1.防火墙
firewall-cmd --add-port=3306/tcp --permanent firewall-cmd --reload
2.SELINUX
sed -i 's/SELINUX=enforcing/SELINUX=disabled/g' /etc/selinux/config setenforce 0
3.软件包需求
yum install -y perl-Class-Load.noarch perl-DBD-MySQL #以下需下载，地址： https://pkgs.org/ yum install -y perl-Log-Dispatch-2.41-1.of.el7.noarch.rpm yum install -y perl-Config-Tiny-2.14-7.el7.noarch.rpm yum install -y perl-Parallel-ForkManager-1.06-1.of.el7.noarch.rpm #MHA包需下载，地址 https://github.com/yoshinorim mha4mysql-manager-0.58-0.el7.centos.noarch.rpm #建议安装在独立服务器，也可选择任一从库安装 mha4mysql-node-0.58-0.el7.centos.noarch.rpm #安装在所有节点
4.密码认证插件需要用 mysql_native_password
[mysqld] default_authentication_plugin=mysql_native_password
alter user rep identified with 'mysql_native_password' by 'rep'; alter user root identified with 'mysql_native_password' by 'root ';
5. 检查配置是否成功
建议所有从库都置为只读模式 read_only=1,但这实际并不影响MHA搭建。
[root@mysql3 soft]# masterha_check_repl --conf=/etc/masterha_default.cnf Mon Jun 10 05:38:47 2019 - [info] Reading default configuration from /etc/masterha_default.cnf.. Mon Jun 10 05:38:47 2019 - [info] Reading application default configuration from /etc/masterha_default.cnf.. Mon Jun 10 05:38:47 2019 - [info] Reading server configuration from /etc/masterha_default.cnf.. Mon Jun 10 05:38:47 2019 - [info] MHA::MasterMonitor version 0.58. Mon Jun 10 05:38:47 2019 - [info] GTID failover mode = 1 Mon Jun 10 05:38:47 2019 - [info] Dead Servers: Mon Jun 10 05:38:47 2019 - [info] Alive Servers: Mon Jun 10 05:38:47 2019 - [info] mysql1(192.168.20.81:3306) Mon Jun 10 05:38:47 2019 - [info] mysql2(192.168.20.82:3306) Mon Jun 10 05:38:47 2019 - [info] mysql3(192.168.20.83:3306) Mon Jun 10 05:38:47 2019 - [info] mysql4(192.168.20.84:3306) Mon Jun 10 05:38:47 2019 - [info] Alive Slaves: Mon Jun 10 05:38:47 2019 - [info] mysql2(192.168.20.82:3306) Version=8.0.16 (oldest major version between slaves) log-bin:enabled Mon Jun 10 05:38:47 2019 - [info] GTID ON Mon Jun 10 05:38:47 2019 - [info] Replicating from 192.168.20.81(192.168.20.81:3306) Mon Jun 10 05:38:47 2019 - [info] Primary candidate for the new Master (candidate_master is set) Mon Jun 10 05:38:47 2019 - [info] mysql3(192.168.20.83:3306) Version=8.0.16 (oldest major version between slaves) log-bin:enabled Mon Jun 10 05:38:47 2019 - [info] GTID ON Mon Jun 10 05:38:47 2019 - [info] Replicating from 192.168.20.81(192.168.20.81:3306) Mon Jun 10 05:38:47 2019 - [info] Not candidate for the new Master (no_master is set) Mon Jun 10 05:38:47 2019 - [info] mysql4(192.168.20.84:3306) Version=8.0.16 (oldest major version between slaves) log-bin:enabled Mon Jun 10 05:38:47 2019 - [info] GTID ON Mon Jun 10 05:38:47 2019 - [info] Replicating from 192.168.20.81(192.168.20.81:3306) Mon Jun 10 05:38:47 2019 - [info] Not candidate for the new Master (no_master is set) Mon Jun 10 05:38:47 2019 - [info] Current Alive Master: mysql1(192.168.20.81:3306) Mon Jun 10 05:38:47 2019 - [info] Checking slave configurations.. Mon Jun 10 05:38:47 2019 - [info] read_only=1 is not set on slave mysql2(192.168.20.82:3306). Mon Jun 10 05:38:47 2019 - [info] Checking replication filtering