Let’s talk about how to implement distributed locks in Redis that support almost all locking scenarios.

This article will introduce you to the implementation method of Redis distributed lock. I hope it will be helpful to you!

Hello everyone, today I will share with you the multi-type lock implemented by redisson, the implementation of redis distributed lock that supports almost all locking scenarios, and also supports small MQ and redis various data operations. [Related recommendations: Redis video tutorial]

Theoretical part

In the previous article, two methods of implementing distributed locks through redis were introduced The methods are:

• Through the command that comes with redis: setNX

• Through the redis client: redisson

The author recommends using the redisson client, because redisson supports more lock types, such as interlock, red lock, read-write lock, fair lock, etc., and the implementation of redisson is more Simple, developers only need to call the corresponding API, and do not need to worry about the underlying locking process and the atomicity issue of unlocking.

In the Redis distributed lock, redisson's simple implementation of various lock types is listed, that is, programmatic implementation. Such an implementation can fully meet our daily development needs, but its shortcomings are also obvious.

For example:

• There are too many codes embedded, which is not elegant enough
• Duplicate code
• The use of lock parameters is not intuitive
• It is easy to forget the unlocking steps

Students who have used Spring must have Knowing the @Transactional annotation, Spring supports both programmatic transactions and annotated (declarative) transactions.

Can we also refer to such an implementation?

The answer is: completely OK!

AOP is specialized in doing this kind of thing.

Practical part

1. Introduction of redisson dependency

    org.redisson
    redisson
    3.16.2
Copy to clipboardErrorCopied

2. Custom annotations

/**
 * 分布式锁自定义注解
 */
@Target(ElementType.METHOD)
@Retention(RetentionPolicy.RUNTIME)
@Documented
public @interface Lock {

    /**
     * 锁的模式：如果不设置自动模式，当参数只有一个.使用 REENTRANT 参数多个 MULTIPLE
     */
    LockModel lockModel() default LockModel.AUTO;

    /**
     * 如果keys有多个，如果不设置，则使用 联锁
     *
     * @return
     */
    String[] keys() default {};

    /**
     * key的静态常量：当key的spel的值是LIST、数组时使用+号连接将会被spel认为这个变量是个字符串，只能产生一把锁，达不到我们的目的，
     * 而我们如果又需要一个常量的话。这个参数将会在拼接在每个元素的后面
     *
     * @return
     */
    String keyConstant() default "";

    /**
     * 锁超时时间,默认30000毫秒(可在配置文件全局设置)
     *
     * @return
     */
    long watchDogTimeout() default 30000;

    /**
     * 等待加锁超时时间，默认10000毫秒 -1 则表示一直等待(可在配置文件全局设置)
     *
     * @return
     */
    long attemptTimeout() default 10000;
}

3. Constant class

/**
 * Redisson常量类
 */
public class RedissonConst {
    /**
     * redisson锁默认前缀
     */
    public static final String REDISSON_LOCK = "redisson:lock:";
    /**
     * spel表达式占位符
     */
    public static final String PLACE_HOLDER = "#";
}

4. Enumeration

/**
 * 锁的模式
 */
public enum LockModel {
    /**
     * 可重入锁
     */
    REENTRANT,
    /**
     * 公平锁
     */
    FAIR,
    /**
     * 联锁
     */
    MULTIPLE,
    /**
     * 红锁
     */
    RED_LOCK,
    /**
     * 读锁
     */
    READ,
    /**
     * 写锁
     */
    WRITE,
    /**
     * 自动模式，当参数只有一个使用 REENTRANT 参数多个 RED_LOCK
     */
    AUTO
}

5. Custom exception

/**
 * 分布式锁异常
 */
public class ReddissonException extends RuntimeException {

    public ReddissonException() {
    }

    public ReddissonException(String message) {
        super(message);
    }

    public ReddissonException(String message, Throwable cause) {
        super(message, cause);
    }

    public ReddissonException(Throwable cause) {
        super(cause);
    }

    public ReddissonException(String message, Throwable cause, boolean enableSuppression, boolean writableStackTrace) {
        super(message, cause, enableSuppression, writableStackTrace);
    }
}

6. AOP aspects

   /**
 * 分布式锁aop
 */
@Slf4j
@Aspect
public class LockAop {

    @Autowired
    private RedissonClient redissonClient;

    @Autowired
    private RedissonProperties redissonProperties;

    @Autowired
    private LockStrategyFactory lockStrategyFactory;

