MySQL高可用架构-MMM安装教程
安装指南:
一、架构以及服务器信息
基本安装包含至少2个数据库服务器和1个监视服务器。本例中使用2个监视服务器和5个数据库服务器(服务器系统为CentOS 7)
用途 | IP | 主机名 | Server-id |
MMM-Monitor01 | 192.168.0.30 | mon01 | - |
MMM-Monitor02 | 192.168.0.31 | mon02 | - |
Master01 | 192.168.0.1 | db1 | 1 |
Master02 | 192.168.0.2 | db2 | 2 |
Slave01 | 192.168.0.11 | db3 | 11 |
Slave02 | 192.168.0.12 | db4 | 12 |
Slave03 | 192.168.0.13 | db5 | 13 |
我使用了以下虚拟IP。它们将由MMM分发给主机。
IP | 角色 | 描述 |
192.168.0.100 | writer | 应用程序应连接到此IP以进行写入查询 |
192.168.0.101 | reader | 应用程序应连接到这两个IP之一以进行读取查询 |
192.168.0.102 | reader |
二、配置服务器上的MySQL
首先我们在所有的数据库服务器上安装MySQL。
配置YUM源:
echo '''[mysql-release-$basearch] name = Percona-Release YUM repository - $basearch baseurl = https://mirror.tuna.tsinghua.edu.cn/mysql/yum/mysql57-community-el7/ gpgcheck=0''' > /etc/yum.repos.d/mysql.repo
安装MySQL:
yum -y install mysql mysql-server
然后我们编辑配置文件/etc/mysql/my.cnf并添加以下行 - 确保为所有主机使用不同的服务器ID:
[client] default-character-set = utf8mb4 [mysql] default-character-set = utf8mb4 [mysqld] basedir=/var/lib/mysql datadir=/var/lib/mysql socket=/var/lib/mysql/mysql.sock pid-file=/var/run/mysqld/mysqld.pid # Disabling symbolic-links is recommended to prevent assorted security risks symbolic-links=0 default-storage-engine = InnoDB ########basic settings######## #服务器ID,不能重复 server-id = 1 character_set_server=utf8mb4 skip_name_resolve = 1 max_connections = 1000 #MySQL读入缓冲区的大小 read_buffer_size = 16M #MySQL的随机读缓冲区大小 read_rnd_buffer_size = 8M #MySQL的顺序读缓冲区大小 sort_buffer_size = 8M ####innodb配置################# innodb_buffer_pool_size=2G innodb_log_file_size=256M innodb_flush_method=O_DIRECT #innodb独享表空间,优点很多,缺点会导致单个表文件过大 innodb_file_per_table = 1 # 从库需要加上只读选项,主库不需要此选项,另外super和replica权限的用户不受只读选项的影响。 #read_only=1 ########log settings######## log_error = /var/log/mysql/error.log #开启慢查询日志 slow_query_log = 1 #超出次设定值的SQL即被记录到慢查询日志 long_query_time = 5 slow_query_log_file = /var/log/mysql/slow.log #表示记录下没有使用索引的查询 log_queries_not_using_indexes = 1 #记录管理语句 log_slow_admin_statements = 1 #开启复制从库复制的慢查询的日志 log_slow_slave_statements = 1 #设置每分钟增长的没有使用索引查询的日志数量 log_throttle_queries_not_using_indexes = 10 expire_logs_days = 90 min_examined_row_limit = 100 log-bin = mysql-bin log-bin-index = mysql-bin.index max-binlog-size = 100M # binlog-do-db = test # binlog-ignore-db = mysql sync-binlog = 1 relay-log = mysql-relay relay-log-index = mysql-relay.index log-slave-updates = 1 # replicate-do-db = test # replicate-ignore-db = mysql #设置为主节点人数 auto_increment_increment = 2 #在每台服务器上设置为唯一的递增数字,小于auto_increment_increment(两台就是1,2) auto_increment_offset = 1
最后启动MySQL:
systemctl enable mysql #此处需要先创建mysql的日志文件夹 mkdir /var/log/mysql && chown -R mysql.mysql /var/log/mysqls systemctl start mysql
查看MySQL的日志,可以看到如下信息:
图中红色框内为mysql生成的临时密码,登陆mysql并修改密码:
mysql -uroot -p<password> > set password for 'root'@'localhost'=password('123456')
其他db服务器按照同样步骤配置。
创建用户:
用途 | 描述 | 权限 |
monitor user | mmm监视器使用它来检查MySQL服务器的运行状况 | REPLICATION CLIENT |
agent user | mmm代理用于更改只读模式,复制主机等 | SUPER, REPLICATION CLIENT, PROCESS |
replication user | 用于复制 | REPLICATION SLAVE |
