MySQL高可用架構之MHA

時間 2020-03-31

標籤 mysql 可用架構 mha 欄目 MySQL 简体版

原文原文鏈接

能夠看這篇文章的理論部分，寫的不錯：http://www.educity.cn/wenda/399310.html

操做可參考：http://www.178linux.com/15698html

一、關於MHA

MHA(Master HA)是一款開源的MySQL的高可用程序，它爲MySQL主從複製架構提供了automating master failover功能。MHA在監控到master節點故障時，會提高其中擁有的最新數據的slave節點成爲新的master節點，在此期間，MHA會經過其它從節點獲取額外信息來避免一致性方面的問題。MHA還提供了master節點的在線切換功能，即按需切換master/slave節點。node

MHA服務有兩種角色，MHA Manager(管理節點)和MHA Node(數據節點)：mysql

MHA Manager：一般單獨部署在一臺獨立機器上管理多個master/slave集羣，每一個master/slave集羣稱爲一個application；

MHA node：運行在每臺MySQL服務器上，它經過監控具有解析和清理log功能的腳原本加快故障轉移

二、MHA組件說明

Manager節點：linux

-masterha_check_ssh：MHA依賴的SSH環境檢測工具；
-masterha_check_repl:MySQL複製環境檢測工具；
-masterha_manager:MHA服務主程序；
-masterha_check_status:MHA運行狀態探測工具；
-masterha_master_monitor：MySQL master節點可用性檢測工具；
-masterha_master_switch:master節點切換工具；
-masterha_conf_host:添加或刪除配置的節點；
-masterha_stop：關閉MHA服務的工具；

Node節點：sql

-save_binary_logs：保存和複製master的二進制日誌；
-apply_diff_relay_logs:識別差別的中繼日誌事件並用於其餘slave；
-fiter_mysqlbinlog：去除沒必要要的ROLLBACK事件(MHA已再也不使用這個工具)；
-purge_relay_logs：清除中繼日誌(不會阻塞SQL線程)；

自定義擴展：vim

-secondary_check_script：經過多條網絡路由檢測master的可用性；
-master_ip_failover_script:更新appliction使用的masterip；
-shutdown_script:強制關閉master節點；
-report_script:發送報告；
-init_conf_load_script：加載初始配置參數；
-master_ip_online_change_script:更新master節點ip地址

三、部署及測試

實驗拓撲：服務器

node1:192.168.150.137   MHA manager
node2:192.168.150.138   MHA node    mariadb master
node3:192.168.150.139   MHA node    mariadb slave candidate
node4:192.168.150.140   MHA node    mariadb slave

配置過程：網絡

一、修改每臺服務器的hosts文件
/etc/hosts
127.0.0.1   localhost localhost.localdomain localhost4 localhost4.localdomain4
::1         localhost localhost.localdomain localhost6 localhost6.localdomain6

192.168.150.137 node1.com node1
192.168.150.138 node2.com node2
192.168.150.139 node3.com node3
192.168.150.140 node4.com node4

