• 《综合》MMM集群


    《综合》MMM集群

    1. 部署集群基础环境
    2. MySQL-MMM架构部署
    3. MySQL-MMM架构使用

    1 部署集群基础环境

    1.1 问题

    本案例要求为MySQL集群准备基础环境,完成以下任务操作:

    • 数据库授权
    • 部署MySQL双主多从结构
    • 配置本机hosts解析记录

    1.2 方案

    使用4台RHEL 6虚拟机,如图-1所示。其中192.168.4.10、192.168.4.11作为MySQL双主服务器,192.168.4.12、192.168.4.13作为主服务器的从服务器。

    图-1

    1.3 步骤

    实现此案例需要按照如下步骤进行。

    步骤一:准备环境

    1. [root@master1 ~]# cat /etc/hosts
    2. 127.0.0.1   localhost localhost.localdomain localhost4 localhost4.localdomain4
    3. ::1         localhost localhost.localdomain localhost6 localhost6.localdomain6
    4. 192.168.4.10     master1         master1.tarena.com
    5. 192.168.4.11     master2         master2.tarena.com
    6. 192.168.4.12     slave1            slave1.tarena.com
    7. 192.168.4.13     slave2            slave2.tarena.com
    8. 192.168.4.100   master1         master1.tarena.com
    9. [root@master1 ~]# ping -c 2 master1
    10. PING master1 (192.168.4.10) 56(84) bytes of data.
    11. 64 bytes from master1 (192.168.4.10): icmp_seq=1 ttl=64 time=0.378 ms
    12. 64 bytes from master1 (192.168.4.10): icmp_seq=2 ttl=64 time=0.396 ms
    13. --- master1 ping statistics ---
    14. 2 packets transmitted, 2 received, 0% packet loss, time 1001ms
    15. rtt min/avg/max/mdev = 0.378/0.387/0.396/0.009 ms
    16. [root@master1 ~]#

    步骤二:部署数据库主机

    1)安装启动数据库(4台数据库主机master1,master2,slave1,slave2执行以下操作)

    1. [root@master1 ~]# tar xvf MySQL-5.6.15-1.el6.x86_64.rpm-bundle.tar    //解压软件包
    2. .. ..
    3. [root@master1 ~]# rpm -Uvh MySQL-*.rpm                                //安装MySQL
    4. .. ..
    5. [root@master1 ~]# service mysql start
    6. Starting MySQL.                                            [确定]

    2)初始化配置数据库(4台数据库主机master1,master2,slave1,slave2执行以下操作)

    1. [root@master1 ~]# cat /root/.mysql_secret         //查看随机生成密码
    2. # The random password set for the root user at Thu May  7 22:15:47 2015 (local time): wW1BNAjD
    3. [root@master1 ~]# mysql -uroot -pwW1BNAjD        //使用随机生成密码登陆
    4. Warning: Using a password on the command line interface can be insecure.
    5. Welcome to the MySQL monitor.  Commands end with ; or g.
    6. Your MySQL connection id is 1
    7. Server version: 5.6.15
    8. Copyright (c) 2000, 2013, Oracle and/or its affiliates. All rights reserved.
    9. Oracle is a registered trademark of Oracle Corporation and/or its
    10. affiliates. Other names may be trademarks of their respective
    11. owners.
    12. Type 'help;' or 'hfor help. Type 'c' to clear the current input statement.
    13. mysql> set password=password("pwd123");          //修改数据库root密码
    14. Query OK, 0 rows affected (0.49 sec)
    15. mysql> exit
    16. Bye
    17. [root@master1 ~]#

    步骤三:部署双主多从结构

    1)数据库授权(4台数据库主机master1,master2,slave1,slave2执行以下操作)

    部署主从同步只需要授权一个主从同步用户即可,但是我们要部署MySQL-MMM架构,所以在这里我们将MySQL-MMM所需用户一并进行授权设置。再授权一个测试用户,在架构搭建完成时测试使用。

