利用keepalived構建高可用MySQL-HA

關於MySQL-HA，目前有多種解決方案，比如heartbeat、drbd、mmm、共享存儲，但是它們各有優缺點。heartbeat、drbd配置較爲複雜，需要自己寫腳本才能實現MySQL自動切換，對於不會腳本語言的人來說，這無疑是一種腦裂問題；對於mmm，生產環境中很少有人用，且mmm 管理端需要單獨運行一臺服務器上，要是想實現高可用，就得對mmm管理端做HA，這樣無疑又增加了硬件開支；對於共享存儲，個人覺得MySQL數據還是放在本地較爲安全，存儲設備畢竟存在單點隱患。使用MySQL雙master+keepalived是一種非常好的解決方案，在MySQL-HA環境中，MySQL互爲主從關係，這樣就保證了兩臺MySQL數據的一致性，然後用keepalived實現虛擬IP，通過keepalived自帶的服務監控功能來實現MySQL故障時自動切換。

下面，我把即將上線的一個生產環境中的架構與大家分享一下，看一下這個架構中，MySQL-HA是如何實現的，環境拓撲如下

MySQL-VIP：192.168.1.200
MySQL-master1：192.168.1.201
MySQL-master2：192.168.1.202

OS版本：CentOS 5.4
MySQL版本：5.0.89
Keepalived版本：1.1.20

一、MySQL master-master配置

1、修改MySQL配置文件

兩臺MySQL均如要開啓binlog日誌功能，開啓方法：在MySQL配置文件[MySQLd]段中加上log-bin=MySQL-bin選項

兩臺MySQL的server-ID不能一樣，默認情況下兩臺MySQL的serverID都是1，需將其中一臺修改爲2即可

2、將192.168.1.201設爲192.168.1.202的主服務器

在192.168.1.201上新建授權用戶

MySQL> grant replication slave on *.* to 'replication'@'%' identified by 'replication';
Query OK, 0 rows affected (0.00 sec)

MySQL> show master status;
+------------------+----------+--------------+------------------+
| File             | Position | Binlog_Do_DB | Binlog_Ignore_DB |
+------------------+----------+--------------+------------------+
| MySQL-bin.000003 |      374 |              |                  |
+------------------+----------+--------------+------------------+
1 row in set (0.00 sec)

在192.168.1.202上將192.168.1.201設爲自己的主服務器

MySQL> change master to master_host='192.168.1.201',master_user='replication',master_password='replication',master_log_file='MySQL-bin.000003',master_log_pos=374;
Query OK, 0 rows affected (0.05 sec)

MySQL> start slave;
Query OK, 0 rows affected (0.00 sec)

MySQL> show slave status\G
*************************** 1. row ***************************
             Slave_IO_State: Waiting for master to send event
                Master_Host: 192.168.1.201
                Master_User: replication
                Master_Port: 3306
              Connect_Retry: 60
            Master_Log_File: MySQL-bin.000003
        Read_Master_Log_Pos: 374
             Relay_Log_File: MySQL-master2-relay-bin.000002
              Relay_Log_Pos: 235
      Relay_Master_Log_File: MySQL-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: 374
            Relay_Log_Space: 235
            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
1 row in set (0.00 sec)

3、將192.168.1.202設爲192.168.1.201的主服務器

在192.168.1.202上新建授權用戶

MySQL> grant replication slave on *.* to 'replication'@'%' identified by 'replication';
Query OK, 0 rows affected (0.00 sec)

MySQL> show master status;
+------------------+----------+--------------+------------------+
| File             | Position | Binlog_Do_DB | Binlog_Ignore_DB |
+------------------+----------+--------------+------------------+
| MySQL-bin.000003 |      374 |              |                  |
+------------------+----------+--------------+------------------+
1 row in set (0.00 sec)

在192.168.1.201上，將192.168.1.202設爲自己的主服務器

MySQL> change master to master_host='192.168.1.202',master_user='replication',master_password='replication',master_log_file='MySQL-bin.000003',master_log_pos=374;
Query OK, 0 rows affected (0.05 sec)