二、node2-node4進行mariadb的yum安裝
yum -y install mariadb-server

三、配置ssh互信通訊環境
[root@node1 ~]# ssh-keygen -t rsa -P ''
Generating public/private rsa key pair.
Enter file in which to save the key (/root/.ssh/id_rsa): 
Created directory '/root/.ssh'.
Your identification has been saved in /root/.ssh/id_rsa.
Your public key has been saved in /root/.ssh/id_rsa.pub.
The key fingerprint is:
a2:f2:10:28:cd:ea:7b:d8:f4:95:15:6e:73:a6:9d:4e root@node1.com
The key's randomart p_w_picpath is:
+--[ RSA 2048]----+
|                 |
|         .       |
|        . .      |
| +       = o     |
|o +   . S * .    |
|.. o . + . E     |
|. * o .   o      |
|.. * .     .     |
| oo .            |
+-----------------+
[root@node1 ~]# ls .ssh/id_rsa
id_rsa      id_rsa.pub  
[root@node1 ~]# cat .ssh/id_rsa.pub > .ssh/authorized_keys
[root@node1 ~]# ssh node1
The authenticity of host 'node1 (192.168.150.137)' can't be established.
ECDSA key fingerprint is 2a:e3:03:52:8c:84:02:59:a2:26:a3:b2:f6:74:6c:3c.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added 'node1,192.168.150.137' (ECDSA) to the list of known hosts.
Last login: Wed Mar 29 15:06:52 2017 from 192.168.150.1
[root@node1 ~]# ll .ssh/authorized_keys 
-rw-r--r-- 1 root root 396 3月  29 15:35 .ssh/authorized_keys
[root@node1 ~]# chmod go= .ssh/authorized_keys 
[root@node1 ~]# scp -p .ssh/id_rsa .ssh/authorized_keys node2:/root/.ssh
root@node2's password: 
id_rsa                                                  100% 1675     1.6KB/s   00:00    
authorized_keys                                         100%  396     0.4KB/s   00:00    
[root@node1 ~]# scp -p .ssh/id_rsa .ssh/authorized_keys node3:/root/.ssh
The authenticity of host 'node3 (192.168.150.139)' can't be established.
ECDSA key fingerprint is 2a:e3:03:52:8c:84:02:59:a2:26:a3:b2:f6:74:6c:3c.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added 'node3,192.168.150.139' (ECDSA) to the list of known hosts.
root@node3's password: 
id_rsa                                                  100% 1675     1.6KB/s   00:00    
authorized_keys                                         100%  396     0.4KB/s   00:00    
[root@node1 ~]# scp -p .ssh/id_rsa .ssh/authorized_keys node4:/root/.ssh
root@node4's password: 
Permission denied, please try again.
root@node4's password: 
id_rsa                                                  100% 1675     1.6KB/s   00:00    
authorized_keys                                         100%  396     0.4KB/s   00:00    
[root@node1 ~]# ssh node2
Last login: Wed Mar 29 15:07:05 2017 from 192.168.150.1
[root@node2 ~]# exit
登出
Connection to node2 closed.
[root@node1 ~]# ssh node3
Last login: Wed Mar 29 15:07:18 2017 from 192.168.150.1
[root@node3 ~]# exit
登出
Connection to node3 closed.
[root@node1 ~]# ssh node3
Last login: Wed Mar 29 15:40:05 2017 from node1.com
[root@node3 ~]# exit
登出
Connection to node3 closed.
[root@node1 ~]# ssh node4
Last failed login: Wed Mar 29 15:39:53 CST 2017 from node1.com on ssh:notty
There was 1 failed login attempt since the last successful login.
Last login: Wed Mar 29 15:39:30 2017 from node1.com
[root@node4 ~]# exit
登出
Connection to node4 closed.
[root@node1 ~]# ssh node4
Last login: Wed Mar 29 15:40:13 2017 from node1.com
[root@node4 ~]# exit
登出
Connection to node4 closed.


四、修改mysql參數
master：
[mysqld]
innodb_file_per_table = 1
skip_name_resolve = 1
log-bin = master-bin
relay-log = relay-bin
server_id = 1

slave：
[mysqld]
innode_file_per_table = 1
skip_name_resolve = 1
log-bin = master-bin
relay-log = relay-bin
server_id = 2
read_only = 1
relay_log_purge = 0

五、主庫開啓並建立受權帳號
master:
MariaDB [(none)]> SHOW MASTER STATUS;
+-------------------+----------+--------------+------------------+
| File              | Position | Binlog_Do_DB | Binlog_Ignore_DB |
+-------------------+----------+--------------+------------------+
| master-bin.000003 |      245 |              |                  |
+-------------------+----------+--------------+------------------+
1 row in set (0.00 sec)
MariaDB [(none)]> GRANT REPLICATION SLAVE,REPLICATION CLIENT ON *.* TO 'repluser'@'192.168.%.%' IDENTIFIED BY 'replpass';