    1. [root@master1 ~]# mysql -uroot -ppwd123
    2. Warning: Using a password on the command line interface can be insecure.
    3. Welcome to the MySQL monitor.  Commands end with ; or g.
    4. Your MySQL connection id is 2
    5. Server version: 5.6.15 MySQL Community Server (GPL)
    6. Copyright (c) 2000, 2013, Oracle and/or its affiliates. All rights reserved.
    7. Oracle is a registered trademark of Oracle Corporation and/or its
    8. affiliates. Other names may be trademarks of their respective
    9. owners.
    10. Type 'help;' or 'hfor help. Type 'c' to clear the current input statement.

    数据库授权部分为了方便试验我们直接允许所有地址访问了,真实环境需谨慎

    1. mysql> grant   replication  slave  on  *.*  to  slaveuser@"%" identified by  "pwd123";                                //主从同步授权
    2. Query OK, 0 rows affected (0.00 sec)
    3. mysql> grant  replication  client  on *.*  to  monitor@"%" identified by "monitor";                                         //MMM所需架构用户授权
    4. Query OK, 0 rows affected (0.06 sec)        
    5. mysql> grant  replication client,process,super   on *.*  to  agent@"%" identified by "agent";                                 //MMM所需架构用户授权
    6. Query OK, 0 rows affected (0.00 sec)
    7. mysql> grant  all  on *.*  to  root@"%" identified by "pwd123";  //测试用户授权
    8. Query OK, 0 rows affected (0.00 sec)
    9. mysql>

    2)开启主数据库binlog日志、设置server_id(master1,master2)

    master1设置:

    1. [root@master1 ~]# cat /etc/my.cnf
    2. [mysqld]
    3. datadir=/var/lib/mysql
    4. socket=/var/lib/mysql/mysql.sock
    5. user=mysql
    6. server_id=10                        //设置server_id,该值集群中不可以重复
    7. log-bin                                   //开启bin-log日志
    8. # Disabling symbolic-links is recommended to prevent assorted security risks
    9. symbolic-links=0
    10. [mysqld_safe]
    11. log-error=/var/log/mysqld.log
    12. pid-file=/var/run/mysqld/mysqld.pid
    13. [root@master1 ~]# service mysql restart                   //重启MySQL服务
    14. Shutting down MySQL..                                              [确定]
    15. Starting MySQL..                                                         [确定]
    16. [root@master1 ~]# ls /var/lib/mysql/master1-bin*      //查看binlog日志是否生成
    17. /var/lib/mysql/master1-bin.000001  /var/lib/mysql/master1-bin.index
    18. [root@master1 ~]#

    master2设置:

    1. [root@master2 ~]# cat /etc/my.cnf
    2. [mysqld]
    3. datadir=/var/lib/mysql
    4. socket=/var/lib/mysql/mysql.sock
    5. user=mysql
    6. server_id=11
    7. log-bin
    8. # Disabling symbolic-links is recommended to prevent assorted security risks
    9. symbolic-links=0
    10. [mysqld_safe]
    11. log-error=/var/log/mysqld.log
    12. pid-file=/var/run/mysqld/mysqld.pid
    13. [root@master2 ~]# service mysql restart
    14. Shutting down MySQL..                                      [确定]
    15. Starting MySQL.                                                  [确定]
    16. [root@master2 ~]# ls /var/lib/mysql/master2-bin.*
    17. /var/lib/mysql/master2-bin.000001  /var/lib/mysql/master2-bin.index

    3)从库设置server_id

    slave1设置:

    1. [root@slave1 ~]# cat /etc/my.cnf
    2. [mysqld]
    3. datadir=/var/lib/mysql
    4. socket=/var/lib/mysql/mysql.sock
    5. user=mysql
    6. server_id=12
    7. # Disabling symbolic-links is recommended to prevent assorted security risks
    8. symbolic-links=0
    9. [mysqld_safe]
    10. log-error=/var/log/mysqld.log
    11. pid-file=/var/run/mysqld/mysqld.pid
    12. [root@slave1 ~]# service mysql restart
    13. Shutting down MySQL..                                      [确定]
    14. Starting MySQL..                                                [确定]
    15. [root@slave1 ~]#

    slave2设置:

