MySQL series transaction log redo log learning notes

MySQL Series of transaction logs Redo log Learning notes

Before learning this blog, you need to reserve knowledge ：

• MySQL Architecture
• InnoDB Storage engine
• MySQL Business knowledge

In the last blog , We know undo log, Continue with the last blog , Learn another important InnoDB Transaction log redo log

1、Redo Log

1.1、 What is? Redo log？

Redo ： Do it again ,undo Undo rollback means
Redo log： It is called redo log , When an accident occurs in the database , Data recovery ,redo log The backup is the modified data in the process of transaction execution ,redo log The latest data location during the transaction is backed up

1.2、Redo log working principle

Previous learning , We know undo log Is a log record for rollback , It belongs to logical log , And this blog introduces redo log It's a physical log , As long as it is used to prevent system redundancy and other special situations , When this happens , restart mysql Will read redo log, Thus, some data that is not written to the disk is written to the disk for persistent storage

Based on the diagram of blog in the previous chapter , Expand , Compare the undo log and redo log

• undo log and redo logo All are InnoDB The function of , All transaction logs
• undo log It's a logical log , Record is the operation record log ,redo log It's a physical log , New data is recorded
• undo log It is designed to ensure the atomicity of transactions ,redo log It is set to ensure transaction persistence .undo log stay InnoDB Is used to realize multi version control , perform rollback In operation ,undo log It can be used as a snapshot read reference for transaction rollback , and redo log Is the latest data location for backup , System redundancy , Just restart mysql service , You can persist unsaved data to disk

Insert picture description here

1.3、Redo log Write mechanism

Insert picture description here

• write point： This pointer records the current position , While writing , Move along , Write to the end of the last file and go back to 0 No. file overwrites again
• check point： This pointer records the current position to be erased , It's also going back and forth , Before erasing a record, update the record to a data file

quote 51 Legend of blog ： The picture is from ​ ​https://blog.51cto.com/wangwei007/2287431​​

redo log When it is full , The pointer returns to the origin , Restart overwrite save , If write pos Catch up checkpoint, Indicates full , No new updates can be performed at this time , You have to stop and erase some records , hold checkpoint Push forward the node

1.4、Redo log Configuration parameters

Check... By command redo log Parameters ：

Log in and copy

      
      

       
       show variables like '%innodb_log%';
      
      
      
      

       
       
1.
      
      

1.5、Redo Data persistence strategy

【 Expand knowledge 】, Every InnoDB The storage engine has at least 1 Log redo filegroups (redo file group), Each filegroup has at least 2 individual redo Log files , The default is ​​ib_logfile0​​​ and ​​ib_logfile1​​​, The storage path is ​​${datadir}​​

Redo Buffer Persist to Redo Log The strategy of , Through the variable ​​innodb_flush_log_at_trx_commit​​ The value of

Insert picture description here

• 0： Set the value to 0, Express Log Buffer The data in does not go through OS cache , Call directly fsync Directly brush to disk file to save
• 1： Set the value to 1（ The default value is ）, After the transaction is committed , It will be saved to log buffer, Then save to os buffer cache , At the same time call fsync Sync to disk
• 2： Set the value to 2, The data is not written log buffer, Cache directly to os buffer, Every second , call fsync Swipe data to disk

appendix ： Reference material

• ​ ​MySQL InnoDB redo Log elementary analysis ​​
• Detailed analysis MySQL Transaction log (redo log and undo log)