Summary of database Deadlock: (3.7-3.13)

brief introduction :
Database deadlock is generally a state of waiting for deadlock caused by two or more threads competing for a resource at the same time , For example, user access thread A Lock the record C, Another user accesses the thread B And need
Access records C It will form a deadlock ;

The main cause of deadlock :
1. Insufficient system resources ;( Deadlock is rarely formed when system resources are fully allocated )
2. The order of objects accessed is inconsistent ;
3. Improper allocation of system resources, etc ;

Four necessary conditions for deadlock ( Be short of one cannot ):
1. mutual exclusion : At the same time , An object resource can only be used by one thread , Other threads cannot occupy ;
2. Request and hold conditions : When a process initiates an access request for a new resource , The resources held by itself will not be released ;
3. Conditions of non deprivation : When the resources of a process are not used up , Can't be forcibly deprived of ;
4. Loop wait condition : Two or more processes form the condition of waiting for the same resource in the first place ;

Measures to reduce deadlocks :
1. Multiple threads access objects sequentially at the same time : Between multiple threads , Like threads A And thread B All need to access the table A And table B, Threads A First access the table A Revisit table B, In this process, the thread B In a waiting state ;
2. Avoid user interaction : On the ground , The efficiency of transactions without user interaction is much higher than that with user interaction , If there is a scene of user interaction , When the user needs to enter a string of authentication or other content , This
              When , Users don't do anything else in front of the computer , It will cause a blocking waiting state of resources , Slightly worse performance , In severe cases, access to this resource by other threads will cause deadlock ;
3. Multiple transactions should be as short as possible and in a batch : Because if there are more than one chief of staff , The exclusive lock or update lock held by itself takes a long time , Easy to deadlock , So try to keep things short , Multiple
                                Try to process transactions in a batch transaction , Because the fewer transactions , The less network performance overhead , Faster performance response , Causing a delay in the completion of a transaction
                                The less likely it is ;
4. Use low isolation levels : In some scenes , We can use a lower level of isolation on the premise of completing the task , For example, the level of serialization can be changed to the level of read commit to complete transactions ;
5. Binding connection : When an application binds two or more connections , The lock resource used by the secondary connection can be held like the lock held by the primary connection , There is no deadlock risk , contrary
          It's the same thing ;