MariaDB [(none)]> GRANT ALL ON *.* TO 'mhauser'@'192.168.%.%' IDENTIFIED BY 'mhapass';  #此爲mha的管理帳號

MariaDB [(none)]> SHOW GLOBAL VARIABLES LIKE 'read_only';   #此時主庫是可寫可讀
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| read_only     | OFF   |
+---------------+-------+
1 row in set (0.00 sec)


六、從庫進行主從功能開啓 node三、node4操做相同
[root@node3 ~]# mysql
Welcome to the MariaDB monitor.  Commands end with ; or \g.
Your MariaDB connection id is 2
Server version: 5.5.52-MariaDB MariaDB Server

Copyright (c) 2000, 2016, Oracle, MariaDB Corporation Ab and others.

Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.

MariaDB [(none)]> CHANGE MASTER TO MASTER_HOST='192.168.150.138',MASTER_USER='repluser',MSTER_PASSWORD='replpass',MASTER_LOG_FILE='master-bin.000003',MASTER_LOG_POS=245;
Query OK, 0 rows affected (0.01 sec)

MariaDB [(none)]> SHOW SLAVE STATUS\G
*************************** 1. row ***************************
               Slave_IO_State: 
                  Master_Host: 192.168.150.138
                  Master_User: repluser
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: master-bin.000003
          Read_Master_Log_Pos: 245
               Relay_Log_File: relay-bin.000001
                Relay_Log_Pos: 4
        Relay_Master_Log_File: master-bin.000003
             Slave_IO_Running: No
            Slave_SQL_Running: No
              Replicate_Do_DB: 
          Replicate_Ignore_DB: 
           Replicate_Do_Table: 
       Replicate_Ignore_Table: 
      Replicate_Wild_Do_Table: 
  Replicate_Wild_Ignore_Table: 
                   Last_Errno: 0
                   Last_Error: 
                 Skip_Counter: 0
          Exec_Master_Log_Pos: 245
              Relay_Log_Space: 245
              Until_Condition: None
               Until_Log_File: 
                Until_Log_Pos: 0
           Master_SSL_Allowed: No
           Master_SSL_CA_File: 
           Master_SSL_CA_Path: 
              Master_SSL_Cert: 
            Master_SSL_Cipher: 
               Master_SSL_Key: 
        Seconds_Behind_Master: NULL
Master_SSL_Verify_Server_Cert: No
                Last_IO_Errno: 0
                Last_IO_Error: 
               Last_SQL_Errno: 0
               Last_SQL_Error: 
  Replicate_Ignore_Server_Ids: 
             Master_Server_Id: 0
1 row in set (0.00 sec)

MariaDB [(none)]> START SLAVE;
Query OK, 0 rows affected (0.00 sec)

MariaDB [(none)]> SHOW SLAVE STATUS\G
*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: 192.168.150.138
                  Master_User: repluser
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: master-bin.000003
          Read_Master_Log_Pos: 497
               Relay_Log_File: relay-bin.000002
                Relay_Log_Pos: 782
        Relay_Master_Log_File: master-bin.000003
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
              Replicate_Do_DB: 
          Replicate_Ignore_DB: 
           Replicate_Do_Table: 
       Replicate_Ignore_Table: 
      Replicate_Wild_Do_Table: 
  Replicate_Wild_Ignore_Table: 
                   Last_Errno: 0
                   Last_Error: 
                 Skip_Counter: 0
          Exec_Master_Log_Pos: 497
              Relay_Log_Space: 1070
              Until_Condition: None
               Until_Log_File: 
                Until_Log_Pos: 0
           Master_SSL_Allowed: No
           Master_SSL_CA_File: 
           Master_SSL_CA_Path: 
              Master_SSL_Cert: 
            Master_SSL_Cipher: 
               Master_SSL_Key: 
        Seconds_Behind_Master: 0
Master_SSL_Verify_Server_Cert: No
                Last_IO_Errno: 0
                Last_IO_Error: 
               Last_SQL_Errno: 0
               Last_SQL_Error: 
  Replicate_Ignore_Server_Ids: 
             Master_Server_Id: 1
1 row in set (0.00 sec)