    1. [root@slave2 ~]# cat /etc/my.cnf
    2. [mysqld]
    3. datadir=/var/lib/mysql
    4. socket=/var/lib/mysql/mysql.sock
    5. user=mysql
    6. server_id=13
    7. # Disabling symbolic-links is recommended to prevent assorted security risks
    8. symbolic-links=0
    9. [mysqld_safe]
    10. log-error=/var/log/mysqld.log
    11. pid-file=/var/run/mysqld/mysqld.pid
    12. [root@slave2 ~]# service mysql restart
    13. Shutting down MySQL..                                      [确定]
    14. Starting MySQL.                                                 [确定]
    15. [root@slave2 ~]#

    4)配置主从从从关系

    配置master2、slave1、slave2成为master1的从服务器

    查看master1服务器binlong日志使用节点信息:

    1. [root@master1 ~]# mysql -uroot -ppwd123
    2. .. ..
    3. mysql> show master statusG
    4. *************************** 1. row ***************************
    5. File: master1-bin.000001
    6. Position: 120
    7. Binlog_Do_DB: 
    8. Binlog_Ignore_DB: 
    9. Executed_Gtid_Set: 
    10. 1 row in set (0.00 sec)
    11. mysql>

    设置master2为master1从:

    1. [root@master2 ~]# mysql -uroot -ppwd123
    2. .. ..
    3. mysql> change  master  to                            //设置主服务器信息
    4. -> master_host="192.168.4.10",                    //设置主服务器IP地址
    5. -> master_user="slaveuser",                         //设置主从同步用户
    6. -> master_password="pwd123",                    //设置主从同步密码
    7. -> master_log_file="master1-bin.000001",    //设置主库binlog日志名称
    8. -> master_log_pos=120;                               //设置主从binlog日志使用节点
    9. Query OK, 0 rows affected, 2 warnings (0.06 sec)
    10. mysql> start slave;                                         //启动同步进程
    11. Query OK, 0 rows affected (0.00 sec)
    12. mysql> show slave statusG                           //查看主从是否成功
    13. .. ..

    启动同步进程后查看IO节点和SQL节点是否为Yes如果均为Yes表示主从正常。

    1. Slave_IO_Running: Yes                    //IO节点正常
    2. Slave_SQL_Running: Yes                //SQL节点正常
    3. .. ..
    4. mysql>

    设置slave1为master1从:

    1. [root@slave1 ~]# mysql -uroot -ppwd123
    2. .. ..
    3. mysql> change  master  to
    4. -> master_host="192.168.4.10",
    5. -> master_user="slaveuser",
    6. -> master_password="pwd123",
    7. -> master_log_file="master1-bin.000001",                    
    8. -> master_log_pos=120;
    9. Query OK, 0 rows affected, 2 warnings (0.12 sec)
    10. mysql> start slave;
    11. Query OK, 0 rows affected (0.16 sec)
    12. mysql> show slave statusG
    13. .. ..
    14. Slave_IO_Running: Yes                   //IO节点正常
    15. Slave_SQL_Running: Yes                //SQL节点正常
    16. .. ..
    17. mysql>

    设置slave2为master1从:

    1. [root@slave2 ~]# mysql -uroot -ppwd123
    2. .. ..
    3. mysql> change  master  to
    4. -> master_host="192.168.4.10",
    5. -> master_user="slaveuser",
    6. -> master_password="pwd123",
    7. -> master_log_file="master1-bin.000001",                    
    8. -> master_log_pos=120;
    9. Query OK, 0 rows affected, 2 warnings (0.13 sec)
    10. mysql> start slave;
    11. Query OK, 0 rows affected (0.27 sec)
    12. mysql> show slave statusG
    13. .. ..
    14. Slave_IO_Running: Yes                   //IO节点正常
    15. Slave_SQL_Running: Yes                //SQL节点正常
    16. .. ..
    17. mysql>

