MMM高可用方案简介
MMM（Master-Master Replication Manager for MySQL）主主复制管理器，是一套提供了MySQL主主复制配置的监控、故障迁移和管理的可伸缩的脚本程序。在MMM高可用解决方案中，可以配置双主多从架构，通过MySQL Replication技术可以实现两台MySQL服务器互为主从，并且在任何时候只有一个节点可以写入，避免多节点写入的数据冲突，同时，当可写节点故障时，MMM套件可以立即监控到，然后将服务自动切换到另一个主节点继续提供服务，从而实现MySQL的高可用。
简而言之，通过MMM可以实现监控和管理MySQL主主复制和服务状态，同时也可以监控多个Slave节点的复制和运行状态，并且可以做到任意节点发生故障时实现自动切换的功能。MMM也为MySQL提供了读、写分离架构的良好平台。
MMM套件的优缺点
MMM集群套件具有良好的稳定性、高可用性和可扩展性。当活动的Master节点出现故障时，备用Master节点可以立即接管，而其他的Slave节点也能自动切换到备用Master节点继续进行同步复制，而无需人为干涉；MMM架构需要多个节点、多个IP，对服务器数量有要求，在读写非常繁忙的业务系统下表现不是很稳定，可能会出现复制延时、切换失效等问题。MMM方案并不太适应于对数据安全性要求很高，并且读、写繁忙的环境中。
MMM高可用套件工作原理
MMM套件主要的功能是通过以下三个脚本实现的：
1、mmm_mond：监控进程，运行在管理节点上，主要复制对所有数据库的监控工作，同时决定和处理所有节点的角色切换。
2、mmm_agentd：代理进程，运行在每一台MySQL服务器上，主要完成监控的测试工作和执行简单的远端服务设置。
3、mmm_control：一个简单的管理脚本，用来查看和管理集群运行状态，同时管理mmm_mond进程。
MMM高可用MySQL配置方案
在通过MMM套件实现的双Master架构中，需要5个IP地址，两个Master节点各有一个固定的物理IP，另外两个只读IP（reader IP）和一个可以IP（writer IP），这三个虚拟IP不会固定在任何一个节点上，相反，它会在两个Master节点之间来回切换（如何切换取决于节点的高可用）。在正常情况下Master1有两个虚拟IP（reader IP和writer IP），Master2有一个虚拟IP（reader IP），如果Master1故障，那么所有的reader和writer虚拟IP都会分配到Master上。
环境说明：
| 主机名 | IP地址 | 集群角色 | MySQL版本 | 系统版本 |
| Master1 | 192.168.1.210 | 主Master可读、写 | mysql-5.6.28 | CentOS6.7 |
| Master2 | 192.168.1.211 | 备Master可读、写 | mysql-5.6.28 | CentOS6.7 |
| Slave1 | 192.168.1.250 | Slave节点只读 | mysql-5.6.28 | CentOS6.7 |
| Slave2 | 192.168.1.209 | Slave节点只读 | mysql-5.6.28 | CentOS6.7 |
| Monitor | 192.168.1.21 | MMM服务管理端 | mysql-5.6.28 | CentOS6.7 |
虚拟IP地址：
| writer IP | 192.168.1.230 | 写入VIP，仅支持单节点写入 |
| reader IP | 192.168.1.231 | 只读VIP，每个数据库节点一个读VIP，可以通过LVS、HAproxy等负载均衡软件对读VIP做负载均衡 |
| reader IP | 192.168.1.232 | |
| reader IP | 192.168.1.233 | |
| reader IP | 192.168.1.234 |
MMM的安装和配置
step1：MMM套件的安装
1、在MMM管理端monitor安装MMM所有套件
[root@monitor ~]# rpm -ivh epel-release-6-8.noarch.rpm [root@monitor ~]# yum install mysql-mmm mysql-mmm-agent mysql-mmm-tools mysql-mmm-monitor
2、在各个MySQL节点上安装mysql-mmm-agent服务
[root@master1 ~]# yum install mysql-mmm-agent [root@master2 ~]# yum install mysql-mmm-agent [root@slave1 ~]# yum install mysql-mmm-agent [root@slave2 ~]# yum install mysql-mmm-agent
step2：Master1和两个Slave上配置主从（这里需要提前做好配置，Master1和Master2主主配置也一样）
[root@slave1 ~]# mysql -uroot -p741616710 mysql> grant replication slave on *.* to ‘repl‘@‘192.168.1.210‘ identified by ‘replpasswd‘; mysql> flush privileges; mysql> change master to -> master_host=‘192.168.1.210‘, -> master_user=‘repl‘, -> master_password=‘replpasswd‘, -> master_port=3306, -> master_log_file=‘mysql-bin.000034‘, -> master_log_pos=120; Query OK, 0 rows affected, 2 warnings (0.06 sec)
[root@slave2 ~]# mysql -uroot -p741616710 mysql> grant replication slave on *.* to ‘repl‘@‘192.168.1.210‘ identified by ‘replpasswd‘; mysql> flush privileges; mysql> change master to -> master_host=‘192.168.1.210‘, -> master_user=‘repl‘, -> master_password=‘replpasswd‘, -> master_port=3306, -> master_log_file=‘mysql-bin.000034‘, -> master_log_pos=120; Query OK, 0 rows affected, 2 warnings (0.02 sec)
step3：在所有的MySQL节点的/etc/my.cnf中增加参数
read_only=1
step4：在所有的MySQL节点添加以下两个用户
mysql> grant replication client on *.* to ‘mmm_monitor‘@‘192.168.1.%‘ identified by ‘monitorpasswd‘;
mysql> grant super, replication client, process on *.* to ‘mmm_agent‘@‘192.168.1.%‘ identified by ‘agentpasswd‘;
step5：在MMM管理端monitor上配置mmm_common.conf
[root@monitor ~]# ls /etc/mysql-mmm/mmm_agent.conf mmm_common.conf #在所有的节点配置，配置相同 mmm_mon.conf #仅在MMM管理端配置 mmm_mon_log.conf mmm_tools.conf
vim /etc/mysql-mmm/mmm_common.conf 所有的MMM节点配置相同
