mysql主从复制读写分离的配置方法详解

原作者: [db:作者] 来自: [db:来源] 收藏邀请

一、说明

前面我们说了mysql的安装配置,mysql语句使用以及备份恢复mysql数据;本次要介绍的是mysql的主从复制,读写分离;及高可用MHA;

环境如下:

master:CentOS7_x64 mysql5.721 172.16.3.175 db1
slave1:CentOS7_x64 mysql5.7.21 172.16.3.235 db2
slave2:CentOS7_x64 mysql5.7.21 172.16.3.235 db3
proxysql/MHA:CentOS7_x64 mysql5.7.21 172.16.3.235 proxysql

架构图:

说明:

配置测试时为了方便关闭了防火墙头,selinux安全策略;
现实中请开放防火墙策略;myslqdb的安装已经有脚本一键安装并配置好;这里就不在重复配置;只对对应的角色贴出对应的配置或安装与之相关的软件;

二、主从复制配置

一台主数据库,N从节点;从节点开启两个线程,通过Slave_IO_Running线程和主节点上有权限的账号从主数据库节点复制binlog日志到本地,能过Slave_SQL_Running线程在本地执行binlog日志,达到主从节点内容同步;

master配置:

egrep -v '(^$|^#)' /usr/local/mysql/etc/my.cnf

[mysqld]
datadir=/data1/mysqldb
socket=/tmp/mysql.sock
key_buffer_size   = 16M
max_allowed_packet  = 16M
thread_stack   = 192K
thread_cache_size  = 8
query_cache_limit  = 1M
query_cache_size  = 64M
query_cache_type  = 1
symbolic-links=0
innodb_file_per_table=ON
skip_name_resolve=ON

server-id  = 1
log_bin   = /data1/mysqldb/mysql-bin.log
[mysqld_safe]
log-error=/usr/local/mysql/logs/error.log
pid-file=/data1/mysqldb/mysql.pid
!includedir /usr/local/mysql/etc/my.cnf.d

创建从节点同步账号:

mysql > grant replication client,replication slave on *.* to 'repluser'@'172.16.3.%' identified by 'replpass';
mysql > flush privileges;
mysql >show master logs;
+------------------+-----------+
| Log_name   | File_size |
+------------------+-----------+
| mysql-bin.000001 |  622 |

主节点上的binlog日志文件及位置;请记下;从节点第一次同步时需要用;

slave节点:

egrep -v '(^$|^#)' /usr/local/mysql/etc/my.cnf

[mysqld]
datadir=/data1/mysqldb
socket=/data1/mysqldb/mysql.sock
key_buffer_size   = 16M
max_allowed_packet  = 16M
thread_stack   = 192K
thread_cache_size  = 8
query_cache_limit  = 1M
query_cache_size  = 64M
query_cache_type  = 1
symbolic-links=0

innodb_file_per_table=ON
skip_name_resolve=ON

server-id    = 11   #从节点标识ID 各从节点均不一样 
relay_log = relay-log
read_only=ON

[mysqld_safe]
log-error=/usr/local/mysql/log/error.log
pid-file=/var/run/mysql/mysql.pid
!includedir /usr/local/mysql/etc/my.cnf.d

启动mysq数据库

注意:两台从节点的server-id 值不一样;其他的都一样;因此从节点只展示一个配置文件;
登录数据库并同步数据启动slave

两台slave均要同步并启动

mysql > CHANGE MASTER TO MASTER_HOST="172.16.3.175",MASTER_USER="repluser",MASTER_PASSWORD="replpass",MASTER_PORT=3306,MASTER_LOG_FILE="mysql-bin.000001",MASTER_LOG_POS=622;
mysql > start slave;  #启动从节点()

