www.colben.cn/content/post/mysql-replication.md
2022-05-17 22:38:01 +08:00

8.5 KiB
Raw Blame History

title, date, lastmod, tags, categories
title date lastmod tags categories
MySQL 主从复制 2019-10-30T17:33:13+08:00 2020-04-18T23:40:00+08:00
mysql
master
slave
gtid
replication
同步
database

环境

  • mysql 主服务器centos7.4192.168.1.10,端口 10000
  • mysql 从服务器centos7.4192.168.1.6
  • 要复制的数据库有 data_db、conf_db

在主服务器上创建用于备份的用户 replicator

grant replication slave  on *.* to 'replicator'@'192.168.1.6' identified by 'password';
flush privileges;

在主服务器上修改 my.cnf

# 建议与本机ip地址最后一位一致,与其他互备服务器均不一致
server-id = 10
read-only = 0

# 开启 binlog
log-bin = /var/log/mysql-bin/master
binlog-format = row
# 关注要复制的数据库,存在跨库问题
binlog-do-db = data-db
binlog-do-db = conf-db
# 忽略的数据库,存在跨库问题
#binlog-ignore-db =
#binlog 有效时间
expire-logs-days =
# GTID 复制模式
#gtid-mode = ON
#enforce-gtid-consistency=true
# 双主或多主互备时,会用到以下配置
# 自增主键初始值,与其他互备服务器一致
#auto-increment-offset =
# 自增主键等差值,与其他互备服务器均不一致
#auto-increment-increment =

在主服务器上重启 mysql获取 master 状态

-- 如果 mysql 是全新安装,则无须导出数据库初态,直接查看 binlog pos 即可
-- 锁定要导出的数据库表
flush tables with read lock;

导出数据库初态

# 在主服务器的另一个终端中运行
mysqldump -uroot -p data_db > /tmp/data_db.sql
mysqldump -uroot -p conf_db > /tmp/conf_db.sql
# 复制到从服务器上
scp /tmp/data_db.sql /tmp/conf_db.sql 192.168.1.6:/tmp/

查看 binary 日志位置

show master status\G
-- 记住输出里的如下类似两行不记录也可以这两个信息已经写入了导出的sql文件中
-- File: mysql-bin.000001
-- Position: 137103822
-- 解锁数据库表
unlock tables;

在从服务器上编辑 my.cnf

# 建议与本机ip地址最后一位一致,与其他互备服务器均不一致
server-id = 6
read-only = 1

# 如果该 slave 中也运行了 master或者使用了 gtid 模式复制,则开启 binlog
#log-bin = mysql-bin
#binlog-format = row
# 把 slave 操作也计入 binlog用于链式同步
#log-slave-updates = ON
# 指定要复制的数据库,存在跨库问题
#replicate-do-db = data_db
#replicate-do-db = conf_db
# 指定要复制的数据表,无跨库问题
replicate-do-table = db1.t1
replicate-wild-do-table = db1.%
# 忽略的数据库,存在跨库问题
#replicate-ignore-db =
# 忽略的数据表,无跨库问题
#replicate-ignore-table = db1.t1
#replicate-wild-ignore-table = db1.%
# 中继日志
relay-log = /var/lib/mysql-bin/slave
# 多线程复制
slave-parallel-type = logical-clock
slave-parallel-workers = 4
# GTID 模式
#gtid-mode = ON
#enforce-gtid-consistency=true
# 双主或多主互备时,会用到以下配置
# 自增主键初始值,与其他互备服务器一致
#auto-increment-offset =
# 自增主键等差值,与其他互备服务器均不一致
#auto-increment-increment =

在从服务器上重启 mysql导入初态

-- 创建要导入的数据库
create database data_db default charset utf8mb4;
create database conf_db default charset utf8mb4;

导入数据库

msyql -uroot -p data_db < /tmp/data_db.sql
mysql -uroot -p conf_db < /tmp/conf_db.sql

开启同步

-- 基于 binlog 文件位置复制
change master to master_host = '192.168.1.10',
                        master_port = 3306,
                        master_user = 'replicator',
                        master_password = 'password',
                        master_log_file = 'mysql-bin.000001',
                        master_log_pos = 137103822;
flush privileges;
-- 基于 gtid 复制
change master to master_host = '192.168.1.10',
                        master_port = 3306,
                        master_user = 'replicator',
                        master_password = 'password',
                        master_auto_position = 1;

启动 slave查看 slave 状态

start slave;
-- 在从服务器上查看 slave 状态
show slave status\G
-- 如果看到
-- Slave_IO_Running: Yes
-- Slave_SQL_Running: Yes
-- 则表示 slave 开启成功!