    @Around("@annotation(lock)")
    public Object aroundAdvice(ProceedingJoinPoint proceedingJoinPoint, Lock lock) throws Throwable {
        // 需要加锁的key数组
        String[] keys = lock.keys();
        if (ArrayUtil.isEmpty(keys)) {
            throw new ReddissonException("redisson lock keys不能为空");
        }
        // 获取方法的参数名
        String[] parameterNames = new LocalVariableTableParameterNameDiscoverer().getParameterNames(((MethodSignature) proceedingJoinPoint.getSignature()).getMethod());
        Object[] args = proceedingJoinPoint.getArgs();
        // 等待锁的超时时间
        long attemptTimeout = lock.attemptTimeout();
        if (attemptTimeout == 0) {
            attemptTimeout = redissonProperties.getAttemptTimeout();
        }
        // 锁超时时间
        long lockWatchdogTimeout = lock.watchdogTimeout();
        if (lockWatchdogTimeout == 0) {
            lockWatchdogTimeout = redissonProperties.getLockWatchdogTimeout();
        }
        // 加锁模式
        LockModel lockModel = getLockModel(lock, keys);
        if (!lockModel.equals(LockModel.MULTIPLE) && !lockModel.equals(LockModel.RED_LOCK) && keys.length > 1) {
            throw new ReddissonException("参数有多个，锁模式为->" + lockModel.name() + "，无法匹配加锁");
        }
        log.info("锁模式->{}，等待锁定时间->{}毫秒，锁定最长时间->{}毫秒", lockModel.name(), attemptTimeout, lockWatchdogTimeout);
        boolean res = false;
        // 策略模式获取redisson锁对象
        RLock rLock = lockStrategyFactory.createLock(lockModel, keys, parameterNames, args, lock.keyConstant(), redissonClient);
        //执行aop
        if (rLock != null) {
            try {
                if (attemptTimeout == -1) {
                    res = true;
                    //一直等待加锁
                    rLock.lock(lockWatchdogTimeout, TimeUnit.MILLISECONDS);
                } else {
                    res = rLock.tryLock(attemptTimeout, lockWatchdogTimeout, TimeUnit.MILLISECONDS);
                }
                if (res) {
                    return proceedingJoinPoint.proceed();
                } else {
                    throw new ReddissonException("获取锁失败");
                }
            } finally {
                if (res) {
                    rLock.unlock();
                }
            }
        }
        throw new ReddissonException("获取锁失败");
    }

    /**
     * 获取加锁模式
     *
     * @param lock
     * @param keys
     * @return
     */
    private LockModel getLockModel(Lock lock, String[] keys) {
        LockModel lockModel = lock.lockModel();
        // 自动模式：优先匹配全局配置，再判断用红锁还是可重入锁
        if (lockModel.equals(LockModel.AUTO)) {
            LockModel globalLockModel = redissonProperties.getLockModel();
            if (globalLockModel != null) {
                lockModel = globalLockModel;
            } else if (keys.length > 1) {
                lockModel = LockModel.RED_LOCK;
            } else {
                lockModel = LockModel.REENTRANT;
            }
        }
        return lockModel;
    }
}

Strategy pattern is used here to provide implementation for different lock types.

7. Implementation of lock strategy

First define the abstract base class of the lock strategy(you can also use interface) :

/**
 * 锁策略抽象基类
 */
@Slf4j
abstract class LockStrategy {

    @Autowired
    private RedissonClient redissonClient;

    /**
     * 创建RLock
     *
     * @param keys
     * @param parameterNames
     * @param args
     * @param keyConstant
     * @return
     */
    abstract RLock createLock(String[] keys, String[] parameterNames, Object[] args, String keyConstant, RedissonClient redissonClient);

    /**
     * 获取RLock
     *
     * @param keys
     * @param parameterNames
     * @param args
     * @param keyConstant
     * @return
     */
    public RLock[] getRLocks(String[] keys, String[] parameterNames, Object[] args, String keyConstant) {
        List rLocks = new ArrayList<>();
        for (String key : keys) {
            List valueBySpel = getValueBySpel(key, parameterNames, args, keyConstant);
            for (String s : valueBySpel) {
                rLocks.add(redissonClient.getLock(s));
            }
        }
        RLock[] locks = new RLock[rLocks.size()];
        int index = 0;
        for (RLock r : rLocks) {
            locks[index++] = r;
        }
        return locks;
    }

    /**
     * 通过spring Spel 获取参数
     *
     * @param key            定义的key值 以#开头 例如:#user
     * @param parameterNames 形参
     * @param args           形参值
     * @param keyConstant    key的常亮
     * @return
     */
    List getValueBySpel(String key, String[] parameterNames, Object[] args, String keyConstant) {
        List keys = new ArrayList<>();
        if (!key.contains(PLACE_HOLDER)) {
            String s = REDISSON_LOCK + key + keyConstant;
            log.info("没有使用spel表达式value->{}", s);
            keys.add(s);
            return keys;
        }
        // spel解析器
        ExpressionParser parser = new SpelExpressionParser();
        // spel上下文
        EvaluationContext context = new StandardEvaluationContext();
        for (int i = 0; i < parameterNames.length; i++) {
            context.setVariable(parameterNames[i], args[i]);
        }
        Expression expression = parser.parseExpression(key);
        Object value = expression.getValue(context);
        if (value != null) {
            if (value instanceof List) {
                List valueList = (List) value;
                for (Object o : valueList) {
                    keys.add(REDISSON_LOCK + o.toString() + keyConstant);
                }
            } else if (value.getClass().isArray()) {
                Object[] objects = (Object[]) value;
                for (Object o : objects) {
                    keys.add(REDISSON_LOCK + o.toString() + keyConstant);
                }
            } else {
                keys.add(REDISSON_LOCK + value.toString() + keyConstant);
            }
        }
        log.info("spel表达式key={},value={}", key, keys);
        return keys;
    }
}