settings.. Mon Jun 10 05:38:47 2019 - [info] binlog_do_db= , binlog_ignore_db= Mon Jun 10 05:38:47 2019 - [info] Replication filtering check ok. Mon Jun 10 05:38:47 2019 - [info] GTID (with auto-pos) is supported. Skipping all SSH and Node package checking. Mon Jun 10 05:38:47 2019 - [info] Checking SSH publickey authentication settings on the current master.. Mon Jun 10 05:38:47 2019 - [info] HealthCheck: SSH to mysql1 is reachable. Mon Jun 10 05:38:47 2019 - [info] mysql1(192.168.20.81:3306) (current master) +--mysql2(192.168.20.82:3306) +--mysql3(192.168.20.83:3306) +--mysql4(192.168.20.84:3306) Mon Jun 10 05:38:47 2019 - [info] Checking replication health on mysql2.. Mon Jun 10 05:38:47 2019 - [info] ok. Mon Jun 10 05:38:47 2019 - [info] Checking replication health on mysql3.. Mon Jun 10 05:38:47 2019 - [info] ok. Mon Jun 10 05:38:47 2019 - [info] Checking replication health on mysql4.. Mon Jun 10 05:38:47 2019 - [info] ok. Mon Jun 10 05:38:47 2019 - [warning] master_ip_failover_script is not defined. Mon Jun 10 05:38:47 2019 - [warning] shutdown_script is not defined. Mon Jun 10 05:38:47 2019 - [info] Got exit code 0 (Not master dead). MySQL Replication Health is OK.
6.以上我用的GTID 复制模式
[mysqld] enforce_gtid_consistency=ON gtid_mode=on
--从库执行如下命令，启动复制
change master to master_host='192.168.20.81', master_port=3306, master_auto_position=1; start slave user='rep' password='rep';
7.MHA 相关命令
masterha_check_ssh --conf=/etc/masterha_default.cnf masterha_check_repl --conf=/etc/masterha_default.cnf masterha_master_switch #可用来执行 fail over/online master switch # For master failover masterha_master_switch --master_state=dead --global_conf=/etc/masterha_default.cnf --conf=/usr/local/masterha/conf/app1.cnf --dead_master_host=host1 # For online master switch masterha_master_switch --master_state=alive --global_conf=/etc/masterha_default.cnf --conf=/usr/local/masterha/conf/app1.cnf masterha_master_monitor --global_conf=/etc/masterha_default.cnf --conf=/usr/local/masterha/app1.cnf masterha_secondary_check -s secondary_host1 -s secondary_host2 .. --user=ssh_username --master_host=host --master_ip=ip --master_port=port masterha_manager --global_conf=/etc/masterha_default.cnf --conf=/usr/local/masterha/conf/app1.cnf
# 以上的--global_conf=/etc/masterha_default.cnf 参数均可省略
8.MHA 在线切换测试
[root@mysql3 soft]# masterha_master_switch --master_state=alive --conf=/etc/masterha_default.cnf Mon Jun 10 06:33:03 2019 - [info] MHA::MasterRotate version 0.58. Mon Jun 10 06:33:03 2019 - [info] Starting online master switch.. Mon Jun 10 06:33:03 2019 - [info] Mon Jun 10 06:33:03 2019 - [info] * Phase 1: Configuration Check Phase.. Mon Jun 10 06:33:03 2019 - [info] Mon Jun 10 06:33:03 2019 - [info] Reading default configuration from /etc/masterha_default.cnf.. Mon Jun 10 06:33:03 2019 - [info] Reading application default configuration from /etc/masterha_default.cnf.. Mon Jun 10 06:33:03 2019 - [info] Reading server configuration from /etc/masterha_default.cnf.. Mon Jun 10 06:33:03 2019 - [info] GTID failover mode = 1 Mon Jun 10 06:33:03 2019 - [info] Current Alive Master: mysql1(192.168.20.81:3306) Mon Jun 10 