MariaDB [(none)]> SHOW GLOBAL VARIABLES LIKE 'read_only';   #此時從庫是隻讀模式
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| read_only     | ON    |
+---------------+-------+
1 row in set (0.00 sec)



七、此時一主兩從架構已經配置完成，安裝MHA包
manager節點(node1)：
安裝mha4mysql-manager-0.56-0.el6.noarch.rpm和mha4mysql-node-0.56-0.el6.noarch.rpm
yum install mha4mysql* -y

node節點(node2-node4):
安裝mha4mysql-node-0.56-0.el6.noarch.rpm
yum -y iinstall mha4mysql-node-0.56-0.el6.noarch.rpm

八、初始化MHA
建立配置目錄及配置文件(在node1上執行)
[root@node1 ~]# mkdir /etc/masterha
[root@node1 ~]# vim /etc/masterha/app1.cnf 
[server default]
user=mhauser
password=mhapass
manager_workdir=/data/masterha/app1
master_log=/data/masterha/app1/manager.log
remote_workdir=/data/masterha/app1
ssh_user=root
repl_user=repluser
repl_password=replpass
ping_interval=1

[server1]
hostname=192.168.150.138
candidate_master=1

[server2]
hostname=192.168.150.139
candidate_master=1

[server3]
hostname=192.168.150.140

九、啓動前檢測
ssh互信配置是否OK
[root@node1 ~]# masterha_check_ssh --conf=/etc/masterha/app1.cnf
......
Warning: Permanently added '192.168.150.139' (ECDSA) to the list of known hosts.
Wed Mar 29 17:03:03 2017 - [debug]   ok.
Wed Mar 29 17:03:03 2017 - [info] All SSH connection tests passed successfully.

mysql複製集羣的鏈接配置是否OK
[root@node1 ~]# masterha_check_repl --conf=/etc/masterha/app1.cnf
......
Wed Mar 29 17:04:40 2017 - [info] 
192.168.150.138(192.168.150.138:3306) (current master)
 +--192.168.150.139(192.168.150.139:3306)
 +--192.168.150.140(192.168.150.140:3306)

Wed Mar 29 17:04:40 2017 - [info] Checking replication health on 192.168.150.139..
Wed Mar 29 17:04:40 2017 - [info]  ok.
Wed Mar 29 17:04:40 2017 - [info] Checking replication health on 192.168.150.140..
Wed Mar 29 17:04:40 2017 - [info]  ok.
Wed Mar 29 17:04:40 2017 - [warning] master_ip_failover_script is not defined.
Wed Mar 29 17:04:40 2017 - [warning] shutdown_script is not defined.
Wed Mar 29 17:04:40 2017 - [info] Got exit code 0 (Not master dead).

MySQL Replication Health is OK.

十、啓動MHA
[root@node1 ~]# nohup masterha_manager --conf=/etc/masterha/app1.cnf > /data/masterha/app1/manager.log 2>&
1 &[1] 16989

[root@node1 ~]# tail -f /data/masterha/app1/manager.log 
192.168.150.138(192.168.150.138:3306) (current master)
 +--192.168.150.139(192.168.150.139:3306)
 +--192.168.150.140(192.168.150.140:3306)

Wed Mar 29 21:51:58 2017 - [warning] master_ip_failover_script is not defined.
Wed Mar 29 21:51:58 2017 - [warning] shutdown_script is not defined.
Wed Mar 29 21:51:58 2017 - [info] Set master ping interval 1 seconds.
Wed Mar 29 21:51:58 2017 - [warning] secondary_check_script is not defined. It is highly recommended setti
ng it to check master reachability from two or more routes.Wed Mar 29 21:51:58 2017 - [info] Starting ping health check on 192.168.150.138(192.168.150.138:3306)..
Wed Mar 29 21:51:58 2017 - [info] Ping(SELECT) succeeded, waiting until MySQL doesn't respond..