Then provide the specific implementation of various lock modes:

• Reentrant lock:

/**
 * 可重入锁策略
 */
public class ReentrantLockStrategy extends LockStrategy {

    @Override
    public RLock createLock(String[] keys, String[] parameterNames, Object[] args, String keyConstant, RedissonClient redissonClient) {
        List valueBySpel = getValueBySpel(keys[0], parameterNames, args, keyConstant);
        //如果spel表达式是数组或者集合 则使用红锁
        if (valueBySpel.size() == 1) {
            return redissonClient.getLock(valueBySpel.get(0));
        } else {
            RLock[] locks = new RLock[valueBySpel.size()];
            int index = 0;
            for (String s : valueBySpel) {
                locks[index++] = redissonClient.getLock(s);
            }
            return new RedissonRedLock(locks);
        }
    }
}

• Fair Lock:

/**
 * 公平锁策略
 */
public class FairLockStrategy extends LockStrategy {

    @Override
    public RLock createLock(String[] keys, String[] parameterNames, Object[] args, String keyConstant, RedissonClient redissonClient) {
        return redissonClient.getFairLock(getValueBySpel(keys[0], parameterNames, args, keyConstant).get(0));
    }
}

• Interlocking

/**
 * 联锁策略
 */
public class MultipleLockStrategy extends LockStrategy {

    @Override
    public RLock createLock(String[] keys, String[] parameterNames, Object[] args, String keyConstant, RedissonClient redissonClient) {
        RLock[] locks = getRLocks(keys, parameterNames, args, keyConstant);
        return new RedissonMultiLock(locks);
    }
}

• Red Lock

/**
 * 红锁策略
 */
public class RedLockStrategy extends LockStrategy {

    @Override
    public RLock createLock(String[] keys, String[] parameterNames, Object[] args, String keyConstant, RedissonClient redissonClient) {
        RLock[] locks = getRLocks(keys, parameterNames, args, keyConstant);
        return new RedissonRedLock(locks);
    }
}

• Read lock

/**
 * 读锁策略
 */
public class ReadLockStrategy extends LockStrategy {

    @Override
    public RLock createLock(String[] keys, String[] parameterNames, Object[] args, String keyConstant, RedissonClient redissonClient) {
        RReadWriteLock rwLock = redissonClient.getReadWriteLock(getValueBySpel(keys[0], parameterNames, args, keyConstant).get(0));
        return rwLock.readLock();
    }
}

• Write lock

/**
 * 写锁策略
 */
public class WriteLockStrategy extends LockStrategy {

    @Override
    public RLock createLock(String[] keys, String[] parameterNames, Object[] args, String keyConstant, RedissonClient redissonClient) {
        RReadWriteLock rwLock = redissonClient.getReadWriteLock(getValueBySpel(keys[0], parameterNames, args, keyConstant).get(0));
        return rwLock.writeLock();
    }
}

Finally provides aStrategy FactoryInitialization lock strategy:

/**
 * 锁的策略工厂
 */
@Service
public class LockStrategyFactory {

    private LockStrategyFactory() {
    }

    private static final Map STRATEGIES = new HashMap<>(6);

    static {
        STRATEGIES.put(LockModel.FAIR, new FairLockStrategy());
        STRATEGIES.put(LockModel.REENTRANT, new ReentrantLockStrategy());
        STRATEGIES.put(LockModel.RED_LOCK, new RedLockStrategy());
        STRATEGIES.put(LockModel.READ, new ReadLockStrategy());
        STRATEGIES.put(LockModel.WRITE, new WriteLockStrategy());
        STRATEGIES.put(LockModel.MULTIPLE, new MultipleLockStrategy());
    }

    public RLock createLock(LockModel lockModel, String[] keys, String[] parameterNames, Object[] args, String keyConstant, RedissonClient redissonClient) {
        return STRATEGIES.get(lockModel).createLock(keys, parameterNames, args, keyConstant, redissonClient);
    }
}

8. How to use

    @Lock(keys = "#query.channel") // 支持spel
    @ApiOperation("分页列表")
    @GetMapping
    public ApiPageResult list(VendorProjectItemQuery query, Pagination pagination) {
        return ApiPageResult.success(pagination, vendorProjectItemService.list(query, pagination), vendorProjectItemService.count(query));
    }

You’re done, eat chicken smoothly

For more programming-related knowledge, please visit: Programming Video! !

The above is the detailed content of Let’s talk about how to implement distributed locks in Redis that support almost all locking scenarios.. For more information, please follow other related articles on the PHP Chinese website!