active_master_role writer <host default> cluster_interface eth0 pid_path /var/run/mysql-mmm/mmm_agentd.pid bin_path /usr/libexec/mysql-mmm/ replication_user replication replication_password replication agent_user mmm_agent agent_password agentpasswd </host> <host db1> ip 192.168.1.210 mode master peer db2 </host> <host db2> ip 192.168.1.211 mode master peer db1 </host> <host db3> ip 192.168.1.209 mode slave </host> <host db4> ip 192.168.1.250 mode slave </host> <role writer> hosts db1, db2 ips 192.168.1.230 mode exclusive </role> <role reader> hosts db1, db2, db3, db4 ips 192.168.1.231, 192.168.1.232, 192.168.1.233, 192.168.1.234 mode balanced </role>
step6：在MMM管理节点上配置mmm_mon.conf
[root@monitor ~]# vim /etc/mysql-mmm/mmm_mon.conf
include mmm_common.conf <monitor> ip 127.0.0.1 pid_path /var/run/mysql-mmm/mmm_mond.pid bin_path /usr/libexec/mysql-mmm status_path /var/lib/mysql-mmm/mmm_mond.status ping_ips 192.168.1.1, 192.168.1.2, 192.168.1.210, 192.168.1.211, 192.168.1.209, 192.168.1. 250 flap_duration 3600 flap_count 3 auto_set_online 8 # The kill_host_bin does not exist by default, though the monitor will # throw a warning about it missing. See the section 5.10 "Kill Host # Functionality" in the PDF documentation. # # kill_host_bin /usr/libexec/mysql-mmm/monitor/kill_host # </monitor> <host default> monitor_user mmm_monitor monitor_password monitorpasswd </host> debug 0
step7：在所有的MySQL节点配置mmm_agent.conf
[root@master1 mysql]# vim /etc/mysql-mmm/mmm_agent.conf include mmm_common.conf this db1 #在四台mysql节点上设置对应的db，分别为db1、db2、db3、db4
step8：所有节点设置ENABLED=1
cat /etc/default/mysql-mmm-agent # mysql-mmm-agent defaults ENABLED=1
step9：启动MMM服务
在MMM管理端启动服务
[root@monitor ~]# /etc/init.d/mysql-mmm-monitor start
Starting MMM Monitor Daemon: [ OK ]
在每个mysql节点启动服务
[root@master1 ~]# /etc/init.d/mysql-mmm-agent start
Starting MMM Agent Daemon: [ OK ]
查看集群运行状态
[root@monitor mysql-mmm]# mmm_control show db1(192.168.1.210) master/AWAITING_RECOVERY. Roles: db2(192.168.1.211) master/AWAITING_RECOVERY. Roles: db3(192.168.1.209) slave/AWAITING_RECOVERY. Roles: db4(192.168.1.250) slave/AWAITING_RECOVERY. Roles:
若一直出现上面AWAITING_RECOVERY的状态，可以手动设置各个MySQL节点为online状态
[root@monitor ~]# mmm_control set_online db1 OK: State of ‘db1‘ changed to ONLINE. Now you can wait some time and check its new roles! [root@monitor ~]# mmm_control set_online db2 OK: State of ‘db2‘ changed to ONLINE. Now you can wait some time and check its new roles! [root@monitor ~]# mmm_control set_online db3 OK: State of ‘db3‘ changed to ONLINE. Now you can wait some time and check its new roles! [root@monitor ~]# mmm_control set_online db4 OK: State of ‘db4‘ changed to ONLINE. Now you can wait some time and check its new roles! [root@monitor ~]# mmm_control show db1(192.168.1.210) master/ONLINE. Roles: reader(192.168.1.234), writer(192.168.1.230) db2(192.168.1.211) master/ONLINE. Roles: reader(192.168.1.231) db3(192.168.1.209) slave/ONLINE. Roles: reader(192.168.1.232) db4(192.168.1.250) slave/ONLINE. Roles: reader(192.168.1.233)
检测各个节点的运行状态
[root@monitor ~]# mmm_control checks all db4 ping [last change: 2016/02/27 05:13:57] OK db4 mysql [last change: 2016/02/27 05:13:57] OK db4 rep_threads [last change: 2016/02/27 05:13:57] OK db4 rep_backlog [last change: 2016/02/27 05:13:57] OK: Backlog is null db2 ping [last change: 2016/02/27 05:13:57] OK db2 mysql [last change: 2016/02/27 05:13:57] OK db2 rep_threads [last change: 2016/02/27 05:13:57] OK db2 rep_backlog [last change: 2016/02/27 05:13:57] OK: Backlog is null db3 ping [last change: 2016/02/27 05:13:57] OK db3 mysql [last change: 2016/02/27 05:13:57] OK db3 rep_threads [last change: 2016/02/27 05:13:57] OK db3 rep_backlog [last change: 2016/02/27 05:13:57] OK: Backlog is null db1 ping [last change: 2016/02/27 05:13:57] OK db1 mysql [last change: 2016/02/27 05:13:57] OK db1 rep_threads [last change: 2016/02/27 05:13:57] OK db1 rep_backlog [last change: 2016/02/27 05:13:57] OK: Backlog is null