#查看从节点状态
mysql > SHOW SLAVE STATUS;
*************************** 1. row ***************************
    Slave_IO_State: Waiting for master to send event
     Master_Host: 172.16.3.175     #主节点
     Master_User: repluser       #同步账号
     Master_Port: 3306
    Connect_Retry: 60
    Master_Log_File: mysql-bin.000001
   Read_Master_Log_Pos: 622
    Relay_Log_File: relay-log.000001
    Relay_Log_Pos: 582
  Relay_Master_Log_File: mysql-bin.000001
    Slave_IO_Running: Yes      #同步线程正常
   Slave_SQL_Running: Yes     #本地写线程正常
    Replicate_Do_DB:        #同步过滤为空(可以只同步某个或某些库)
   Replicate_Ignore_DB:        #不同步的库
   Replicate_Do_Table:      
  Replicate_Ignore_Table: 
  Replicate_Wild_Do_Table: 
 Replicate_Wild_Ignore_Table: 
     Last_Errno: 0    #最后同步的错误 0表示正常同步
     Last_Error: 
     Skip_Counter: 0
   Exec_Master_Log_Pos: 622
    Relay_Log_Space: 615
    Until_Condition: None
    Until_Log_File: 
    Until_Log_Pos: 0
   Master_SSL_Allowed: No
   Master_SSL_CA_File: 
   Master_SSL_CA_Path: 
    Master_SSL_Cert: 
   Master_SSL_Cipher: 
    Master_SSL_Key: 
  Seconds_Behind_Master: 0
Master_SSL_Verify_Server_Cert: No
    Last_IO_Errno: 0
    Last_IO_Error: 
    Last_SQL_Errno: 0
    Last_SQL_Error: 
 Replicate_Ignore_Server_Ids: 
    Master_Server_Id: 1
     Master_UUID: 57017c43-36e3-11e8-ac76-080027393fc7
    Master_Info_File: /data1/mysqldb/master.info
     SQL_Delay: 0
   SQL_Remaining_Delay: NULL
  Slave_SQL_Running_State: Slave has read all relay log; waiting for more updates
   Master_Retry_Count: 86400
     Master_Bind: 
  Last_IO_Error_Timestamp: 
  Last_SQL_Error_Timestamp: 
    Master_SSL_Crl: 
   Master_SSL_Crlpath: 
   Retrieved_Gtid_Set: 
   Executed_Gtid_Set: 
    Auto_Position: 0
   Replicate_Rewrite_DB: 
     Channel_Name: 
   Master_TLS_Version: 
1 row in set (0.00 sec)
ERROR: 
No query specified

测试主从同步

在master导入测试数据;修改数据并查看slave 中的数据是否一致;

[root@db1 ~]# mysql < Testdb.sql
登录数据库
[root@db1 ~]# mysql -uroot -p
mysql> show databases;
+--------------------+
| Database   |
+--------------------+
 study    |
+--------------------+
5 rows in set (0.00 sec)
###study测试数据库导入成功
mysql> use study;
Database changed
mysql> show tables;
+-----------------+
| Tables_in_study |
+-----------------+
| class   |
| course   |
| part   |
| score   |
| student   |
| tb31   |
| tb32   |
| teacher   |
| test1   |
| test2   |
| user_info  |
+-----------------+
11 rows in set (0.00 sec)
#删除test1 test2表

slave从节点上查看

mysql> show tables;
+-----------------+
| Tables_in_study |
+-----------------+
| class   |
| course   |
| part   |
| score   |
| student   |
| tb31   |
| tb32   |
| teacher   |
| user_info  |
+-----------------+
9 rows in set (0.00 sec)

数据已经可以正常同步;注意主从同步只需要第一次手动启动;之后都随mysql服务自动启动;主从同步架构只方便了数据的同步,此时如果没有第三方工具介入想做到读写分离就需要在程序中去做,难免出错;而出错了,就需要手动同步数据；这里通过proxysql来做读写分离;

三、proxysql之读写分离

以上已经完成了主从复制配置;然而这只是一个基本配置,加上一个proxysql实现mysql读写分离,proxysql类似haproxy七层代理路由功能且支持MySQL 协议的的数据库代理；是dba开发给dba使用的;用户请求发向proxysql,如果是写请求发往主节点;读请求发下从节点组中;以此实现读写分离;一定程序上减轻了主数据库的io压力;
下载安装proxysql
目前最新版本是1.4.7-1(由于最新版本有问题)
我们这里下载使用1.3.6-1的基于CentOS7的rpm包;下载到本地并yum安装

[root@proxysql ~]# yum install proxysql-1.3.6-1-centos7.x86_64.rpm -y
[root@proxysql ~]# rpm -ql proxysql
/etc/init.d/proxysql
/etc/proxysql.cnf  #主配置文件
/usr/bin/proxysql
/usr/share/proxysql/tools/proxysql_galera_checker.sh
/usr/share/proxysql/tools/proxysql_galera_writer.pl

