MYSQL高可用架构实践

一 . 基础环境设置
 

1. 准备四台机器 

10.0.0.150 管理机 Manager节点
10.0.0.151 MASTER Node节点
10.0.0.152 SLAVE Node节点
10.0.0.153 SLAVE Node节点

 2.建立秘钥对

(1) 在10.0.0.150机器上操作
[root@manage ~]# ssh-keygen -t dsa
[root@manage ~]# ssh-copy-id  -i  /root/.ssh/id_dsa.pub root@10.0.0.151
[root@manage ~]# ssh-copy-id  -i  /root/.ssh/id_dsa.pub root@10.0.0.152
[root@manage ~]# ssh-copy-id  -i  /root/.ssh/id_dsa.pub root@10.0.0.153
 
(2) 在10.0.0.151机器上操作
[root@mha01 ~]#ssh-keygen -t dsa
[root@mha01 ~]# ssh-copy-id  -i  /root/.ssh/id_dsa.pub root@10.0.0.152
[root@mha01 ~]# ssh-copy-id  -i  /root/.ssh/id_dsa.pub root@10.0.0.153
 
(3) 在10.0.0.152机器上操作
[root@mha02 ~]# ssh-keygen -t dsa
[root@mha02 ~]# ssh-copy-id  -i  /root/.ssh/id_dsa.pub root@10.0.0.151
[root@mha02 ~]# ssh-copy-id  -i  /root/.ssh/id_dsa.pub root@10.0.0.153
 
(4) 在10.0.0.153机器上操作
[root@mha03 ~]# ssh-keygen -t dsa
[root@mha03 ~]# ssh-copy-id  -i  /root/.ssh/id_dsa.pub root@10.0.0.151
[root@mha03 ~]# ssh-copy-id  -i  /root/.ssh/id_dsa.pub root@10.0.0.152
 
 
二. 搭建数据库主从关系

 1.  修改数据库配置文件my.cnf,添加在 [mysqld] 模块内

          [root@mha01 ~]# /etc/init.d/mysqld stop      (关闭mysql,否则改配置后无法重启)
          [root@mha01 ~]# vim /etc/my.cnf
 
             server-id  = 151                 #以IP结尾作为server_id区分
   log-bin=mysql-bin             #开启主库从库的binlog日志
   relay_log_purge=0            #slave数据不丢失,relay log值为0表示手工清除
 
2.  初始化数据库并登录Mysql
 
(1)在所有Node节点增加MHA管理账户和复制账户
grant replication slave on *.* to 'rep'@'10.0.0.%' identified by '123456';
grant all on *.* to 'mhamanager'@'10.0.0.%' identified by '123456';
 (2)检查所有Node节点授权
 show grants for 'rep'@'10.0.0.%';
 show grants for 'mhamanager'@'%';
(3)确定所有的slave为只读,master为读写
Master> set global read_only=0;    #MASTER数据库操作
Slave > set global read_only=1;     #SLAVE数据库操作

(4)初始化数据库

[root@mha01 ~]#    /etc/init.d/mysqld start
[root@mha01 ~]#   mysql -uroot  -p123456
mysql> reset master;
mysql> reset slave;
(5)开启从库slave模式,并查看状态(只在从库上操作)
 
mysql -uroot -p123456 <<EOF
                      CHANGE MASTER TO
  MASTER_HOST='10.0.0.152',
  MASTER_PORT=3306,
  MASTER_USER='rep',
  MASTER_PASSWORD='123456',
  MASTER_LOG_FILE='mysql-bin.000001',
  MASTER_LOG_POS=107;
EOF
 
mysql> start slave;
          mysql> show slave status\G;
 
三. 搭建MHA安装环境及配置
1.Manager节点配置安装
mkdir  -p /etc/mha                                   #mha配置文件目录
mkdir  -p /data/mha/scripts                    #mha所需脚本总目录 
mkdir  -p /data/mha/scripts/mhm          #mha 故障转移所需脚本目录
mkdir  -p /data/mha/manager/app1      #mha 故障转移数据日志存储目录
mkdir  -p /data/mha/log/app1                #mha 日志目录
 
