主节点高可用
MHA是一款开源的MySQL的高可用程序,他为MySQL主从复制架构提供了automating master failover功能。MHA在监控到master节点故障时,会提升其中拥有最新数据的slave节点成为新的master节点。在此期间,MHA会通过其他从节点获取额外信息来避免一致性方面的问题。MHA还提供了master节点切换功能,即按需切换master/slave节点。
MHA服务有两种角色,MHA Manager(管理节点)和HMA Node(数据节点)
- MHA Manager:通常部署在一台独立机器上管理多个master/slave集群,每个master/slave集群乘坐一个application;
- MHA node:运行在每台MySQL服务器上(master/slave/manager),他通过监控解析和清理logs功能的脚本来加快故障转移。
MHA组件说明
Manager节点:
-masterha_check_ssh:MHA依赖的SSH环境检测工具;
-masterha_check_repl:MySQL复制环境检测工具;
-masterha_manager:MHA服务主程序;
-masterha_check_status:MHA运行状态探测工具;
-masterha_master_monitor:MySQL master节点可用性检测工具;
-masterha_master_switch:master节点切换工具;
-masterha_conf_host:添加或删除配置的节点;
-masterha_stop:关闭MHA服务的工具;
Node节点:
-save_binary_logs:保存和复制master的二进制日志;
-apply_diff_relay_logs:识别差异的中继日志事件并用于其他slave;
-fiter_mysqlbinlog:去除不必要的ROLLBACK事件(MHA已不再使用这个工具);
-purge_relay_logs:清除中继日志(不会阻塞SQL线程);
自定义扩展:
-secondary_check_script:通过多条网络路由检测master的可用性;
-master_ip_failover_script:更新appliction使用的masterip;
-shutdown_script:强制关闭master节点;
-report_script:发送报告;
-init_conf_load_script:加载初始配置参数;
-master_ip_online_change_script:更新master节点ip地址
高可用MHA架构部署及测试
节点角色
mysql-master 172.16.252.82
mysql-slave1 172.16.252.92
mysql-slave2 172.16.252.100
manager 172.16.252.67
配置各节点之间可以通过主机名通信
[root@manager ~]# vim /etc/hosts
172.16.252.82 mysql-master
172.16.252.92 mysql-slave1
172.16.252.100 mysql-slave2
172.16.252.67 manager
其他各主机依次配置主机名解析文件
mysql-master
[root@mysql-master ~]# vim /etc/my.cnf.d/server.cnf
[mysqld]
server_id = 1
relay-log = relay-log
log-bin = master-log
skip_name_resolve = ON
innodb_file_per_table = ON
[root@mysql-master ~]# systemctl start mariadb
[root@mysql-master ~]# mysql
授权一个有权限做主从复制的用户
MariaDB [(none)]> GRANT REPLICATION CLIENT,REPLICATION SLAVE ON *.* TO 'repluser'@'172.16.%.%' IDENTIFIED BY 'replpass';
MariaDB [(none)]> FLUSH PRIVILEGES;
查看master节点的状态,记录二进制文件的位置信息,用于从节点的复制起点
MariaDB [(none)]> SHOW MASTER STATUS;
+-------------------+----------+--------------+------------------+
| File | Position | Binlog_Do_DB | Binlog_Ignore_DB |
+-------------------+----------+--------------+------------------+
| master-log.000001 | 501 | | |
+-------------------+----------+--------------+------------------+
mysql-slave1
[root@mysql-slave1 ~]# vim /etc/my.cnf.d/server.cnf
[mysqld]
server_id = 2
relay-log = relay-log
log-bin = master-log
skip_name_resolve = ON
innodb_file_per_table = ON
read_only = ON
relay_log_purge = OFF
[root@mysql-slave1 ~]# systemctl start mariadb
[root@mysql-slave ~]# mysql
MariaDB [mydb]> SHOW GLOBAL VARIABLES LIKE 'read_only'; \\查看数据库为只读
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| read_only | ON |
+---------------+-------+
连接主节点数据库
MariaDB [(none)]> CHANGE MASTER TO MASTER_HOST='192.168.198.139',MASTER_USER='repluser',MASTER_PASSWORD='replpass',MASTER_LOG_FILE='master-log.000001',MASTER_LOG_POS=501;
启动从节点复制进程
MariaDB [(none)]> START SLAVE;
查看SLAVE节点的状态
MariaDB [(none)]> SHOW SLAVE STATUS\G;
连接manager的用户
MariaDB [(none)]> GRANT REPLICATION CLIENT,REPLICATION SLAVE ON *.* TO 'repluser'@'172.16.%.%' IDENTIFIED BY 'replpass';
mysql-slave2
root@mysql-slave2 ~]# vim /etc/my.cnf.d/server.cnf
[mysqld]
server_id = 3