配置如下:

在配置proxysql之前需要在主节点配置授权账号以作proxysql对主从节点操作;另外proxysql上的mysql客户端工具需要和主从节点上的保持一致;
在主节点master上授权登录账号：

mysql > GRANT ALL ON *.* TO 'myadmin'@'172.16.3.%' identified by 'mypass';

proxysql.cnf配置

[root@proxysql ~]# egrep -v '(^$|^#)' /etc/proxysql.cnf

datadir="/var/lib/proxysql"
admin_variables=
{ 
 admin_credentials="admin:admin"    #proxysql自己的管理用户名密码
 mysql_ifaces="127.0.0.1:6032;/tmp/proxysql_admin.sock"
}
mysql_variables=
{
 threads=4       #线程数,建议和cpu核心数一致
 max_connections=2048  #最大连接
 default_query_delay=0
 default_query_timeout=36000000
 have_compress=true
 poll_timeout=2000
 interfaces="0.0.0.0:3306;/tmp/proxysql.sock"    #对外接口
 default_schema="information_schema"
 stacksize=1048576
 server_version="5.5.30"
 connect_timeout_server=3000
 monitor_username="monitor"
 monitor_password="monitor"
 monitor_history=600000
 monitor_connect_interval=60000
 monitor_ping_interval=10000
 monitor_read_only_interval=1500
 monitor_read_only_timeout=500
 ping_interval_server_msec=120000
 ping_timeout_server=500
 commands_stats=true
 sessions_sort=true
 connect_retries_on_failure=10
}

#####主从节点的配置
mysql_servers =      
(
 {
  address = "172.16.3.175" # no default, required . If port is 0 , address is interpred as a Unix Socket Domain
  port = 3306   # no default, required . If port is 0 , address is interpred as a Unix Socket Domain
  hostgroup = 1   # 设置组号
  status = "ONLINE"  # default: ONLINE
  weight = 1   # default: 1
  compression = 0  # default: 0
  max_connections = 200   ###定义最大的连接
 },
 {
  address = "172.16.3.235" # no default, required . If port is 0 , address is interpred as a Unix Socket Domain
  port = 3306   # no default, required . If port is 0 , address is interpred as a Unix Socket Domain
  hostgroup = 2   # no default, required
  status = "ONLINE"  # default: ONLINE
  weight = 1   # default: 1
  compression = 0  # default: 0
  max_connections=1000
 },
 {
  address = "172.16.3.241" # no default, required . If port is 0 , address is interpred as a Unix Socket Domain
  port = 3306   # no default, required . If port is 0 , address is interpred as a Unix Socket Domain
  hostgroup = 2   # no default, required
  status = "ONLINE"  # default: ONLINE
  weight = 1   # default: 1
  compression = 0  # default: 0
  max_connections=1000
 }
)
mysql_users:
(
 {
  username = "myadmin" # no default , required
  password = "mypass" # default: ''
  default_hostgroup = 1 # default: 0
  max_connections=1000
  default_schema="test"
  active = 1   #是否激活
 }
)
mysql_query_rules:
(
)
scheduler=
(
)
mysql_replication_hostgroups=
(
  {
    writer_hostgroup=1   #定义写组号1
    reader_hostgroup=2  #定义读组号2
    comment="test repl 1" #注释内容
  }
)

启动proxysql服务

[root@proxysql ~]# service proxysql start

测试proxysql

模拟通过proxysql使用数据库
[root@proxysql ]# mysql -h172.16.3.175 -umyadmin -pmypass
mysql: [Warning] Using a password on the command line interface can be insecure.
Welcome to the MySQL monitor. Commands end with ; or \g.
Your MySQL connection id is 17406
Server version: 5.7.21-log MySQL Community Server (GPL)
Copyright (c) 2000, 2018, Oracle and/or its affiliates. All rights reserved.
Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Other names may be trademarks of their respective
owners.
Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.
mysql> 