MySQL> start slave;
Query OK, 0 rows affected (0.00 sec)

MySQL> show slave status\G
*************************** 1. row ***************************
             Slave_IO_State: Waiting for master to send event
                Master_Host: 192.168.1.202
                Master_User: replication
                Master_Port: 3306
              Connect_Retry: 60
            Master_Log_File: MySQL-bin.000003
        Read_Master_Log_Pos: 374
             Relay_Log_File: MySQL-master1-relay-bin.000002
              Relay_Log_Pos: 235
      Relay_Master_Log_File: MySQL-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: 374
            Relay_Log_Space: 235
            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
1 row in set (0.00 sec)

4、MySQL同步測試

如上述均正確配置，現在任何一臺MySQL上更新數據都會同步到另一臺MySQL，MySQL同步在此不再演示

二、keepalived安裝及配置

1、192.168.1.201服務器上keepalived安裝及配置

安裝keepalived

#tar zxvf keepalived-1.1.20.tar.gz
#cd keepalived-1.1.20
#./configure --prefix=/usr/local/keepalived --with-kernel-dir=/usr/src/kernels/2.6.18-164.el5-i686
#make && make install

配置keepalived

我們自己在新建一個配置文件，默認情況下keepalived啓動時會去/etc/keepalived目錄下找配置文件

#mkdir /etc/keepalived
#vi /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
     notification_email {
     [email protected]
     }
     notification_email_from [email protected]
     smtp_server 127.0.0.1
     smtp_connect_timeout 30
     router_id MySQL-ha
     }

vrrp_instance VI_1 {
     state BACKUP   #兩臺配置此處均是BACKUP
     interface eth0
     virtual_router_id 51
     priority 100   #優先級，另一臺改爲90
     advert_int 1
     nopreempt  #不搶佔，只在優先級高的機器上設置即可，優先級低的機器不設置
     authentication {
     auth_type PASS
     auth_pass 1111
     }
     virtual_ipaddress {
     192.168.1.200
     }
     }

virtual_server 192.168.1.200 3306 {
     delay_loop 2   #每個2秒檢查一次real_server狀態
     lb_algo wrr   #LVS算法
     lb_kind DR    #LVS模式
     persistence_timeout 60   #會話保持時間
     protocol TCP
     real_server 192.168.1.201 3306 {
     weight 3
     notify_down /usr/local/MySQL/bin/MySQL.sh  #檢測到服務down後執行的腳本
     TCP_CHECK {
     connect_timeout 10    #連接超時時間
     nb_get_retry 3       #重連次數
     delay_before_retry 3   #重連間隔時間
     connect_port 3306   #健康檢查端口
     }
     }

編寫檢測服務down後所要執行的腳本

#vi /usr/local/MySQL/bin/MySQL.sh
#!/bin/sh
pkill keepalived
#chmod +x /usr/local/MySQL/bin/MySQL.sh

注：此腳本是上面配置文件notify_down選項所用到的，keepalived使用notify_down選項來檢查real_server的服務狀態，當發現real_server服務故障時，便觸發此腳本；我們可以看到，腳本就一個命令，通過pkill keepalived強制殺死keepalived進程，從而實現了MySQL故障自動轉移。另外，我們不用擔心兩個MySQL會同時提供數據更新操作，因爲每臺MySQL上的keepalived的配置裏面只有本機MySQL的IP+VIP，而不是兩臺MySQL的IP+VIP

啓動keepalived

#/usr/local/keepalived/sbin/keepalived –D
#ps -aux | grep keepalived

測試

找一臺局域網PC，然後去ping  MySQL的VIP，這時候MySQL的VIP是可以ping的通的

停止MySQL服務，看keepalived健康檢查程序是否會觸發我們編寫的腳本

2、192.168.1.202上keepalived安裝及配置

安裝keepalived

#tar zxvf keepalived-1.1.20.tar.gz
#cd keepalived-1.1.20
#./configure --prefix=/usr/local/keepalived --with-kernel-dir=/usr/src/kernels/2.6.18-164.el5-i686
#make && make install

