MySQL master-slave replication

Copy how it works

1. master Log changes to binary log (binary log) in （ These records are called binary log events ,binary log events）;
2. slave take master Of binary log events Copy to its trunk log (relay log);
3. slave Redo events in relay log , Will change the data that reflects itself .

The above is just an overview , In fact, every step is complicated

Configure replication

by MySQL Server configuration replication is very simple . But because of the different scenes , The basic steps are different . The most basic scenario is a completely new master library and slave library , In general, it is divided into the following steps ：

1. Create a replication account on each server .
2. Configure the main database and the standby database .
3. Notify the standby database to connect to the primary database and copy data from the primary database .

Let's assume that most of the configurations are default values , Both the primary database and the standby database are newly installed and have the same data （ Default MySQL database ） Next we show how to configure replication step by step ： Suppose you have a server server1（IP Address 10.100.2.122） And the server server2（IP Address 10.100.2.143）.

Create a copy account

MySQL Will give some special permissions to the replication thread . Running in the standby database I/O The thread will create a to the main library TCP/IP Connect , This means that you must create a user in the main library , And give them appropriate authority . For the library I/O The thread connects to the main library with this user name and reads its binary log . Create a user account with the following statement ：

GRANT REPLICATION SLAVE, REPLICATION CLIENT ON *.*
        TO [email protected]'10.100.2.%' IDENTIFIED BY 'p4ssword',;

We create this account in both the primary database and the standby database . Note that we restrict this account to the local network , Because this is a privileged account （ Although the account cannot be executed select Or modify data , But you can still get some data from the binary log ）.
In fact, to copy an account, you only need to have REPLICATION SLAVE jurisdiction , You don't have to have... On each end of the server REPLICATION CLIENT jurisdiction , Then why do we give these two permissions to the master / All the spare parts are given ？ There are two reasons for this ：

1. The accounts used to monitor and manage replication need to be REPLICATION CLIENT jurisdiction , And it's easier to use the same account for both purposes （ Instead of creating a separate account for a particular purpose ）.
2. If an account is established on the main database , When cloning data from the primary database to the secondary database , The standby database is set up —— The configuration needed to become the main library . In this way, the roles of the primary and standby databases can be easily exchanged if necessary .

This is the norm , But this is just a demonstration and case , The user was not created , Using two databases directly root Permission account

Configure the main database and the standby database

Next, you need to enable some settings on the main database , Suppose that the main database is a server server1, You need to open the binary log and specify a unique server ID（server ID）, In the main warehouse my.ini（linux The system name is my.cnf） Add or modify the following contents in the document ：

log-bin
server-id = 2

The actual value is up to you , This is just for simplicity , Of course, you can also set more required configurations . Using the default value may result in... With other servers ID Conflict , A common practice in the company is to use servers IP End of address 8 position , But make sure it is constant and unique （ for example , The servers are all in a subnet ）. We'd better choose some meaningful conventions and follow them , But here it's free .

It should be noted that , my mysql Version is 5.7, In different systems , Different versions have different names , Such as log_bin,server_id; After the save , Don't forget to restart mysql Service makes it effective

The standby database also needs to be in my.ini Add similar configuration in , also You also need to restart the server .

log_bin
server_id = 1
read_only = 1

Technically , These options are not always necessary . For some of these options, we just explicitly list the default values . In fact only server_id It's necessary , because server-id The default is 1, So I haven't modified the slave library here . Here we also use log_bin, And uses the default name . Of course, if you like , It can also be set to other values .

Best use read_only configuration option , This option will prevent any threads without privileges from modifying data （ So it's best not to give users more permissions than they need ）. but read_only Options are often not very practical , Especially for those applications that need to create tables in the standby database .

Start replication

We are executing this statement on the slave Library ,host Fill in the main database ip Address , The account password is the same , For the binary log file of the main library, we need show Just copy and paste ; The name can actually be configured , In the above log-bin Prefix can be configured in , Do not configure to use the device name as the prefix by default .

CHANGE MASTER TO MASTER_HOST='10.100.2.122',
         MASTER_USER='root',
         MASTER_PASSWORD='123456',
         MASTER_LOG_FILE='DESKTOP-V351QGI-bin.000003',
         MASTER_LOG_POS=0;

After execution , We can see Slave The state of , If there is any mistake, it will be in the result LAST_ERROR It shows that

SHOW SLAVE STATUS;

If there is no mistake , perform START SLAVE, Master-slave replication begins , The data inserted or modified in the master database will be asynchronously executed in the slave database ;