十一、啓動後查看master節點狀態
[root@node1 ~]# masterha_check_status --conf=/etc/masterha/app1.cnf
app1 (pid:16623) is running(0:PING_OK), master:192.168.150.138


十二、進行故障轉移測試
(1)maser節點關閉mariadb
[root@node2 ~]# killall mysqld mysqld_safe

(2)此時在manager上能夠看到轉移的日誌
----- Failover Report -----

app1: MySQL Master failover 192.168.150.138(192.168.150.138:3306) to 192.168.150.139(192.168.150.139:3306)
 succeeded
Master 192.168.150.138(192.168.150.138:3306) is down!

Check MHA Manager logs at node1.com:/data/masterha/app1/manager.log for details.

Started automated(non-interactive) failover.
The latest slave 192.168.150.139(192.168.150.139:3306) has all relay logs for recovery.
Selected 192.168.150.139(192.168.150.139:3306) as a new master.
192.168.150.139(192.168.150.139:3306): OK: Applying all logs succeeded.
192.168.150.140(192.168.150.140:3306): This host has the latest relay log events.
Generating relay diff files from the latest slave succeeded.
192.168.150.140(192.168.150.140:3306): OK: Applying all logs succeeded. Slave started, replicating from 19
2.168.150.139(192.168.150.139:3306)192.168.150.139(192.168.150.139:3306): Resetting slave info succeeded.
Master failover to 192.168.150.139(192.168.150.139:3306) completed successfully.


故障轉移後，manager會自動中止，此時查看master狀態
[root@node1 ~]# masterha_check_status --conf=/etc/masterha/app1.cnf
app1 is stopped(2:NOT_RUNNING).

(3)查看其它兩個庫狀態
node3   已成功接管master，並可讀寫
MariaDB [(none)]> SHOW MASTER STATUS;
+-------------------+----------+--------------+------------------+
| File              | Position | Binlog_Do_DB | Binlog_Ignore_DB |
+-------------------+----------+--------------+------------------+
| master-bin.000003 |      245 |              |                  |
+-------------------+----------+--------------+------------------+
1 row in set (0.00 sec)

MariaDB [(none)]> SHOW GLOBAL VARIABLES LIKE 'read_only';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| read_only     | OFF   |
+---------------+-------+
1 row in set (0.00 sec)

node4
MariaDB [(none)]> SHOW SLAVE STATUS\G
*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: 192.168.150.139
                  Master_User: repluser
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: master-bin.000003
          Read_Master_Log_Pos: 245
               Relay_Log_File: relay-bin.000002
                Relay_Log_Pos: 530
        Relay_Master_Log_File: master-bin.000003
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
              Replicate_Do_DB: 
          Replicate_Ignore_DB: 
           Replicate_Do_Table: 
       Replicate_Ignore_Table: 
      Replicate_Wild_Do_Table: 
  Replicate_Wild_Ignore_Table: 
                   Last_Errno: 0
                   Last_Error: 
                 Skip_Counter: 0
          Exec_Master_Log_Pos: 245
              Relay_Log_Space: 818
              Until_Condition: None
               Until_Log_File: 
                Until_Log_Pos: 0
           Master_SSL_Allowed: No
           Master_SSL_CA_File: 
           Master_SSL_CA_Path: 
              Master_SSL_Cert: 
            Master_SSL_Cipher: 
               Master_SSL_Key: 
        Seconds_Behind_Master: 0
Master_SSL_Verify_Server_Cert: No
                Last_IO_Errno: 0
                Last_IO_Error: 
               Last_SQL_Errno: 0
               Last_SQL_Error: 
  Replicate_Ignore_Server_Ids: 
             Master_Server_Id: 2
1 row in set (0.00 sec)

MariaDB [(none)]> SHOW GLOBAL VARIABLES LIKE 'read_only';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| read_only     | ON    |
+---------------+-------+
1 row in set (0.00 sec)