relay-log = relay-log
log-bin = master-log
skip_name_resolve = ON
innodb_file_per_table = ON
read_only = ON
relay_log_purge = OFF
[root@mysql-slave2 ~]# systemctl start mariadb
[root@mysql-slave ~]# mysql
MariaDB [mydb]> SHOW GLOBAL VARIABLES LIKE 'read_only'; \\查看数据库为只读
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| read_only | ON |
+---------------+-------+
连接主节点数据库
MariaDB [(none)]> CHANGE MASTER TO MASTER_HOST='192.168.198.139',MASTER_USER='repluser',MASTER_PASSWORD='replpass',MASTER_LOG_FILE='master-log.000001',MASTER_LOG_POS=501;
启动从节点复制进程
MariaDB [(none)]> START SLAVE;
查看SLAVE节点的状态
MariaDB [(none)]> SHOW SLAVE STATUS\G;
连接manager的用户
MariaDB [(none)]> GRANT REPLICATION CLIENT,REPLICATION SLAVE ON *.* TO 'repluser'@'172.16.%.%' IDENTIFIED BY 'replpass';
mysql-master(master/slave上都需有此用户权限来连接manager)
MariaDB [(none)]> GRANT ALL ON *.* TO 'mha'@'172.16.%.%' IDENTIFIED BY 'mhapass';
MariaDB [(none)]> GRANT REPLICATION CLIENT,REPLICATION SLAVE ON *.* TO 'repluser'@'172.16.%.%' IDENTIFIED BY 'replpass';
MHA安装
github官网下载manager到本地
[root@manager ~]# ls mha4mysql-node-0.56-0.el6.noarch.rpm mha4mysql-manager-0.56-0.el6.noarch.rpm
mha4mysql-manager-0.56-0.el6.noarch.rpm
mha4mysql-node-0.56-0.el6.noarch.rpm
[root@manager ~]# yum install ./mha4mysql-* 安装MHA程序包
复制mha4mysql-node-0.56-0.el6.noarch.rpm程序包到各节点上
[root@manager ~]# scp mha4mysql-node-0.56-0.el6.noarch.rpm mysql-master:/root
[root@manager ~]# scp mha4mysql-node-0.56-0.el6.noarch.rpm mysql-slave1:/root
[root@manager ~]# scp mha4mysql-node-0.56-0.el6.noarch.rpm mysql-slave2:/root
mysql-master、mysql-slave1和mysql-slave2节点依次安装mha4mysql-node-0.56-0.el6.noarch.rpm程序包
[root@mysql-master ~]# yum -y install ./mha4mysql-node-0.56-0.el6.noarch.rpm
[root@mysql-slave1 ~]# yum -y install ./mha4mysql-node-0.56-0.el6.noarch.rpm
[root@mysql-slave2 ~]# yum -y install ./mha4mysql-node-0.56-0.el6.noarch.rpm
配置各节点之间SSH互信的通信环境
MHA
[root@manager ~]# ssh-keygen -t rsa -P '' 生成ssh秘钥文件
[root@manager ~]# ssh-copy-id -i .ssh/id_rsa.pub root@manager
[root@manager ~]# scp -p .ssh/id_rsa .ssh/authorized_leys root@mysql-master:/root/.ssh
[root@manager ~]# scp -p .ssh/id_rsa .ssh/authorized_leys root@mysql-slave1:/root/.ssh
[root@manager ~]# scp -p .ssh/id_rsa .ssh/authorized_leys root@mysql-slave2:/root/.ssh
[root@mysql-master ~]# ssh -o StrictHostKeyChecking=no mysql-slave1 直接连接mysql-slave主机,无询问,无提示
初始化MHA
manager
[root@manager ~]# mkdir /data/masterha
[root@manager ~]# mkdir /data/masterha/app
[root@manager ~]# mkdir /etc/masterha
[root@manager ~]# vim /etc/masterha/app.cnf
[server default]
user=mha
password=mhapass \\node节点连接manager的用户
manager_workdir=/data/masterha/app
manager_log=/data/masterha/app/manager.log
remote_workdir=/data/masterha/app
ssh_user=root
repl_user=repluser
repl_password=replpass
ping_interval=1
[server1]
hostname=172.16.252.82
candidate_master=1
[server2]
hostname=172.16.252.92
candidate_master=1
[server3]
hostname=172.16.252.100
candidate_master=1
测试后端主机的集群
[root@manager ~]# masterha_check_repl --conf=/etc/masterha/app.cnf
测试运行MHA程序
[root@manager ~]# masterha_manager --conf=/etc/masterha/app.cnf
查看MHA集群的状态
[root@manager ~]# masterha_check_status