mysql > show databases;
mysql> show databases;
+--------------------+
| Database   |
+--------------------+
| information_schema |
| mysql    |
| performance_schema |
| study    |
| sys    |
+--------------------+
5 rows in set (0.00 sec)
###删除study数据库user_info中6 -12之间的数据
删除之前:
mysql> select * from user_info;
+-----+-------+------+--------+----------+
| nid | name | age | gender | part_nid |
+-----+-------+------+--------+----------+
| 1 | san | 20 | 男  |  1 |
| 2 | dong | 29 | 男  |  2 |
| 4 | Ling | 28 | 男  |  4 |
| 5 | ling | 28 | 男  |  3 |
| 6 | dong | 30 | 男  |  1 |
| 7 | b  | 11 | 女  |  1 |
| 8 | c  | 12 | 女  |  1 |
| 9 | d  | 18 | 女  |  4 |
| 10 | e  | 22 | 男  |  3 |
| 11 | f  | 23 | 男  |  2 |
| 12 | dongy | 22 | 男  |  1 |
+-----+-------+------+--------+----------+
11 rows in set (0.00 sec)
删除之后:
mysql> delete from user_info where nid >6 and nid <12;
Query OK, 5 rows affected (0.03 sec)

mysql> select * from user_info;
+-----+-------+------+--------+----------+
| nid | name | age | gender | part_nid |
+-----+-------+------+--------+----------+
| 1 | san | 20 | 男  |  1 |
| 2 | dong | 29 | 男  |  2 |
| 4 | Ling | 28 | 男  |  4 |
| 5 | ling | 28 | 男  |  3 |
| 6 | dong | 30 | 男  |  1 |
| 12 | dongy | 22 | 男  |  1 |
+-----+-------+------+--------+----------+
6 rows in set (0.00 sec)

到主从节点上查看,会发现以上的查和修改数据都被proxysql正确的代理到后端处理了;
以上看了并不直观;为了查看proxysql与各主从节点通讯我们在主从节点上安装tcpdump并过滤包

主节点:
类似如下:
[root@db1 ~]# tcpdump -i enp0s3 -nn tcp port 3306
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on enp0s3, link-type EN10MB (Ethernet), capture size 262144 bytes
18:04:34.678861 IP 172.16.3.254.42191 > 172.16.3.175.3306: Flags [S], seq 3385407732, win 29200, options [mss 1460,sackOK,TS val 17576713 ecr 0,nop,wscale 7], length 0
18:04:34.678908 IP 172.16.3.175.3306 > 172.16.3.254.42191: Flags [S.], seq 1579426335, ack 3385407733, win 28960, options [mss 1460,sackOK,TS val 29413673 ecr 17576713,nop,wscale 7], length 0
18:04:34.680902 IP 172.16.3.254.42191 > 172.16.3.175.3306: Flags [.], ack 1, win 229, options [nop,nop,TS val 17576715 ecr 29413673], length 0
18:04:34.681264 IP 172.16.3.175.3306 > 172.16.3.254.42191: Flags [P.], seq 1:83, ack 1, win 227, options [nop,nop,TS val 29413675 ecr 17576715], length 82
....
从节点:
类似如下:
[root@db2 data1]# tcpdump -i enp0s3 -nn tcp port 3306
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on enp0s3, link-type EN10MB (Ethernet), capture size 262144 bytes
18:02:57.932043 IP 172.16.3.254.42733 > 172.16.3.235.3306: Flags [S], seq 76520456, win 29200, options [mss 1460,sackOK,TS val 17479189 ecr 0,nop,wscale 7], length 0
...........

proxysql命令行管理接口:支持运行时修改

 [root@proxysql]# mysql -u admin -padmin -h 127.0.0.1 -P6032 --prompt='Admin> '
Admin> show databases;
+-----+---------+-------------------------------+
| seq | name | file       |
+-----+---------+-------------------------------+
| 0 | main |        |
| 2 | disk | /var/lib/proxysql/proxysql.db |
| 3 | stats |        |
| 4 | monitor |        |
+-----+---------+-------------------------------+
4 rows in set (0.00 sec)

以上stats,monitor,main都从配置文件中获取的数据库;可以通过类似mysql运行时修改;而不需要重启;

至此我们基于proxysql主从复制读写分离架构已经完成;
双主或多主模型是无须实现读写分离，仅需要负载均衡：haproxy, nginx, lvs等;
proxysql并没有解决,当主数据岩机时的问题;此时就需要量MHA来解决 ;后续再介绍;