step10：查看各节点虚拟IP分配情况
Master1
[root@master1 ~]# ip a |grep eth0 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000 inet 192.168.1.210/24 brd 192.168.1.255 scope global eth0 inet 192.168.1.234/32 scope global eth0 inet 192.168.1.230/32 scope global eth0
Mster2
[root@master2 ~]# ip a |grep eth0 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000 inet 192.168.1.211/24 brd 192.168.1.255 scope global eth0 inet 192.168.1.231/32 scope global eth0
Slave1
[root@slave1 ~]# ip a |grep eth0 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000 inet 192.168.1.250/24 brd 192.168.1.255 scope global eth0 inet 192.168.1.213/32 scope global eth0 inet 192.168.1.233/32 scope global eth0
Slave2
[root@slave2 ~]# ip a |grep eth0 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000 inet 192.168.1.209/24 brd 192.168.1.255 scope global eth0 inet 192.168.1.232/32 scope global eth0
step11：测试MMM实现MySQL高可用
1、授权一个可以通过远程使用VIP登陆集群的用户
mysql> grant all on *.* to ‘hm‘@‘192.168.1.%‘ identified by ‘741616710‘;
2、使用VIP192.168.1.230登陆，并做相关测试，在各个节点查看是否同步数据库
[root@monitor ~]# mysql -uhm -p741616710 -h192.168.1.230
mysql> show variables like ‘hostname%‘; +---------------+---------+ | Variable_name | Value | +---------------+---------+ | hostname | master1 | +---------------+---------+ 1 row in set (0.01 sec) mysql> create database test1; Query OK, 1 row affected (0.00 sec) mysql> use test1 Database changed mysql> create table tt1(id int, name varchar(20)); Query OK, 0 rows affected (0.13 sec) mysql> insert into tt1(id,name) values(1,‘july‘),(2,‘dime‘); Query OK, 2 rows affected (0.04 sec) Records: 2 Duplicates: 0 Warnings: 0 mysql> select * from tt1; +------+------+ | id | name | +------+------+ | 1 | july | | 2 | dime | +------+------+ 2 rows in set (0.00 sec)
step12：测试MMM故障转移功能
1、关闭Master1上的MySQL服务，查看状态
[root@monitor ~]# mmm_control show db1(192.168.1.210) master/HARD_OFFLINE. Roles: db2(192.168.1.211) master/ONLINE. Roles: reader(192.168.1.231), writer(192.168.1.230) db3(192.168.1.209) slave/ONLINE. Roles: reader(192.168.1.232), reader(192.168.1.234) db4(192.168.1.250) slave/ONLINE. Roles: reader(192.168.1.233)
[root@monitor ~]# mmm_control set_online db1 OK: State of ‘db1‘ changed to ONLINE. Now you can wait some time and check its new roles! [root@monitor ~]# mmm_control show db1(192.168.1.210) master/ONLINE. Roles: reader(192.168.1.232) db2(192.168.1.211) master/ONLINE. Roles: reader(192.168.1.231), writer(192.168.1.230) db3(192.168.1.209) slave/ONLINE. Roles: reader(192.168.1.234) db4(192.168.1.250) slave/ONLINE. Roles: reader(192.168.1.233)
2、Master1故障恢复后，若想让VIP继续回到Master1上，则可以按以下手动设置
[root@monitor ~]# mmm_control move_role writer db1 OK: Role ‘writer‘ has been moved from ‘db2‘ to ‘db1‘. Now you can wait some time and check new roles info!
[root@monitor ~]# mmm_control show db1(192.168.1.210) master/ONLINE. Roles: reader(192.168.1.232), writer(192.168.1.230) db2(192.168.1.211) master/ONLINE. Roles: reader(192.168.1.231) db3(192.168.1.209) slave/ONLINE. Roles: reader(192.168.1.234) db4(192.168.1.250) slave/ONLINE. Roles: reader(192.168.1.233)
原文地址：http://7424593.blog.51cto.com/7414593/1745513