> GRANT REPLICATION CLIENT ON *.* TO 'mmm_monitor'@'192.168.0.%' IDENTIFIED BY 'monitor_password'; > GRANT SUPER, REPLICATION CLIENT, PROCESS ON *.* TO 'mmm_agent'@'192.168.0.%' IDENTIFIED BY 'agent_password'; > GRANT REPLICATION SLAVE ON *.* TO 'replication'@'192.168.0.%' IDENTIFIED BY 'replication_password';
注意:对于允许用户连接的主机会更加严格:mmm_monitor用于192.168.0.30/31。mmm_agent和replication用于192.168.0.1 - 192.168.0.20。
注意:不要用超过32个字符的replication_password
三、db之间的数据同步配置
配置同步的准备工作
我假设db1包含正确的数据。如果你有一个空数据库,你仍然需要同步我们刚刚创建的帐户。
首先确保在创建备份时没有人在改变数据。
> FLUSH TABLES WITH READ LOCK;
然后获取二进制日志中的当前位置。当我们在db2,db3,db4和db5上设置复制时,将需要这个值。
> SHOW MASTER STATUS; +------------------+----------+--------------+------------------+ | File | Position | Binlog_Do_DB | Binlog_Ignore_DB | +------------------+----------+--------------+------------------+ | mysql-bin.000006 | 567 | | | +------------------+----------+--------------+------------------+ 1 row in set (0.00 sec)
不要关闭这个mysql-shell。如果关闭它,将会删除数据库锁,不能保证数据库没有用户写入。
再打开第二个控制台并键入:
mysqldump -u root -p --all-databases> /tmp/database-backup.sql
现在我们可以删除数据库锁。转到第一个shell:
> UNLOCK TABLES;
将数据库备份复制到db2,db3,db4和db5:
scp /tmp/database-backup.sql <user> @ 192.168.0.2:/tmp scp /tmp/database-backup.sql <user> @ 192.168.0.11:/tmp scp /tmp/database-backup.sql <user> @ 192.168.0.12:/tmp scp /tmp/database-backup.sql <user> @ 192.168.0.13:/tmp
然后将其导入db2,db3,db4和db5:
mysql -u root -p < /tmp/database-backup.sql mysql -u root -p < /tmp/database-backup.sql mysql -u root -p < /tmp/database-backup.sql mysql -u root -p < /tmp/database-backup.sql
然后刷新db2,db3,db4和db5上的权限。我们更改了用户表,而mysql必须重新读取此表。
flush privileges;
flush privileges;
flush privileges;
flush privileges;
两个数据库现在都包含相同的数据。我们现在可以设置复制以保持这种方式。
注意:导入只是从转储文件中添加记录。你应该在导入转储文件之前删除所有数据库。
安装复制
使用以下命令在db2,db3,db4和db5上配置复制:
> CHANGE MASTER TO master_host ='192.168.0.1',master_port = 3306,master_user ='replication', master_password ='replication_password',master_log_file ='<file>',master_log_pos = <position>; > CHANGE MASTER TO master_host ='192.168.0.1',master_port = 3306,master_user ='replication', master_password ='replication_password',master_log_file ='<file>',master_log_pos = <position>; > CHANGE MASTER TO master_host ='192.168.0.1',master_port = 3306,master_user ='replication', master_password ='replication_password',master_log_file ='<file>',master_log_pos = <position>; > CHANGE MASTER TO master_host ='192.168.0.1',master_port = 3306,master_user ='replication', master_password ='replication_password',master_log_file ='<file>',master_log_pos = <position>;
请在db1“show master status”查看<file>和<position>。
在所有4台主机上启动slave-process:
> START SLAVE; > START SLAVE; > START SLAVE; > START SLAVE;
现在检查复制是否在所有主机上正确运行:
> SHOW SLAVE STATUS\G *************************** 1. row *************************** Slave_IO_State: Waiting for master to send event Master_Host: 192.168.0.1 Master_User: replication Master_Port: 3306 Connect_Retry: 60 … > SHOW SLAVE STATUS\G *************************** 1. row *************************** Slave_IO_State: Waiting for master to send event Master_Host: 192.168.0.1 Master_User: replication Master_Port: 3306 Connect_Retry: 60 … > SHOW SLAVE STATUS\G *************************** 1. row *************************** Slave_IO_State: Waiting for master to send event Master_Host: 192.168.0.1 Master_User: replication Master_Port: 3306 Connect_Retry: 60 … > SHOW SLAVE STATUS\G *************************** 1. row *************************** Slave_IO_State: Waiting for master to send event Master_Host: 192.168.0.1 Master_User: replication Master_Port: 3306 Connect_Retry: 60 …
现在我们需要配置db1为db2的从服务器,也就是主主之间的复制
在db2上执行如下命令:
> SHOW MASTER STATUS; +------------------+----------+--------------+------------------+ | File | Position | Binlog_Do_DB | Binlog_Ignore_DB | +------------------+----------+--------------+------------------+ | mysql-bin.000003 | 189 | | | +------------------+----------+--------------+------------------+ 1 row in set (0.00 sec)
现在我们可以使用下面的命令来配置db1的slave进程:
> CHANGE MASTER TO master_host = '192.168.0.2', master_port=3306, master_user='replication', master_password='replication_password', master_log_file='<file>', master_log_pos=<position>;
启动db1上的从属进程:
> START SLAVE;
现在检查复制是否在db1上正确运行:
SHOW SLAVE STATUS\G *************************** 1. row *************************** Slave_IO_State: Waiting for master to send event Master_Host: 192.168.0.2 Master_User: <replication> Master_Port: 3306 Connect_Retry: 60 …
现在应该完成节点之间的复制。通过在db1和db2中插入一些数据来检查它,并检查数据是否会出现在所有其他节点上。
四、安装MMM
创建用户
创建将成为MMM脚本和配置文件所有者的用户。这将提供一种更安全地管理监视器脚本的方法。
useradd --comment “MMM Script owner” --shell / sbin / nologin mmmd
监控主机
yum install mysql-mmm mysql-mmm-agent mysql-mmm-tools mysql-mmm-monitor -y
数据库主机
yum install -y mysql-mmm-agent
配置MMM
所有通用配置选项都分组在一个名为/etc/mysql-mmm/mmm_common.conf的单独文件中。该文件在系统中的所有主机上都是相同的:
active_master_role writer <host default> cluster_interface eno16777984 pid_path /run/mysql-mmm-agent.pid bin_path /usr/libexec/mysql-mmm/ replication_user replication replication_password replication_password agent_user mmm_agent agent_password agent_password </host> <host db1> ip 192.168.0.1 mode master peer db2 </host> <host db2> ip 192.168.0.2 mode master peer db1 </host> <host db3> ip 192.168.0.11 mode slave </host> <host db4> ip 192.168.0.12 mode slave </host> <host db5> ip 192.168.0.13 mode slave </host> <role writer> hosts db1, db2 ips 192.168.0.100 mode exclusive </role> <role reader> hosts db3, db4, db5 ips 192.168.0.101,192.168.0.102 mode balanced </role>
不要忘记将此文件复制到所有其他主机(包括监控主机)。
在数据库主机上,我们需要编辑/etc/mysql-mmm/mmm_agent.conf。在其他主机上相应地更改“db1”:
include mmm_common.conf
this db1
在监视器主机上(两台,同样的配置,ip需要修改为主机对应地址),我们需要编辑/etc/mysql-mmm/mmm_mon.conf:
include mmm_common.conf <monitor> ip 192.168.0.30 pid_path /run/mysql-mmm-monitor.pid bin_path /usr/libexec/mysql-mmm status_path /var/lib/mysql-mmm/mmm_mond.status ping_ips 192.168.0.254, 192.168.0.1, 192.168.0.2, 192.168.0.11, 192.168.0.12, 192.168.0.13 auto_set_online 10 # 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 monitor_password </host> debug 0
ping_ips是一些被ping的ips,用于确定监视器的网络连接是否正常。我使用了交换机地址(192.168.0.254)和四个数据库服务器。
启动MMM
启动agent
systemctl enable mysql-mmm-agent