(4)提供新的從節點已修復複製集羣
master界定啊故障後，須要從新準備好一個新的MySQL節點。基於來自於master節點的備份恢復後，將其從新配置爲mster的從節點便可。新加入節點IP爲原master節點IP，不然還得修改appl.cnf中相應的設置，最後再次啓動manager，並再次檢查狀態。

[root@node2 ~]# rm -rf /var/lib/mysql/*
[root@node2 ~]# vim /etc/my.cnf
添加從庫兩選項
read_only = 1
relay_log_purge = 0
[root@node2 ~]# systemctl start mariadb.service
[root@node2 ~]# mysql
Welcome to the MariaDB monitor.  Commands end with ; or \g.
Your MariaDB connection id is 2
Server version: 5.5.52-MariaDB MariaDB Server

Copyright (c) 2000, 2016, Oracle, MariaDB Corporation Ab and others.

Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.

MariaDB [(none)]> GRANT REPLICATION SLAVE,REPLICATION CLIENT ON *.* TO 'repluser'@'192.168.%.%' IDENTIFIED
 BY 'replpass';Query OK, 0 rows affected (0.00 sec)

MariaDB [(none)]> GRANT ALL ON *.* TO 'mhauser'@'192.168.%.%' IDENTIFIED BY 'mhapass';
Query OK, 0 rows affected (0.00 sec)

MariaDB [(none)]> FLUSH PRIVILEGES;
Query OK, 0 rows affected (0.00 sec)

MariaDB [(none)]> CHANGE MASTER TO MASTER_HOST='192.168.150.139',MASTER_USER='repluser',MASTER_PASSWORD='r
eplpass',MASTER_LOG_FILE='master-bin.000003',MASTER_LOG_POS=245;
Query OK, 0 rows affected (0.01 sec)

MariaDB [(none)]> SHOW SLAVE\G
ERROR 1064 (42000): You have an error in your SQL syntax; check the manual that corresponds to your MariaD
B server version for the right syntax to use near '' at line 1MariaDB [(none)]> START SLAVE;
Query OK, 0 rows affected (0.01 sec)

MariaDB [(none)]> SHOW SLAVE STATUS\G
*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: 192.168.150.139
                  Master_User: repluser
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: master-bin.000003
          Read_Master_Log_Pos: 245
               Relay_Log_File: relay-bin.000002
                Relay_Log_Pos: 530
        Relay_Master_Log_File: master-bin.000003
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
              Replicate_Do_DB: 
          Replicate_Ignore_DB: 
           Replicate_Do_Table: 
       Replicate_Ignore_Table: 
      Replicate_Wild_Do_Table: 
  Replicate_Wild_Ignore_Table: 
                   Last_Errno: 0
                   Last_Error: 
                 Skip_Counter: 0
          Exec_Master_Log_Pos: 245
              Relay_Log_Space: 818
              Until_Condition: None
               Until_Log_File: 
                Until_Log_Pos: 0
           Master_SSL_Allowed: No
           Master_SSL_CA_File: 
           Master_SSL_CA_Path: 
              Master_SSL_Cert: 
            Master_SSL_Cipher: 
               Master_SSL_Key: 
        Seconds_Behind_Master: 0
Master_SSL_Verify_Server_Cert: No
                Last_IO_Errno: 0
                Last_IO_Error: 
               Last_SQL_Errno: 0
               Last_SQL_Error: 
  Replicate_Ignore_Server_Ids: 
             Master_Server_Id: 2
1 row in set (0.00 sec)

再次開啓manager查看狀態，狀態所有OK，主庫變動爲node3
[root@node1 ~]# nohup masterha_manager --conf=/etc/masterha/app1.cnf > /data/masterha/app1/manager.log 2>&1 &

[root@node1 ~]# masterha_check_status --conf=/etc/masterha/app1.cnf
app1 (pid:17229) is running(0:PING_OK), master:192.168.150.139