Distributed lock database implementation

 

1：for update Principle

select Retrieved data , for update Added a lock , Others cannot modify this data , You can't lock this data anymore . Other threads can retrieve , But I'm using it for update It is impossible to lock these data , Because this lock , It has been locked by the previous thread . Others cannot lock it , During locking , No one else can modify these data . because update Update the data , Is to obtain the lock of these data

2： Code implementation

    @Transactional(rollbackFor = Exception.class)
    @RequestMapping("/database")
    public String databaseLock () throws InterruptedException {
        log.info(" How to get in ");
        BusinessLock businessLock = businessLockMapper.getLock("demo");
        if (ObjectUtils.isEmpty(businessLock)) {
            log.info(" Distributed lock not found ");
        }
        log.info(" Enter the lock ");
        Thread.sleep(6000);
        log.info(" Method execution complete ");
        return " Method execution complete ";
    }

 mapper Inside sql

        SELECT * FROM business_lock WHERE  business_code = #{businessCode}  FOR UPDATE

•   Open two instances , The ports are   8080  8088
• The figure below 8080  application    

1. 58:16 second       How to get in
2. 58:16 second       Enter the lock
3. 58:22 second       Method execution complete

 

• The figure below 8088  application    

1. 58:17 second       How to get in
2. 58:22 second       Enter the lock
3. 58:28 second       Method execution complete

In the code, we let the application that gets the lock sleep 6 second ,8080  and 8088 Enter the method at about the same time ,

however 8080 Earlier, all got the lock , and 8088 Just block in the database , Wait for lock .

stay 16 second 8080 Obtain lock wait 6 second 22 Second release lock , At this time   8088 Just got the lock . It shows that locks are effective in many applications .

Tutorial source code +sql download

3： use native measuring for update

• First, automatically submit the database of the current page and turn it off

Why turn it off first ？ After you retrieve it now or for update Lock up , After the lock is added, the transaction will be submitted automatically , The transaction is committed . The lock will automatically release . Other conversations , This data will also be retrieved

• The figure below ： The default is to automatically commit transactions  

• The figure below ： At this time, auto commit transactions are not turned off . Both can execute locking pages . It doesn't play the role we want
• The figure below ： Close the auto commit transaction of the two windows
•   Retrieve again with   for update Keywords sql . At this time, the lock is locked by the left window , The right window cannot be locked . At this time   The left window executes commit Commit transaction operations , The right window immediately finds the data . It means that the lock is effective

• The left window executes for update , The right window is not used   for update Instead, the query is executed directly Data can be queried  

4： summary