MySQL asynchronous replication, semi-synchronous replication, enhanced semi-synchronous replication (the most complete in history)

Background: From the little buddy question

My friend's problem:

When mysql master-slave synchronization, what is the concept of semi-synchronization and enhanced semi-synchronization, I read that it is a bit unclear on the Internet, and I have not found a suitable explanation.

insert image description here

Here, Nien will give you a systematic and systematic review. Also include this question and reference answers in our "Nin Java Interview Collection PDF" MySQL topic, for the reference of later friends, and improve everyone's 3-high architecture, design, and development levels.

Note: This article is continuously updated in PDF. For the latest PDF files of Nien’s architecture notes and interview questions, please get them from the link below: Code Cloud

At present, MySQL mainly has three replication methods

1) Asynchronous replication

2) Semi-synchronous replication

3) Enhanced semi-synchronous replication

Recommended use: It is recommended to use asynchronous replication for high performance requirements. If the financial business is running, it is recommended to use enhanced semi-synchronous replication, and use ROW+GTID+5.7 or above

asynchronous replication

The default method of MySQL replication

illustrate:

If you don't understand this well, this is rocketmq's asynchronous brushing + asynchronous replication

For details, please refer to Nien's rocketmq high availability video, Chapter 17 video

semi-synchronous replication

rpl_semi_sync_master_wait_point=after_commit

illustrate:

If you don't understand this well, this is rocketmq's synchronous brushing + asynchronous replication

For details, please refer to Nien's rocketmq high availability video, Chapter 17 video

Advantages and disadvantages of semisynchronous replication

Disadvantage 1: phantom reading

When a user initiates a transaction, the transaction has been written to the redo log and the binlog log, but the transaction has not been written to the slave library. At this time, it is at the waiting slave dump. At this time, another user can read this data, and he himself could not;

Cons 2: Data Loss

After a transaction crashes at the waiting slave dump, the master library will have one more piece of data than the slave library

Question: At this time, if the master-slave replication is restored, can the data be synchronized to the slave server normally? (not tested!!)

Enhanced Semisynchronous Replication

rpl_semi_rsync_master_wait_point=after_sync

Improvement 1: Solve phantom reading

When a user initiates a transaction, after the transaction is written into the binary, it synchronizes with the slave library. At this time, other users cannot read the data, which solves the phantom reading

Improvement 2: Solve data loss

After a transaction crashes at the waiting slave dump, you can observe whether the last gtid value of the master database exists on the slave database. If it exists, this data will be restored normally. If it does not exist, delete the redundant GTID value of the master database. Then recover to ensure the integrity of the data;

illustrate:

If you don't understand this well, this is rocketmq's synchronous brushing + synchronous replication

For details, please refer to Nien's rocketmq high availability video, Chapter 17 video

deployment test

Asynchronous replication environment: one master and one slave

主IP：192.168.20.206

从IP：192.168.20.212

3 steps are required:

1. Load the semi-synchronous replication plug-in to the main library, and modify the my.cnf parameters of the main library;

2. Load the semi-synchronous replication plug-in for the slave library, and modify the my.cnf parameters of the slave library;

3. Execute in the slave library and use the stop salve io_thread and start slave io_thread commands to get it done

1.1) The main library loads the semi-synchronous replication plug-in

install plugin rpl_semi_sync_master soname 'semisync_master.so';

1.2) The main library opens the semi-synchronous replication mode

set global rpl_semi_sync_master_enabled=1;

set global rpl_semi_sync_master_timeout=1000;

2.3) The main library modifies the my.cnf file

vim /etc/my.cnf

[mysqld]

###:for replcation

rpl_semi_sync_master_enabled=1

rpl_semi_sync_master_timeout=1000

2.1) Load the semi-synchronous replication plugin from the library

install plugin rpl_semi_sync_SLAVE soname 'semisync_slave.so';

2.2) Open the semi-synchronous replication mode from the library

set global rpl_semi_sync_slave_enabled=1;

2.3) Modify the my.cnf file from the library

vim /etc/my.cnf

[mysqld]

###:for replcation

rpl_semi_sync_slave_enabled=1

3.3) Use stop slave io_thread and start slave io_thread to execute in the slave library;

stop slave io_thread;

start slave io_thread;

Semi-synchronous replication monitoring command:

show plugins;

show global status like '%semi%';

show global variables like '%semi%';

references

https://blog.csdn.net/weixin_35555531/article/details/113948452

The Path to Technical Freedom PDF is available at:

Realize your architectural freedom:

" Have a thorough understanding of the 8-figure-1 template, everyone can do the architecture "

" 10Wqps review platform, how to structure it? This is what station B does! ! ! "

" Alibaba Two Sides: How to optimize the performance of tens of millions and billions of data?" Textbook-level answers are coming "

" Peak 21WQps, 100 million DAU, how is the small game "Sheep a Sheep" structured? "

" How to Scheduling 10 Billion-Level Orders, Come to a Big Factory's Superb Solution "

" Two Big Factory 10 Billion-Level Red Envelope Architecture Scheme "

… more architecture articles, being added

Please go to the official account of "Technical Freedom Circle" to get the PDF file updates of the above Nien architecture notes and interview questions↓↓↓