06:33:03 2019 - [info] Alive Slaves: Mon Jun 10 06:33:03 2019 - [info] mysql2(192.168.20.82:3306) Version=8.0.16 (oldest major version between slaves) log-bin:enabled Mon Jun 10 06:33:03 2019 - [info] GTID ON Mon Jun 10 06:33:03 2019 - [info] Replicating from 192.168.20.81(192.168.20.81:3306) Mon Jun 10 06:33:03 2019 - [info] Primary candidate for the new Master (candidate_master is set) Mon Jun 10 06:33:03 2019 - [info] mysql3(192.168.20.83:3306) Version=8.0.16 (oldest major version between slaves) log-bin:enabled Mon Jun 10 06:33:03 2019 - [info] GTID ON Mon Jun 10 06:33:03 2019 - [info] Replicating from 192.168.20.81(192.168.20.81:3306) Mon Jun 10 06:33:03 2019 - [info] Not candidate for the new Master (no_master is set) Mon Jun 10 06:33:03 2019 - [info] mysql4(192.168.20.84:3306) Version=8.0.16 (oldest major version between slaves) log-bin:enabled Mon Jun 10 06:33:03 2019 - [info] GTID ON Mon Jun 10 06:33:03 2019 - [info] Replicating from 192.168.20.81(192.168.20.81:3306) Mon Jun 10 06:33:03 2019 - [info] Not candidate for the new Master (no_master is set) It is better to execute FLUSH NO_WRITE_TO_BINLOG TABLES on the master before switching. Is it ok to execute on mysql1(192.168.20.81:3306)? (YES/no): YES Mon Jun 10 06:34:00 2019 - [info] Executing FLUSH NO_WRITE_TO_BINLOG TABLES. This may take long time.. Mon Jun 10 06:34:00 2019 - [info] ok. Mon Jun 10 06:34:00 2019 - [info] Checking MHA is not monitoring or doing failover.. Mon Jun 10 06:34:00 2019 - [info] Checking replication health on mysql2.. Mon Jun 10 06:34:00 2019 - [info] ok. Mon Jun 10 06:34:00 2019 - [info] Checking replication health on mysql3.. Mon Jun 10 06:34:00 2019 - [info] ok. Mon Jun 10 06:34:00 2019 - [info] Checking replication health on mysql4.. Mon Jun 10 06:34:00 2019 - [info] ok. Mon Jun 10 06:34:00 2019 - [info] Searching new master from slaves.. Mon Jun 10 06:34:00 2019 - [info] Candidate masters from the configuration file: Mon Jun 10 06:34:00 2019 - [info] mysql1(192.168.20.81:3306) Version=8.0.16 log-bin:enabled Mon Jun 10 06:34:00 2019 - [info] GTID ON Mon Jun 10 06:34:00 2019 - [info] mysql2(192.168.20.82:3306) Version=8.0.16 (oldest major version between slaves) log-bin:enabled Mon Jun 10 06:34:00 2019 - [info] GTID ON Mon Jun 10 06:34:00 2019 - [info] Replicating from 192.168.20.81(192.168.20.81:3306) Mon Jun 10 06:34:00 2019 - [info] Primary candidate for the new Master (candidate_master is set) Mon Jun 10 06:34:00 2019 - [info] Non-candidate masters: Mon Jun 10 06:34:00 2019 - [info] mysql3(192.168.20.83:3306) Version=8.0.16 (oldest major version between slaves) log-bin:enabled Mon Jun 10 06:34:00 2019 - [info] GTID ON Mon Jun 10 06:34:00 2019 - [info] Replicating from 192.168.20.81(192.168.20.81:3306) Mon Jun 10 06:34:00 2019 - [info] Not candidate for the new Master (no_master is set) Mon Jun 10 06:34:00 2019 - [info] mysql4(192.168.20.84:3306) Version=8.0.16 (oldest major version between slaves) log-bin:enabled Mon Jun 10 06:34:00 2019 - [info] GTID ON Mon Jun 10 06:34:00 2019 - [info] Replicating from 192.168.20.81(192.168.20.81:3306) Mon Jun 10 06:34:00 2019 - [info] Not