麒麟v10 上部署 TiDB v5.1.2 生产环境优化实践
738
2023-05-06
MySQL主从复制读写分离与高可用配置
一、说明
前面我们说了mysql的安装配置(并提供一键安装脚本),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 db1slave1:CentOS7_x64 mysql5.7.21 172.16.3.235 db2slave2:CentOS7_x64 mysql5.7.21 172.16.3.235 db3proxysql/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
启动proxysql服务
[root@proxysql ~]# service proxysql start
测试proxysql
模拟通过proxysql使用数据库
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;更多命令行的配置请参考github;
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;
版权声明:本文内容由网络用户投稿,版权归原作者所有,本站不拥有其著作权,亦不承担相应法律责任。如果您发现本站中有涉嫌抄袭或描述失实的内容,请联系我们jiasou666@gmail.com 处理,核实后本网站将在24小时内删除侵权内容。