In stand-alone applications , When multithreading synchronous access to a shared variable , It can be used. JVM Of synchronized Lock or lock Lock handling .
But when in a distributed cluster system , Multithreading 、 Multiple processes are distributed on different machines , This will invalidate the concurrent control lock policy in the case of the original stand-alone deployment .

Therefore, a cross machine mutual exclusion mechanism is proposed to control the access of shared resources —— Distributed lock

One 、 Conditions for distributed

1. In a distributed system environment , A method can only be accessed by one thread on a machine at a time
2. Highly available acquire and release locks
3. Efficient lock acquisition and release
4. Reentrant feature （ Re enter , Used concurrently by multiple tasks , No more than worrying about data errors ）
5. Failure mechanism with lock , Prevent deadlock
6. Features of non blocking lock , The lock was not acquired , Direct return to get lock failed

Two 、 Three ways of implementation

Distributed systems should comply with CAP theory ：

C（Consistency）： Uniformity
A（Availability）： Usability
P（Partition Tolerance）： Partition tolerance

These three cannot be achieved at the same time , In most scenarios , All need sacrifice Strong consistency In exchange for the system High availability , Systems often only need to guarantee “ Final consistency ”.

The three ways to implement distributed locks are mainly ：

• Implementation of distributed lock based on Database
• be based on redis Implement distributed locks
• be based on Zookeeper Implement distributed locks

Based on the fact that the author has not known Zookeeper, So it will not be based on Zookeeper Implement distributed locks , Wait until you learn later ~

3、 ... and 、 Implementation of distributed lock based on Database

Create a lock table , By manipulating the data of the table .

DROP TABLE IF EXISTS `methodLock`;
CREATE TABLE `methodLock` (
  `id` int(11) unsigned NOT NULL AUTO_INCREMENT COMMENT ' Primary key ',
  `method_name` varchar(64) NOT NULL COMMENT ' Locked method name ',
  `desc` varchar(255) NOT NULL COMMENT ' Notes ',
  `update_time` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
  PRIMARY KEY (`id`),
  UNIQUE KEY `uidx_method_name` (`method_name`) USING BTREE
) ENGINE=InnoDB AUTO_INCREMENT=3 DEFAULT CHARSET=utf8 COMMENT=' Method in locking ';

Yes method_name Did Uniqueness constraint
When a method or resource needs to be locked , Insert record in lock table , If there are multiple requests submitted to the database at the same time , The database will guarantee that only one operation can succeed

INSERT INTO methodLock (method_name, desc) VALUES ('methodName', ' Locked method name or resource name ');

Delete records when releasing the lock

DELETE FROM methodLock WHERE method_name='methodName';

Using database to realize distributed lock is simple , But some problems still need to be solved ：

1. The availability and performance of distributed locks are directly affected by the availability and performance of databases ;
2. No, Lock failure The mechanism of , The data is not deleted when the lock is released due to server downtime , In this way, the lock cannot be obtained after the server is restored . So you need to have field records Failure time , Delete invalid data through scheduled tasks ;
3. Do not have Blocking lock characteristic , Fail to get lock and return directly , So we need to optimize the acquisition logic , Loop multiple times to get
4. Do not have Reentrant Characteristics of , When a machine is locked , Unable to insert data successfully again . So you need to have field records machine and Threads Information , When acquiring the lock, it is related to the currently acquired machine and site information , If identical , Then get the lock directly .

Four 、 be based on Redis Implement distributed locks

redis It has very high performance
redis The command supports this better , It's easy to use .

4.1 Realization thought

• When getting the lock , adopt setnx Lock , The lock value Value is a randomly generated UUID, Judge when releasing the lock through this
• When getting the lock , use expire Command to add a timeout to the lock , After this time, the lock is abandoned
• When you release the lock , adopt UUID Decide whether to lock , If it's time to lock , execute delete Lock release

4.2 Use command

• setnx： The order is at the designated key When there is no , by key Set the specified value , Set up the success , return 1 . Setup failed , return 0
• expire： Set up key The expiration time of ,key No longer available after expiration . Unit in seconds
• delete： Delete key value