原创作品,允许转载,转载时请务必以超链接形式标明文章 原始出处 、作者信息和本声明。否则将追究法律责任。http://hcymysql.blog.51cto.com/5223301/982561

MySQL架构为master-slave(主从),master故障自动切换到slave上。当然也可以设置为双master,但这里有个弊端:就是当主的压力很大时,从上延时很大,比如落后2000秒,此时主挂了,从接管(VIP漂移到从),用户刚才发表的文章,此时因为同步延时大,还没复制过来,于是用户又发表了一篇文章,当原来的master修好后,因从的IO和SQL线程还在开启状态,还会继续同步刚才没有同步复制完的数据,这时有可能把用户新发表的文章更改掉,造成用户数据丢失。

考虑到这种情况,我这里还是用的master-slave(主从)架构。

keepalive安装很简单,这里不再啰嗦。主要看下配置文件和脚本:

  1. # more /etc/keepalived/keepalived.conf
  2. global_defs {
  3. router_id KeepAlive_Mysql
  4. }
  5. vrrp_script check_run {
  6. script "/root/sh/mysql_check.sh"
  7. interval 300
  8. }
  9. vrrp_sync_group VG1 {
  10. group {
  11. VI_1
  12. }
  13. }
  14. vrrp_instance VI_1 {
  15. state BACKUP
  16. interface eth0
  17. virtual_router_id 51
  18. priority 100
  19. advert_int 1
  20. nopreempt
  21. authentication {
  22. auth_type PASS
  23. auth_pass 1111
  24. }
  25. track_script {
  26. check_run
  27. }
  28. notify_master /root/sh/master.sh
  29. notify_backup /root/sh/backup.sh
  30. notify_stop /root/sh/stop.sh
  31. virtual_ipaddress {
  32. 192.168.8.150
  33. }
  34. }
notify_master <STRING>|<QUOTED-STRING>    # 状态改变为MASTER后执行的脚本
notify_backup <STRING>|<QUOTED-STRING>    # 状态改变为BACKUP后执行的脚本
notify_fault <STRING>|<QUOTED-STRING>    # 状态改变为FAULT后执行的脚本
notify_stop <STRING>|<QUOTED-STRING>    # VRRP停止后后执行的脚本
notify <STRING>|<QUOTED-STRING>        # (1)任意状态改变后执行的脚本

下面解释下这4个脚本的用法:

mysql_check.sh(健康检查脚本,当发现mysql连接不上,会把keepalive进程关闭,并切换。)
  1. # more mysql_check.sh
  2. #!/bin/bash
  3. . /root/.bash_profile
  4. count=1
  5. while true
  6. do
  7. mysql -e "show status;" > /dev/null 2>&1
  8. i=$?
  9. ps aux | grep mysqld | grep -v grep > /dev/null 2>&1
  10. j=$?
  11. if [ $i = 0 ] && [ $j = 0 ]
  12. then
  13. exit 0
  14. else
  15. if [ $i = 1 ] && [ $j = 0 ]
  16. then
  17. exit 0
  18. else
  19. if [ $count -gt 5 ]
  20. then
  21. break
  22. fi
  23. let count++
  24. continue
  25. fi
  26. fi
  27. done
  28. /etc/init.d/keepalived stop
 master.sh(状态改变为MASTER后执行的脚本)  (首先判断同步复制是否执行完毕,如果未执行完毕,等1分钟后,不论是否执行完毕,都跳过,并停止同步复制进程。) (其次,更改前端程序连接的业务账号admin的权限和密码,并记录当前切换以后的日志和POS点。)
  1. # more master.sh
  2. #!/bin/bash
  3. . /root/.bash_profile
  4. Master_Log_File=$(mysql -e "show slave status\G" | grep -w Master_Log_File | awk -F": " '{print $2}')
  5. Relay_Master_Log_File=$(mysql -e "show slave status\G" | grep -w Relay_Master_Log_File | awk -F": " '{print $2}')
  6. Read_Master_Log_Pos=$(mysql -e "show slave status\G" | grep -w Read_Master_Log_Pos | awk -F": " '{print $2}')
  7. Exec_Master_Log_Pos=$(mysql -e "show slave status\G" | grep -w Exec_Master_Log_Pos | awk -F": " '{print $2}')
  8. i=1
  9. while true
  10. do
  11. if [ $Master_Log_File = $Relay_Master_Log_File ] && [ $Read_Master_Log_Pos -eq $Exec_Master_Log_Pos ]
  12. then
  13. echo "ok"
  14. break
  15. else
  16. sleep 1
  17. if [ $i -gt 60 ]
  18. then
  19. break
  20. fi
  21. continue
  22. let i++
  23. fi
  24. done
  25. mysql -e "stop slave;"
  26. mysql -e "flush logs;GRANT ALL PRIVILEGES ON *.* TO 'admin'@'%' IDENTIFIED BY 'admin';flush privileges;"
  27. mysql -e "show master status;" > /tmp/master_status_$(date "+%y%m%d-%H%M").txt
backup.sh(状态改变为BACKUP后执行的脚本)
  1. # more backup.sh
  2. #!/bin/bash
  3. . /root/.bash_profile
  4. mysql -e "GRANT ALL PRIVILEGES ON *.* TO 'admin'@'%' IDENTIFIED BY '1q2w3e4r';flush privileges;"
 stop.sh(keepalived停止后后执行的脚本) (首先把admin密码更改掉) (其次,设置参数,保证不丢失数据) (最后,查看是否还有写操作,不论是否执行完毕,1分钟后都退出。)
  1. # more stop.sh
  2. #!/bin/bash
  3. . /root/.bash_profile
  4. mysql -e "GRANT ALL PRIVILEGES ON *.* TO 'admin'@'%' IDENTIFIED BY '1q2w3e4r';flush privileges;"
  5. mysql -e "set global innodb_support_xa=1;"
  6. mysql -e "set global sync_binlog=1;"
  7. mysql -e "set global innodb_flush_log_at_trx_commit=1;"
  8. M_File1=$(mysql -e "show master status\G" | awk -F': ' '/File/{print $2}')
  9. M_Position1=$(mysql -e "show master status\G" | awk -F': ' '/Position/{print $2}')
  10. sleep 1
  11. M_File2=$(mysql -e "show master status\G" | awk -F': ' '/File/{print $2}')
  12. M_Position2=$(mysql -e "show master status\G" | awk -F': ' '/Position/{print $2}')
  13. i=1
  14. while true
  15. do
  16. if [ $M_File1 = $M_File2 ] && [ $M_Position1 -eq $M_Position2 ]
  17. then
  18. echo "ok"
  19. break
  20. else
  21. sleep 1
  22. if [ $i -gt 60 ]
  23. then
  24. break
  25. fi
  26. continue
  27. let i++
  28. fi
  29. done