MySQL8 gtid 互为主从配置

[mysqld]
# ---- 固定配置 ----
datadir = /var/lib/mysql
socket = /var/lib/mysql/mysql.sock
pid-file = /var/lib/mysql/mysql.pid
log-timestamps = SYSTEM
log-error = /var/log/mysql/error.log
slow-query-log = TRUE
slow-query-log-file = /var/log/mysql/slow.log
default-authentication-plugin = mysql_native_password

# ---- 动态配置 ----
mysqlx = OFF
character-set-server = utf8mb4
default-storage-engine = innodb
lower-case-table-names = 1
#skip-name-resolve = 1
#max-user-connections = 600
#innodb-buffer-pool-size = 8G
#innodb-buffer-pool-instances = 8

# master
# 确认不同节点该 id 唯一
server-id = 1
log-bin = /var/lib/mysql-bin/master
binlog-format = ROW
#binlog-do-db = db1
binlog-expire-logs-seconds = 172800
gtid-mode = ON
enforce-gtid-consistency = TRUE

# slave
replicate-wild-ignore-table = information_schema.%
replicate-wild-ignore-table = mysql.%
replicate-wild-ignore-table = performance_schema.%
replicate-wild-ignore-table = sys.%
replica-parallel-workers = 2
log-replica-updates = FALSE
relay-log = /var/lib/mysql-bin/slave
relay-log-recovery = TRUE
#read-only = ON

MySQL8 gtid 多源复制从库配置

  • 修改 my.cnf

    [mysqld]
    # ---- 固定配置 ----
    datadir = /data/mysql
    socket = /var/lib/mysql/mysql.sock
    pid-file = /var/lib/mysql/mysql.pid
    log-timestamps = SYSTEM
    log-error = /var/log/mysql/error.log
    slow-query-log-file = /var/log/mysql/slow.log
    slow-query-log = TRUE
    default-authentication-plugin = mysql_native_password
    
    # ---- 动态配置 ----
    mysqlx = OFF
    character-set-server = utf8mb4
    default-storage-engine = innodb
    lower-case-table-names = 1
    #skip-name-resolve = 1
    #max-user-connections = 600
    #innodb-buffer-pool-size = 8G
    #innodb-buffer-pool-instances = 8
    
    # master
    server-id = 39
    log-bin = /var/lib/mysql/master
    binlog-format = ROW
    #binlog-do-db = db1
    gtid-mode = ON
    enforce-gtid-consistency = TRUE
    binlog-expire-logs-seconds = 172800
    
    # slave
    #replicate-wild-do-table = db1.%
    replicate-wild-ignore-table = information_schema.%
    replicate-wild-ignore-table = mysql.%
    replicate-wild-ignore-table = performance_schema.%
    replicate-wild-ignore-table = sys.%
    slave-parallel-workers = 2
    log-slave-updates = FALSE
    relay-log = /var/lib/mysql/slave
    relay-log-recovery = TRUE
    super_read_only = ON
    master_info_repository  = table
    relay_log_info_repository = table
    
  • 导出主库数据

    mysqldump -uroot -h<主库1> -p --single-transaction --set-gtid-purged=on --databases db1 > 1_db1.sql
    mysqldump -uroot -h<主库2> -p --single-transaction --set-gtid-purged=on --databases db2 > 2_db2.sql
    
  • 在导出的文件(1_db1.sql,2_db2.sql)中找到 "SET @@GLOBAL.gtid_purged ..." 语句,记录下来,并在该文件中删除

  • 导入 1_db1.sql 和 2_db2.sql

    mysql -uroot -p < 1_db1.sql
    mysql -uroot -p < 2_db2.sql
    
  • 合并这两个 "SET @@GLOBAL.gtid_purged ..." 语句(gtid set 做并集),导入 gtid set 并集

    set @@global.gtid_purged = '<gtid set 并集>'
    
  • 加入两个主库的同步配置

    change master to master_host='<主库1>',
        master_port=3306,
        master_user='<主库上的 replication 账户>',
        master_password='<主库上的 replication 账户的密码>',
        master_auto_position=1 for channel '<master_1>';
    change master to master_host='<主库2>',
        master_port=3306,
        master_user='<主库上的 replication 账户>',
        master_password='<主库上的 replication 账户的密码>',
        master_auto_position=1 for channel '<master_2>';
    
  • 启动从库

    start slave for channel '<master_1>';
    start slave for channel '<master_2>';
    
  • 查看从库

    show slave status for channel '<master_1>'\G
    show slave status for channel '<master_2>'\G
    
  • 停止从库

    -- 停止全部 slave
    stop slave;
    -- 停止指定 slave
    stop slave for channel '...';
    
  • 重置从库

    -- 重置全部 slave
    reset slave;
    -- 重置指定 slave
    reset slave for channel '...';
    
  • 监控表: performance_schema.replication_connection_status