Redis based distributed lock

Distributed lock

The reason for using distributed locks

Distributed lock is a way to ensure synchronous access to shared resources between distributed systems

matters needing attention

When acquiring the lock , Use SET name value NX EX Time Ensure atomicity when acquiring locks , And the lock is automatically released in case of downtime

When releasing the lock , Run the following script , Avoid releasing locks on other threads

//  Release the lock   Compare value Whether it is equal or not   Avoid accidental release 
if redis.call("get",KEYS[1]) == ARGV[1] then
    return redis.call("del",KEYS[1])
else
    return 0
end

Redlock Algorithm

Redlock yes Redis The author of Antirez The proposed cluster mode is distributed lock , be based on N Completely independent Redis Nodes realize high availability of distributed locks

Suppose there is N Completely independent Redis node , stay N individual Redis Use and on instances Redis In single instance, the same method is used to acquire and release the lock .

In the process of acquiring and releasing locks, the client will perform the following operations ：

• Get current system time , In Milliseconds

• Use the same... In turn Key And random values in N Lock requested on nodes

  When Redis When requesting a lock , The client will set a timeout for network connection and response , This timeout should be less than the lock expiration time , This prevents the client from waiting .

• The client uses the current time minus the time to acquire the lock to get the time to acquire the lock , If and only if more than half of Redis Node gets lock , And the use time is less than the lock aging time , Lock is success

• If you get the lock ,key The real effective time of is equal to The effective time minus the time taken to acquire the lock

• . If for some reason , Lock acquisition failed （ The lock is not obtained in more than half of the instances, or the lock removal time has exceeded the effective time ）, The client should be in all Redis Unlock the instance , No matter what Redis Whether the instance is locked successfully , Because the server response message may have been lost, but it actually succeeded , After all, one more release will not be a problem

（ Reference resources https://www.cnblogs.com/backnullptr/p/12155812.html）