candidate for the new Master (no_master is set) Mon Jun 10 06:34:00 2019 - [info] Searching from candidate_master slaves which have received the latest relay log events.. Mon Jun 10 06:34:00 2019 - [info] From: mysql1(192.168.20.81:3306) (current master) +--mysql2(192.168.20.82:3306) +--mysql3(192.168.20.83:3306) +--mysql4(192.168.20.84:3306) To: mysql2(192.168.20.82:3306) (new master) +--mysql3(192.168.20.83:3306) +--mysql4(192.168.20.84:3306) Starting master switch from mysql1(192.168.20.81:3306) to mysql2(192.168.20.82:3306)? (yes/NO): yes Mon Jun 10 06:35:02 2019 - [info] Checking whether mysql2(192.168.20.82:3306) is ok for the new master.. Mon Jun 10 06:35:02 2019 - [info] ok. Mon Jun 10 06:35:02 2019 - [info] ** Phase 1: Configuration Check Phase completed. Mon Jun 10 06:35:02 2019 - [info] Mon Jun 10 06:35:02 2019 - [info] * Phase 2: Rejecting updates Phase.. Mon Jun 10 06:35:02 2019 - [info] FLUSH NO_WRITE_TO_BINLOG TABLES is not defined. If you do not disable writes on the current master manually, applications keep writing on the current master. Is it ok to proceed? (yes/NO): yes Mon Jun 10 06:35:53 2019 - [info] Locking all tables on the orig master to reject updates from everybody (including root): Mon Jun 10 06:35:53 2019 - [info] Executing FLUSH TABLES WITH READ LOCK.. Mon Jun 10 06:35:53 2019 - [info] ok. Mon Jun 10 06:35:53 2019 - [info] Orig master binlog:pos is binlog.000017:1285. Mon Jun 10 06:35:53 2019 - [info] Waiting to execute all relay logs on mysql2(192.168.20.82:3306).. Mon Jun 10 06:35:53 2019 - [info] master_pos_wait(binlog.000017:1285) completed on mysql2(192.168.20.82:3306). Executed 0 events. Mon Jun 10 06:35:53 2019 - [info] done. Mon Jun 10 06:35:54 2019 - [info] Getting new master's binlog name and position.. Mon Jun 10 06:35:54 2019 - [info] binlog.000015:235 Mon Jun 10 06:35:54 2019 - [info] All other slaves should start replication from here. Statement should be: CHANGE MASTER TO MASTER_HOST='mysql2 or 192.168.20.82', MASTER_PORT=3306, MASTER_AUTO_POSITION=1, MASTER_USER='rep', MASTER_PASSWORD='xxx'; Mon Jun 10 06:35:54 2019 - [info] Setting read_only=0 on mysql2(192.168.20.82:3306).. Mon Jun 10 06:35:54 2019 - [info] ok. Mon Jun 10 06:35:54 2019 - [info] Mon Jun 10 06:35:54 2019 - [info] * Switching slaves in parallel.. Mon Jun 10 06:35:54 2019 - [info] Mon Jun 10 06:35:54 2019 - [info] -- Slave switch on host mysql3(192.168.20.83:3306) started, pid: 9450 Mon Jun 10 06:35:54 2019 - [info] Mon Jun 10 06:35:54 2019 - [info] -- Slave switch on host mysql4(192.168.20.84:3306) started, pid: 9451 Mon Jun 10 06:35:54 2019 - [info] Mon Jun 10 06:35:54 2019 - [info] Log messages from mysql3 ... Mon Jun 10 06:35:54 2019 - [info] Mon Jun 10 06:35:54 2019 - [info] Waiting to execute all relay logs on mysql3(192.168.20.83:3306).. Mon Jun 10 06:35:54 2019 - [info] master_pos_wait(binlog.000017:1285) completed on mysql3(192.168.20.83:3306). Executed 0 events. Mon Jun 10 06:35:54 2019 - [info] done. Mon Jun 10 06:35:54 2019 - [info] Resetting slave mysql3(192.168.20.83:3306) and starting replication from the new master mysql2(192.168.20.82:3306).. Mon