systemctl start mysql-mmm-agent
启动monitor
systemctl enable mysql-mmm-monitor
systemctl start mysql-mmm-monitor
等待几秒钟让mmmd_mon启动。几秒钟后,您可以使用mmm_control检查集群的状态:
$ mmm_control show db1(192.168.0.1) master/AWAITING_RECOVERY. Roles: db2(192.168.0.2) master/AWAITING_RECOVERY. Roles: db3(192.168.0.11) slave/AWAITING_RECOVERY. Roles: db4(192.168.0.12) slave/AWAITING_RECOVERY. Roles: db4(192.168.0.13) slave/AWAITING_RECOVERY. Roles:
因为它是第一次启动,监视器不知道我们的主机,所以它将所有主机设置为AWAITING_RECOVERY状态并记录一条警告消息:
$ tail /var/log/mysql-mmm/mmm_mon.warn … 0000/10/28 23:15:28 WARN Detected new host 'db1': Setting its initial state to 'AWAITING_RECOVERY'. Use 'mmm_control set_online db1' to switch it online. 0000/10/28 23:15:28 WARN Detected new host 'db2': Setting its initial state to 'AWAITING_RECOVERY'. Use 'mmm_control set_online db2' to switch it online. 0000/10/28 23:15:28 WARN Detected new host 'db3': Setting its initial state to 'AWAITING_RECOVERY'. Use 'mmm_control set_online db3' to switch it online. 0000/10/28 23:15:28 WARN Detected new host 'db4': Setting its initial state to 'AWAITING_RECOVERY'. Use 'mmm_control set_online db4' to switch it online. 0000/10/28 23:15:28 WARN Detected new host 'db5': Setting its initial state to 'AWAITING_RECOVERY'. Use 'mmm_control set_online db5' to switch it online.
由于我们配置文件中setonline的时间设置为10s,等待10s或者手动online:
$ mmm_control set_online db1 OK: State of 'db1' changed to ONLINE. Now you can wait some time and check its new roles! $ mmm_control set_online db2 OK: State of 'db2' changed to ONLINE. Now you can wait some time and check its new roles! $ mmm_control set_online db3 OK: State of 'db3' changed to ONLINE. Now you can wait some time and check its new roles! $ mmm_control set_online db4 OK: State of 'db4' changed to ONLINE. Now you can wait some time and check its new roles! $ mmm_control set_online db5 OK: State of 'db5' changed to ONLINE. Now you can wait some time and check its new roles!
OK,现在整个集群已经配置完毕,但是还是存在一个问题,mmm的monitor节点目前只有一个,那么就存在单点故障,如何处理这个问题那,这就是我们第二个monitor节点的用途了。
方案罗列如下:
1、使用keepalived进行monitor高可用的保障。
2、现在使用的方案,自己编写脚本监控节点状态,脚本如下:
cat /opt/scripts/mysql-mmm-backup.sh #!/bin/bash remote_port_status=`nc -z -w 5 192.168.0.1 9988;echo $?` local_mmm_svc_status=`systemctl status mysql-mmm-monitor.service | grep Active | awk -F '[(,)]' '{print $2}'` if [ $remote_port_status -eq 0 ];then echo "远端服务正常!" if [ $local_mmm_svc_status = 'running' ];then echo "本地服务需要关闭!" systemctl stop mysql-mmm-monitor.service echo "远端服务器Mysql-mmm-monitor服务恢复,关闭备用服务!" | mail -s 'Mysql-mmm' admin@local.com fi else echo "远端服务异常!" if [ $local_mmm_svc_status = 'dead' ];then echo "本地服务需要开启!" systemctl start mysql-mmm-monitor.service echo "远端服务器Mysql-mmm-monitor服务异常,启动备用服务!" | mail -s 'Mysql-mmm' admin@local.com fi fi #配置cron任务 vim /etc/crontab */1 * * * * root for i in {1..59};do /opt/scripts/mysql-mmm-backup.sh;sleep 1; done >> /opt/scripts/mysql-mmm-backup.log
注:MMM安装借鉴文章http://mysql-mmm.org/mmm2:guide
本文来自博客园,作者:sunnydoy,转载请注明原文链接:https://www.cnblogs.com/sunnydou/p/28f747c35471965e0610eda5efa06a15.html