    5)配置主主从从关系,将master1配置为master2的从

    查看master2的binlog使用信息:

    1. [root@master2 ~]# mysql -uroot -ppwd123
    2. .. ..
    3. mysql> show master statusG
    4. *************************** 1. row ***************************
    5. File: master2-bin.000001
    6. Position: 120
    7. Binlog_Do_DB: 
    8. Binlog_Ignore_DB: 
    9. Executed_Gtid_Set: 
    10. 1 row in set (0.00 sec)
    11. mysql>

    设置master1成为master2的从:

    1. [root@master1 ~]# mysql -uroot -ppwd123
    2. .. ..
    3. mysql> change  master  to
    4. -> master_host="192.168.4.11",
    5. -> master_user="slaveuser",
    6. -> master_password="pwd123",
    7. -> master_log_file="master2-bin.000001",                    
    8. -> master_log_pos=120;
    9. Query OK, 0 rows affected, 2 warnings (0.31 sec)
    10. mysql> start slave;
    11. Query OK, 0 rows affected (0.27 sec)
    12. mysql> show slave statusG
    13. .. ..
    14. Slave_IO_Running: Yes                    //IO节点正常
    15. Slave_SQL_Running: Yes                //SQL节点正常
    16. .. ..
    17. mysql>

    6)测试主从架构是否成功

    master1更新数据,查看其它主机是否同步:

    1. [root@master1 ~]# mysql -uroot -ppwd123
    2. .. ..
    3. mysql> show databases;
    4. +--------------------+
    5. | Database           |
    6. +--------------------+
    7. | information_schema |
    8. | mysql              |
    9. | performance_schema |
    10. | test               |
    11. +--------------------+
    12. 4 rows in set (0.00 sec)
    13. mysql> create database tarena;
    14. Query OK, 1 row affected (0.06 sec)
    15. mysql> show databases;
    16. +--------------------+
    17. | Database           |
    18. +--------------------+
    19. | information_schema |
    20. | mysql              |
    21. | performance_schema |
    22. | tarena             |
    23. | test               |
    24. +--------------------+
    25. 5 rows in set (0.00 sec)
    26. mysql>

    master2主机查看:

    1. [root@master2 ~]# mysql -uroot -ppwd123 -e "show databases"
    2. Warning: Using a password on the command line interface can be insecure.
    3. +--------------------+
    4. | Database           |
    5. +--------------------+
    6. | information_schema |
    7. | mysql              |
    8. | performance_schema |
    9. | tarena             |
    10. | test               |
    11. +--------------------+
    12. [root@master2 ~]#

    slave1主机查看:

    1. [root@slave1 ~]# mysql -uroot -ppwd123 -e "show databases"
    2. Warning: Using a password on the command line interface can be insecure.
    3. +--------------------+
    4. | Database           |
    5. +--------------------+
    6. | information_schema |
    7. | mysql              |
    8. | performance_schema |
    9. | tarena             |
    10. | test               |
    11. +--------------------+
    12. [root@slave1 ~]#

    slave2主机查看:

    1. [root@slave2 ~]# mysql -uroot -ppwd123 -e "show databases"
    2. Warning: Using a password on the command line interface can be insecure.
    3. +--------------------+
    4. | Database           |
    5. +--------------------+
    6. | information_schema |
    7. | mysql              |
    8. | performance_schema |
    9. | tarena             |
    10. | test               |
    11. +--------------------+
    12. [root@slave2 ~]#

     

    2 MySQL-MMM架构部署

    2.1 问题

    本案例要求熟悉实现MySQL-MMM的架构部署,主要包括以下任务:

    1. 安装依赖包
    2. 安装软件包
    3. 配置MySQL-MMM

    2.2 方案

    使用5台RHEL 6虚拟机,如图-2所示。其中192.168.4.10、192.168.4.11作为MySQL双主服务器,192.168.4.12、192.168.4.13作为主服务器的从服务器,192.168.4.100作为MySQL-MMM架构中管理监控服务器,实现监控MySQL主从服务器的工作状态及决定故障节点的移除或恢复工作,架构搭建完成后使用客户机192.168.4.120进行访问,客户机需要安装MySQL-client软件包。

     

    -2

    2.3 步骤

    实现此案例需要按照如下步骤进行。

    步骤一:安装MySQL-MMM

    1)安装依赖关系(MySQL集群内5台服务器master1,master2,slave1,slave2,monitor)均需安装

    1. [root@master1 ~]# yum -y install gcc* perl-Date-Manip  perl-Date-Manip  perl-Date-Manip perl-XML-DOM-XPath perl-XML-Parser perl-XML-RegExp rrdtool perl-Class-Singleton perl perl-DBD-MySQL perl-Params-Validate perl-MailTools perl-Time-HiRes
    2. .. ..

    2)安装MySQL-MMM软件依赖包(MySQL集群内5台服务器master1,master2,slave1,slave2,monitor)均需安装,软件包讲师提供

    安装安装Log-Log4perl 类

    1. [root@mysql-master1 ~]# rpm -ivh perl-Log-Log4perl-1.26-1.el6.rf.noarch.rpm
    2. warning: perl-Log-Log4perl-1.26-1.el6.rf.noarch.rpm: Header V3 DSA/SHA1 Signature, key ID 6b8d79e6: NOKEY
    3. Preparing...                ######################### [100%]
    4. 1:perl-Log-Log4perl        ########################  [100%]

    安装Algorithm-Diff类

    1. [root@mysql-master1 ~]# tar -zxvf Algorithm-Diff-1.1902.tar.gz  //解压安装包
    2. .. .. 
    3. [root@mysql-master1 ~]# cd Algorithm-Diff-1.1902                //切换到安装目录
    4. [root@mysql-master1 Algorithm-Diff-1.1902]# perl  Makefile.PL    //生成makefile文件
    5. Checking if your kit is complete...
    6. Looks good
    7. Writing Makefile for Algorithm::Diff
    8. [root@mysql-master1 Algorithm-Diff-1.1902]# make && make install        //编译,编译安装
    9. .. ..
    10. [root@mysql-master1 Algorithm-Diff-1.1902]# cd                //切换到软件包目录
    11. [root@mysql-master1 ~]#

    安装Proc-Daemon类

    1. [root@mysql-master1 ~]# tar -zxvf Proc-Daemon-0.03.tar.gz    //解压安装包
    2. .. ..
    3. [root@mysql-master1 ~]# cd Proc-Daemon-0.03                    //切换到安装目录
    4. [root@mysql-master1 Proc-Daemon-0.03]# perl    Makefile.PL    //生成makefile文件
    5. Checking if your kit is complete...
    6. Looks good
    7. Writing Makefile for Proc::Daemon
    8. [root@mysql-master1 Proc-Daemon-0.03]# make && make install    //编译,编译安装
    9. .. ..
    10. [root@mysql-master1 Proc-Daemon-0.03]# cd                    //切换到软件包目录
    11. [root@mysql-master1 ~]#

    安装Net-ARP虚拟IP分配工具:

    1. [root@mysql-master1 ~]# gunzip Net-ARP-1.0.8.tgz    //使用gunzip解压tgz格式的安装包
    2. [root@mysql-master1 ~]# tar xvf Net-ARP-1.0.8.tar            //解压tar安装包
    3. .. ..
    4. [root@mysql-master1 ~]# cd Net-ARP-1.0.8                    //切换到安装目录
    5. [root@mysql-master1 Net-ARP-1.0.8]# perl Makefile.PL        //生成makefile文件
    6. Module Net::Pcap is required for make test!
    7. Checking if your kit is complete...
    8. Looks good
    9. Writing Makefile for Net::ARP
    10. [root@mysql-master1 Net-ARP-1.0.8]# make && make install    //编译,编译安装
    11. .. ..
    12. [root@mysql-master1 Net-ARP-1.0.8]# cd                        //切换到软件包目录
    13. [root@mysql-master1 ~]#