2.MHA manager节点安装 需要安装到管理节点
          yum install -y perl* perl-*
yum install -y python*
yum install cpan
yum install perl-DBD-MySQL
yum install perl-Config-Tiny
yum install perl-Log-Dispatch
yum install perl-Parallel-ForkManager
必须先安装node程序
          yum install mha4mysql-node-0.56-0.el6.noarch.rpm
然后安装Manager
          yum install mha4mysql-manager-0.56-0.el6.noarch.rpm
 
3.Node节点配置安装
mkdir  -p /data/mha/manager/app1      #mha 故障转移数据日志存储目录
mkdir  -p /data/mha/scripts                  #mha 节点所需维护脚本目录
 
MHA node节点安装,MHA node的node服务需要安装在所有的mysql服务器上
yum install -y perl* perl-*
yum install -y python*
yum install cpan
yum install perl-DBD-MySQL
yum install mha4mysql-node-0.56-0.el6.noarch.rpm
 
 
 
 
 
 
目录 /etc/mha 
 
[root@db-hma-manager-p-zw mha]# more app1.cnf 
 
[server default]
manager_log=/data/mha/log/app1/manager.log
manager_workdir=/data/mha/manager/app1
master_ip_failover_script=/data/mha/scripts/master_ip_failover --vip=192.168.7.234 --devic_1=em1  --devic_2=em1  --key=1 --net_mask=255.255.255.0 --mysql_conf=/etc/my.cnf --gatewa
y=192.168.7.1
master_ip_online_change_script=/data/mha/scripts/master_ip_online_change --vip=192.168.7.234 --key=1 --devic_1=em1  --devic_2=em1  --net_mask=255.255.255.0 --ssh_user=root --gatew
ay=192.168.7.1 --orig_master_ssh_port=22 --new_master_ssh_port=22
ping_interval=5
remote_workdir=/data/mha/manager/app1
repl_password=pass4slave
repl_user=replication
ssh_user=root
ssh_port=22
user=mhamanager
password=pass4mha
 
[server1]
candidate_master=1
hostname= 192.168.7.230
ignore_fail=1
master_binlog_dir="/data/data/mysqllogs/binlog"
port=3306
#check_repl_delay=0
check_repl_delay=0
 
[server2]
candidate_master=1
#no_master=1
hostname= 192.168.7.231
ignore_fail=1
master_binlog_dir="/data/data/mysqllogs/binlog"
port=3306
check_repl_delay=1 
 
[server3]
#candidate_master=1
no_master=1
hostname= 192.168.7.232
ignore_fail=1
master_binlog_dir="/data/data/mysqllogs/binlog"
port=3306
check_repl_delay=1 
 
[binlog1]
#candidate_master=1
no_master=1
hostname= 192.168.7.232
ignore_fail=1
master_binlog_dir="/data/data/mysqllogs/binlog"
 
http://www.cnblogs.com/wingsless/p/4033093.html

mysql> 6.10 修复宕机的Master

通常情况下自动切换以后,原master可能已经废弃掉,待原master主机修复后,如果数据完整的情况下,可能想把原来master重新作为新主库的slave,这时我们可以借助当时自动切换时刻的MHA日志来完成对原master的修复。下面是提取相关日志的命令:

[root@192.168.0.20 app1]# grep -i "All other slaves should start" /var/log/masterha/wssc/manager.log

Fri Feb 27 14:59:53 2015 - [info] All other slaves should start replication from here. Statement should be: CHANGE MASTER TO

MASTER_HOST='192.168.5.102', MASTER_PORT=3306,

MASTER_LOG_FILE='master-bin.000021', MASTER_LOG_POS=120,

MASTER_USER='repl_user', MASTER_PASSWORD='xxx';

posted @ 2018-03-08 09:33  秋天的风吹过原野  阅读(383)  评论(0编辑  收藏  举报