistry.RedisLock(x$0);
    });
}

主要是根据lockKey去查locks这个map中是否已经存在这个key
如果存在就返回内部类RedisLock
如果不存在就创建一个RedisLock，以lockKey为key，RedisLock为value放入map中

备注：每个分布式应用自己都会创建一个RedisLockRegistry实例，同一个应用的多个线程共享RedisLock类

tryLock方法

public boolean tryLock(long time, TimeUnit unit) throws InterruptedException {
	long now = System.currentTimeMillis();
	// 尝试拿取本地锁
	if (!this.localLock.tryLock(time, unit)) {
		return false;
	} else {
		try {
			long expire = now + TimeUnit.MILLISECONDS.convert(time, unit);

			boolean acquired;
			// 当前时间还没过期并且还未获得redis锁，睡眠100ms继续重试
			while(!(acquired = this.obtainLock()) && System.currentTimeMillis() 100L);
			}

			if (!acquired) {
				this.localLock.unlock();
			}

			return acquired;
		} catch (Exception var9) {
			this.localLock.unlock();
			this.rethrowAsLockException(var9);
			return false;
		}
	}
}

主要过程

先获得本地锁，拿不到直接返回失败
当前时间还没过期并且还没拿到redis锁，睡眠100ms继续重试

如果拿到redis锁，结束循环，返回成功

如果超时了还没拿到，释放锁，返回失败

拿redis锁的过程

private boolean obtainLock() {
	boolean success = (Boolean)RedisLockRegistry.this.redisTemplate.execute(RedisLockRegistry.this.obtainLockScript, Collections.singletonList(this.lockKey), new Object[]{RedisLockRegistry.this.clientId, String.valueOf(RedisLockRegistry.this.expireAfter)});
	if (success) {
		this.lockedAt = System.currentTimeMillis();
	}

	return success;
}

通过obtainLock方法，执行lua脚本来获取

redisTemplate.execute()参数说明：

第一个obtainLockScript参数就是要执行的lua脚本；

local lockClientId = redis.call('GET', KEYS[1])
if lockClientId == ARGV[1] then
  redis.call('PEXPIRE', KEYS[1], ARGV[2])
  return true
elseif not lockClientId then
  redis.call('SET', KEYS[1], ARGV[1], 'PX', ARGV[2])
  return true
end
return false

第二个参数就是表示在脚本中所用到的那些 Redis 键(key)，这些键名参数可以在 Lua 中通过全局变量 KEYS 数组，用1为基址的形式访问( KEYS[1] ， KEYS[2] ，以此类推)；

第三个参是附加参数 arg [arg …] ，可以在 Lua 中通过全局变量 ARGV 数组访问，访问的形式和 KEYS 变量类似( ARGV[1] 、 ARGV[2] ，诸如此类)

为什么要用本地锁

为了可重入
为了减轻redis服务器的压力

为什么要用lua脚本

保证原子性
减少网络开销
替代redis的事务功能

unlock方法

public void unlock() {
    if (!this.localLock.isHeldByCurrentThread()) {
        throw new IllegalStateException("You do not own lock at " + this.lockKey);
    } else if (this.localLock.getHoldCount() > 1) {
        this.localLock.unlock();
    } else {
        try {
            if (Thread.currentThread().isInterrupted()) {
                RedisLockRegistry.this.executor.execute(this::removeLockKey);
            } else {
                this.removeLockKey();
            }
            if (RedisLockRegistry.logger.isDebugEnabled()) {
                RedisLockRegistry.logger.debug("Released lock; " + this);
            }
        } catch (Exception var5) {
            ReflectionUtils.rethrowRuntimeException(var5);
        } finally {
            this.localLock.unlock();
        }
    }
}