    安装Mysql-MMM软件包:

    1. [root@mysql-master1 ~]# tar xvf mysql-mmm-2.2.1.tar.gz        //解压安装包
    2. .. ..
    3. [root@mysql-master1 ~]# cd mysql-mmm-2.2.1                    //切换到安装目录
    4. [root@mysql-master1 mysql-mmm-2.2.1]# make && make install    //编译,编译安装
    5. .. ..
    6. [root@mysql-master1 mysql-mmm-2.2.1]#

    步骤二:修改配置文件

    1)修改公共配置文件

    本案例中MySQL集群的5台服务器(master1、master2、slave1、slave2、monitor)都需要配置,可以先配好一台后使用scp复制。

    1. [root@master1 ~]# vim  /etc/mysql-mmm/mmm_common.conf
    2. active_master_role    writer
    3. <host default>
    4. cluster_interface        eth0                //设置主从同步的用户
    5. pid_path                /var/run/mmm_agentd.pid
    6. bin_path                /usr/lib/mysql-mmm/
    7. replication_user        slaveuser            //设置主从同步的用户
    8. replication_password    pwd123            //设置主从同步用户密码
    9. agent_user            agent                //mmm-agent控制数据库用户
    10. agent_password        agent                //mmm-agent控制数据库用户密码
    11. </host>
    12. <host master1>                            //设置第一个主服务器
    13. ip                    192.168.4.10            //master1 IP 地址
    14. mode                    master
    15. peer                    master2                //指定另外一台主服务器
    16. </host>
    17. <host master2>                            //指定另外一台主服务器
    18. ip                    192.168.4.11
    19. mode                    master
    20. peer                    master1
    21. </host>
    22. <host slave1>                                //设置第一台从服务器
    23. ip                    192.168.4.12            //slave1 IP 地址
    24. mode                    slave                //本段落配置的是slave服务器
    25. </host>
    26. <host slave2>
    27. ip                    192.168.4.13
    28. mode                    slave
    29. </host>
    30. <role writer>                                //设置写入服务器工作模式
    31. hosts                master1,master2        //提供写的主服务器
    32. ips                    192.168.4.200        //设置VIP地址
    33. mode                    exclusive            //排他模式
    34. </role>
    35. <role reader>                                //设置读取服务器工作模式
    36. hosts                slave1,slave2        //提供读的服务器信息
    37. ips                    192.168.4.201,192.168.4.202    //多个虚拟IP
    38. mode                    balanced                        //均衡模式
    39. </role>
    40. [root@master1 ~]#

    2)修改管理主机配置文件(monitor主机配置)

    1. [root@monitor ~]# vim /etc/mysql-mmm/mmm_mon.conf
    2. include mmm_common.conf
    3. <monitor>
    4. ip                        192.168.4.100        //设置管理主机IP地址
    5. pid_path                /var/run/mmm_mond.pid
    6. bin_path                /usr/lib/mysql-mmm/
    7. status_path                /var/lib/misc/mmm_mond.status
    8. ping_ips                192.168.4.10,192.168.4.11,192.168.4.12,192.168.4.13
    9. //设置被监控数据库
    10. </monitor>
    11. <host default>
    12. monitor_user            monitor                    //监控数据库MySQL用户
          monitor_password        monitor                    //监控数据库MySQL用户密码
    13. </host>
    14. debug 0
    15. [root@monitor ~]#