Jun 10 06:35:54 2019 - [info] Executed CHANGE MASTER. Mon Jun 10 06:35:54 2019 - [info] Slave started. Mon Jun 10 06:35:54 2019 - [info] End of log messages from mysql3 ... Mon Jun 10 06:35:54 2019 - [info] Mon Jun 10 06:35:54 2019 - [info] -- Slave switch on host mysql3(192.168.20.83:3306) succeeded. Mon Jun 10 06:35:54 2019 - [info] Log messages from mysql4 ... Mon Jun 10 06:35:54 2019 - [info] Mon Jun 10 06:35:54 2019 - [info] Waiting to execute all relay logs on mysql4(192.168.20.84:3306).. Mon Jun 10 06:35:54 2019 - [info] master_pos_wait(binlog.000017:1285) completed on mysql4(192.168.20.84:3306). Executed 0 events. Mon Jun 10 06:35:54 2019 - [info] done. Mon Jun 10 06:35:54 2019 - [info] Resetting slave mysql4(192.168.20.84:3306) and starting replication from the new master mysql2(192.168.20.82:3306).. Mon Jun 10 06:35:54 2019 - [info] Executed CHANGE MASTER. Mon Jun 10 06:35:54 2019 - [info] Slave started. Mon Jun 10 06:35:54 2019 - [info] End of log messages from mysql4 ... Mon Jun 10 06:35:54 2019 - [info] Mon Jun 10 06:35:54 2019 - [info] -- Slave switch on host mysql4(192.168.20.84:3306) succeeded. Mon Jun 10 06:35:54 2019 - [info] Unlocking all tables on the orig master: Mon Jun 10 06:35:54 2019 - [info] Executing UNLOCK TABLES.. Mon Jun 10 06:35:54 2019 - [info] ok. Mon Jun 10 06:35:54 2019 - [info] All new slave servers switched successfully. Mon Jun 10 06:35:54 2019 - [info] Mon Jun 10 06:35:54 2019 - [info] * Phase 5: New master cleanup phase.. Mon Jun 10 06:35:54 2019 - [info] Mon Jun 10 06:35:54 2019 - [info] mysql2: Resetting slave info succeeded. Mon Jun 10 06:35:54 2019 - [info] Switching master to mysql2(192.168.20.82:3306) completed successfully.
执行过程中提供了几次YES输入，还有几个重要的语句：
检查阶段： master 执行 FLUSH NO_WRITE_TO_BINLOG TABLES 搜寻candidate_master=1 的slave 节点，选择其为新master 执行阶段： FLUSH NO_WRITE_TO_BINLOG TABLES is not defined. If you do not disable writes on the current master manually, applications keep writing on the current master. Is it ok to proceed? (yes/NO): yes Locking all tables on the orig master to reject updates from everybody (including root): Executing FLUSH TABLES WITH READ LOCK.. All other slaves should start replication from here. Statement should be: CHANGE MASTER TO MASTER_HOST='mysql2 or 192.168.20.82', MASTER_PORT=3306, MASTER_AUTO_POSITION=1, MASTER_USER='rep', MASTER_PASSWORD='xxx'; Setting read_only=0 on mysql2(192.168.20.82:3306).. Unlocking all tables on the orig master: Executing UNLOCK TABLES..
9.在线切换注意点
1. 会从 candidate_master=1的slave节点中选择新master
2.被选择的新master 修改为read_only=0
3.原来的master 修改为read_only=1
4. 切换后，原master 不会转为新master 的 slave。
5.配置文件不会发生改变，所以重启需注意。
10.将原master 改为新master 的slave
--将原master 修改为新master 的从库 change master to master_host='192.168.20.82',master_port=3306,master_auto_position=1; start slave user='rep' password='rep'; show slave status G
11. 切换回原来的master
# masterha_master_switch --master_state=alive --conf=/etc/masterha_default.cnf 同样的需要把node2 改为slave change master to master_host='192.168.20.81',master_port=3306,master_auto_position=1; start slave user='rep' password='rep'; show slave status G