配置keepalived

這臺配置和上面基本一樣，但有三個地方不同：優先級爲90、無搶佔設置、real_server爲本機IP

#mkdir /etc/keepalived
#vi /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
     notification_email {
     [email protected]
     }
     notification_email_from [email protected]
     smtp_server 127.0.0.1
     smtp_connect_timeout 30
     router_id MySQL-ha
     }

vrrp_instance VI_1 {
     state BACKUP
     interface eth0
     virtual_router_id 51
     priority 90
     advert_int 1
     authentication {
     auth_type PASS
     auth_pass 1111
     }
     virtual_ipaddress {
     192.168.1.200
     }
     }

virtual_server 192.168.1.200 3306 {
     delay_loop 2
     lb_algo wrr
     lb_kind DR
     persistence_timeout 60
     protocol TCP
     real_server 192.168.1.202 3306 {
     weight 3
     notify_down /usr/local/MySQL/bin/MySQL.sh
     TCP_CHECK {
     connect_timeout 10
     nb_get_retry 3
     delay_before_retry 3
     connect_port 3306
     }
     }

編寫檢測服務down後所要執行的腳本

#vi /usr/local/MySQL/bin/MySQL.sh
#!/bin/sh
pkill keepalived
#chmod +x /usr/local/MySQL/bin/MySQL.sh

啓動keepalived
#/usr/local/keepalived/sbin/keepalived –D
#ps -aux | grep keepalived

測試

停止MySQL服務，看keepalived健康檢查程序是否會觸發我們編寫的腳本

三、測試

MySQL遠程登錄測試

我們找一臺安裝有MySQL客戶端的windows，然後登錄VIP，看是否能登錄，在登錄之兩臺MySQL服務器都要授權允許從遠程登錄

MySQL> grant all privileges on *.* to 'root'@'%' identified by '123456';
Query OK, 0 rows affected (0.00 sec)

MySQL> flush privileges;
Query OK, 0 rows affected (0.00 sec)

使用客戶端登錄VIP測試

C:\MySQL\bin>MySQL.exe -uroot -p123456 -h192.168.1.200 -P3306
Welcome to the MySQL monitor.  Commands end with ; or \g.
Your MySQL connection id is 224
Server version: 5.0.89-log Source distribution

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

MySQL>

● keepalived故障轉移測試

※在windows客戶端一直去ping  VIP，然後關閉192.168.1.201上的keepalived，正常情況下VIP就會切換到192.168.1.202上面去

※開啓192.168.1.201上的keepalived，關閉192.168.1.202上的keepalived，看是否能自動切換，正常情況下VIP又會屬於192.168.1.201

注：keepalived切換速度還是非常塊的，整個切換過程只需1-3秒

● MySQL故障轉移測試

※在192.168.1.201上關閉MySQL服務，看VIP是否會切換到192.168.1.202上

※開啓192.168.1.201上的MySQL和keepalived，然後關閉192.168.1.202上的MySQL，看VIP是否會切換到192.168.1.201上

下面是用windows客戶端連接的MySQL的VIP，在切換時我執行了一個MySQL查詢命令，從執行show databases到顯示出結果時間爲3-5秒（大家可以看到上面有個錯誤提示，不過不用擔心，因爲我們的keepalived切換大概爲3秒左右，這3秒左右VIP是誰都不屬於的）

MySQL> show databases;
ERROR 2006 (HY000): MySQL server has gone away
No connection. Trying to reconnect...
Connection id:    592
Current database: *** NONE ***

+--------------------+
| Database           |
+--------------------+
| information_schema |
| MySQL              |
| test               |
+--------------------+
3 rows in set (9.01 sec)

後話：世間萬事萬物，都不具備絕對的完美，就像上面的MySQL-HA一樣，keepalived只能做到對3306的健康檢查，但是做不到比如像MySQL複製中的slave-SQL、slave-IO進程的檢查。所以要想做到一些細緻的健康檢查，還得需要藉助額外的監控工具，比如nagios，然後用nagios實現短信、郵件報警，從而能夠有效地解決問題。