    3)修改客户端配置文件

    master1配置

    1. [root@master1 ~]# cat /etc/mysql-mmm/mmm_agent.conf
    2. include mmm_common.conf
    3. this master1

    master2配置

    1. [root@master2 ~]# cat /etc/mysql-mmm/mmm_agent.conf
    2. include mmm_common.conf
    3. this master2

    slave1配置

    1. [root@slave1 ~]# cat /etc/mysql-mmm/mmm_agent.conf
    2. include mmm_common.conf
    3. this slave1

    slave2配置

    1. [root@slave2 ~]# cat /etc/mysql-mmm/mmm_agent.conf
    2. include mmm_common.conf
    3. this slave2

     

    3 MySQL-MMM架构使用

    3.1 问题

    本案例要求基于普通版的MySQL服务器改造MMM架构,完成以下任务操作:

    • 启动MMM集群架构
    • 设置集群中服务器为online状态

    3.2 方案

    MySQL-MMM架构部署完成后需要启动,数据库端启动mmm-agent进程,管理端启动mmm-monitor进程,启动完成后设置所有数据库主机状态为online。

    3.3 步骤

    实现此案例需要按照如下步骤进行。

    步骤一:启动MMM集群架构

    1)启动mmm-agent进程

    master1操作:

    1. [root@master1 ~]# /etc/init.d/mysql-mmm-agent start
    2. Daemon bin: '/usr/sbin/mmm_agentd'
    3. Daemon pid: '/var/run/mmm_agentd.pid'
    4. Starting MMM Agent daemon... Ok

    master2操作:

    1. [root@master2 ~]# /etc/init.d/mysql-mmm-agent start
    2. Daemon bin: '/usr/sbin/mmm_agentd'
    3. Daemon pid: '/var/run/mmm_agentd.pid'
    4. Starting MMM Agent daemon... Ok

    slave1操作:

    1. [root@master2 ~]# /etc/init.d/mysql-mmm-agent start
    2. Daemon bin: '/usr/sbin/mmm_agentd'
    3. Daemon pid: '/var/run/mmm_agentd.pid'
    4. Starting MMM Agent daemon... Ok

    slave2操作:

    1. [root@slave2 ~]# /etc/init.d/mysql-mmm-agent start
    2. Daemon bin: '/usr/sbin/mmm_agentd'
    3. Daemon pid: '/var/run/mmm_agentd.pid'
    4. Starting MMM Agent daemon... Ok

    2)启动mmm-monitor进程

    monitor主机操作:

    1. [root@monitor ~]# /etc/init.d/mysql-mmm-monitor start
    2. Daemon bin: '/usr/sbin/mmm_mond'
    3. Daemon pid: '/var/run/mmm_mond.pid'
    4. Starting MMM Monitor daemon: Ok

    步骤二:设置集群中服务器为online状态

    控制命令只能在管理端monitor服务器上执行。

    查看当前集群中各服务器状态:

    1. [root@monitor ~]# mmm_control show
    2. master1(192.168.4.10) master/AWAITING_RECOVERY. Roles: 
    3. master2(192.168.4.11) master/AWAITING_RECOVERY. Roles: 
    4. slave1(192.168.4.12) slave/AWAITING_RECOVERY. Roles: 
    5. slave2(192.168.4.13) slave/AWAITING_RECOVERY. Roles: 

    设置4台数据库主机状态为online:

    1. [root@monitor ~]# mmm_control set_online master1
    2. OK: State of 'master1' changed to ONLINE. Now you can wait some time and check its new roles!
    3. [root@monitor ~]# mmm_control set_online master2
    4. OK: State of 'master2' changed to ONLINE. Now you can wait some time and check its new roles!
    5. [root@monitor ~]# mmm_control set_online slave1
    6. OK: State of 'slave1' changed to ONLINE. Now you can wait some time and check its new roles!
    7. [root@monitor ~]# mmm_control set_online slave2
    8. OK: State of 'slave2' changed to ONLINE. Now you can wait some time and check its new roles!
    9. [root@monitor ~]#

    再次查看当前集群中各服务器状态:

    1. [root@monitor ~]# mmm_control show
    2. master1(192.168.4.10) master/ONLINE. Roles: writer(192.168.4.200)
    3. master2(192.168.4.11) master/ONLINE. Roles: 
    4. slave1(192.168.4.12) slave/ONLINE. Roles: reader(192.168.4.201)
    5. slave2(192.168.4.13) slave/ONLINE. Roles: reader(192.168.4.202)
    6. [root@monitor ~]#

    步骤三:测试MySQL-MMM架构

    1)客户机安装MySQL-client软件包

    1. [root@client ~]# tar xvf MySQL-5.6.15-1.el6.x86_64.rpm-bundle.tar
    2. .. ..
    3. [root@client ~]# rpm -ivh MySQL-client-5.6.15-1.el6.x86_64.rpm
    4. .. ..

    2)MySQL-MMM虚拟IP访问测试

    1. [root@client ~]# mysql -h192.168.4.200 -uroot -ppwd123 -e "show databases"
    2. Warning: Using a password on the command line interface can be insecure.
    3. +--------------------+
    4. | Database           |
    5. +--------------------+
    6. | information_schema |
    7. | mysql              |
    8. | performance_schema |
    9. | tarena             |
    10. | test               |
    11. +--------------------+
    12. [root@client ~]#
    13. [root@client ~]# mysql -h192.168.4.200 -uroot -ppwd123 -e "show databases"
    14. Warning: Using a password on the command line interface can be insecure.
    15. +--------------------+
    16. | Database           |
    17. +--------------------+
    18. | information_schema |
    19. | mysql              |
    20. | performance_schema |
    21. | tarena             |
    22. | test               |
    23. +--------------------+
    24. [root@client ~]#
    25. [root@client ~]# mysql -h192.168.4.202 -uroot -ppwd123 -e "show databases"
    26. Warning: Using a password on the command line interface can be insecure.
    27. +--------------------+
    28. | Database           |
    29. +--------------------+
    30. | information_schema |
    31. | mysql              |
    32. | performance_schema |
    33. | tarena             |
    34. | test               |
    35. +--------------------+
    36. [root@client ~]#

    3)主数据库宕机测试

    1. [root@master1 ~]# service mysql stop                    //停止master1上服务
    2. Shutting down MySQL....                                    [确定]
    3. [root@master1 ~]#
    4. [root@monitor ~]# mmm_control show                        //查看集群内服务器状态

    通过输出信息可以看到虚拟IP从master1切换到master2:

    1. master1(192.168.4.10) master/HARD_OFFLINE. Roles: 
    2. master2(192.168.4.11) master/ONLINE. Roles: writer(192.168.4.200)
    3. slave1(192.168.4.12) slave/ONLINE. Roles: reader(192.168.4.201)
    4. slave2(192.168.4.13) slave/ONLINE. Roles: reader(192.168.4.202)
    5. [root@monitor ~]#
    6. [root@client ~]# mysql -h192.168.4.200 -uroot -ppwd123 -e "show databases"    //访问虚拟IP测试
    7. Warning: Using a password on the command line interface can be insecure.
    8. +--------------------+
    9. | Database           |
    10. +--------------------+
    11. | information_schema |
    12. | mysql              |
    13. | performance_schema |
    14. | tarena             |
    15. | test               |
    16. +--------------------+
    17. [root@client ~]#
  • 相关阅读:
    insert 批量插入
    mybatisplus 数据去重
    【新品预告】北京迅为3A5000核心板即将来袭
    弗里德曼迪亚科尼斯规则,偏度,峰度
    第三方支付功能测试点【杭州多测师_王sir】【杭州多测师】
    数据库测试需要注意的点【杭州多测师_王sir】【杭州多测师】
    Python题目:2个list列表相加【杭州多测师】【杭州多测师_王sir】
    蚂蚁金服笔试题:需求文档有什么可以量化的【杭州多测师】【杭州多测师_王sir】
    IDEA: 菜单栏消失的解决办法
    使用arm Neon操作,提高内存拷贝速度
  • 原文地址:https://www.cnblogs.com/qingbai/p/12017775.html
Copyright © 2020-2023  润新知