MySQL高可用架构之Keepalived+主从架构部署

针对目前公司的数据库架构要做统计规划和调整，所以针对不同的业务环境要选择合适的架构就比较重要，同时作为运维人员又要考虑到维护的便捷性（不能做过多架构维护），最终停留在mha+vip/主从+vip/mycat集群/域名+主从这几种架构或组合，这里就分布对不同架构部署做下记录并对比起优劣，针对不同场景做出合适的选择。

本文主要针对主备自动切换，即VIP+主从方式，其中VIP通过keepalived控制。

一、环境描述

服务器IP	操作系统	数据库	硬件配置	备注
192.168.1.23	RHEL6.6_x86_64	MySQL 5.7.21	内存：16GB CPU：4cores 硬盘:100GB	VIP：192.168.1.27 Keepalived:1.4.4
192.168.1.24	RHEL6.6_x86_64	MySQL 5.7.21	内存：16GB CPU：4cores 硬盘:100GB

二、业务场景

1、优点

1)部署简单，管理方便。

2)可实现主备库VIP切换，对业务无感知。

2、缺点

1)存在脑裂风险

2)存在切换主备，状态无检测风险

3)跨VLAN不支持

3、业务场景

1)双节点数据库服务器，仅仅为了实现主备VIP切换业务无感知

2)多个从节点，通过LVS，做负载均衡（现在不推荐这种架构实现负载均衡）

三、数据库安装和主从部署

略

四、keepalived安装部署

1.keepalived下载

下载地址：http://www.keepalived.org/software/keepalived-1.4.4.tar.gz

2.keepalived安装

主库：
[root@node1 tools]# tar -xzvf keepalived-1.4.4.tar.gz
[root@node1 tools]# cd keepalived-1.4.4
[root@node1 keepalived-1.4.4]# ./configure  --prefix=/usr/local/keepalived
[root@node1 keepalived-1.4.4]# make -j 4 
[root@node1 keepalived-1.4.4]# make install
[root@node1 keepalived-1.4.4]# cp -rp ./keepalived/etc/init.d/keepalived /etc/init.d/
[root@node1 keepalived-1.4.4]# chkconfig  --add keepalived
[root@node1 keepalived-1.4.4]# mkdir /etc/keepalived
[root@node2 keepalived-1.4.4]# cp -rp ./bin/keepalived  /usr/bin/
[root@node1 keepalived-1.4.4]# cp ./keepalived/etc/keepalived/keepalived.conf /etc/keepalived
[root@node1 keepalived-1.4.4]# cp -rp /usr/local/keepalived/etc/sysconfig/keepalived  /etc/sysconfig/
[root@node1 keepalived-1.4.4]# service keepalived status
keepalived is stopped
备库：
安装keepalived同上
###备注
*** WARNING - this build will not support IPVS with IPv6. Please install libnl/libnl-3 dev libraries to support IPv6 with IPVS.
###

3.keepalived配置

主库配置：

[root@node1 keepalived-1.4.4]# cp -rp  /etc/keepalived/keepalived.conf /etc/keepalived/keepalived.conf_bak
[root@node1 keepalived-1.4.4]# >/etc/keepalived/keepalived.conf
[root@node1 keepalived-1.4.4]# vi /etc/keepalived/keepalived.conf

! Configuration File for keepalived

global_defs {
   notification_email {
     rangle@163.com
   }
   notification_email_from [email protected]
   smtp_server 127.0.0.1
   smtp_connect_timeout 30
   router_id Node_Master
}

vrrp_instance VI_1 {
    state BACKUP          ##可配置master和backup模式，为了防止脑裂现象，主备均需要设置为backup模式，master模式会抢占VIP
    interface eth1
    virtual_router_id 43  ##VRRP组名，两个节点的设置必须一样，以指明各个节点属于同一VRRP组
    priority 100
    nopreempt             ##配合backup,防止切换后，主库服务恢复正常后，IP漂移过来
    advert_int 1          ###组播信息发送间隔，两个节点设置必须一样
    authentication {      ###设置验证信息，两个节点必须一致
        auth_type PASS
        auth_pass 1111
    }
    virtual_ipaddress {   ###指定虚拟IP, 两个节点设置必须一样
        192.168.1.27 label eth1:1
    }
}
virtual_server 192.168.1.27 3306 {
    delay_loop 6
    lb_algo wrr          #lvs调度算法rr|wrr|lc|wlc|lblc|sh|dh
    lb_kind DR           #负载均衡转发规则NAT|DR|RUN
    persistence_timeout 50
    protocol TCP

    real_server 192.168.1.23 3306 {
      weight 1
      notify_down /etc/keepalived/mysql.sh 
      TCP_CHECK {
      connect_timeout 10
      nb_get_retry 3 
      delay_before_retry 3
      connect_port 3306 
        }
    }
}

备库配置：

[root@node2 keepalived-1.4.4]# cp -rp  /etc/keepalived/keepalived.conf /etc/keepalived/keepalived.conf_bak
[root@node2 keepalived-1.4.4]# >/etc/keepalived/keepalived.conf
[root@node2 keepalived-1.4.4]# vi /etc/keepalived/keepalived.conf

! Configuration File for keepalived
  global_defs {
      notification_email {
      rangle@163.com
      }
      notification_email_from rangle@163.com
      smtp_server 127.0.0.1
      smtp_connect_timeout 30
      router_id Node_backup
     }
 vrrp_instance VI_1 {
      state BACKUP
      interface eth1
      virtual_router_id 43
      priority 90                   ##主优先级高于备，主设置100，备可设置90,（1-254之间）
      advert_int 1
      authentication {
      auth_type PASS
      auth_pass 1111
      }
      virtual_ipaddress {
      192.168.1.27 label eth1:1
      }
      }
 virtual_server 192.168.1.27 3306 {
      delay_loop 2                 ##每隔2秒检查真实服务器状态
      lb_algo wrr                  ##LVS算法，如果仅做高可用可不设
      lb_kind DR                   ##LVS模式，如果仅做高可用可不设
      persistence_timeout 60 
      protocol TCP 

      real_server 192.168.1.24 3306{
      weight 3
      notify_down /etc/keepalived/mysql.sh     ##如果mysql服务宕掉，执行的脚本
      echo '3' >  /etc/keepalived/t.log
      TCP_CHECK {
      connect_timeout 10
      nb_get_retry 3     
      delay_before_retry 3
      connect_port 3306 
      }
      }
}

mysql.sh配置内容：

#!/bin/bash

run_status=`service keepalived status|grep running|wc -l`
pro_status=`ps -ef |grep keepalived|grep -v grep |wc -l`

service keepalived stop

if [ ${run_status} != 0 ] || [ ${pro_status} != 0 ]
then
   pkill keepalived
fi

五、keepalived优化

1、keepalived+mysql风险点

1）脑裂风险

现象：

如果由于主备内部网络通信（vrrp组播）故障（网络波动、防火墙等）导致VIP切换，但实际主备服务均正常，此时VIP会存在两个节点上，这个就会导致主备数据同时写入导致数据不一致。

优化方案：

****主库节点增加脑裂检查脚本，通过本机增加网关链路的检查，增加仲裁节点，判断是否本机对外的网络出现问题，此时在配合VRRP组播，如果网络存在问题则直接关闭keepalived和mysql服务，然后发送告警邮件、短信。

2）主从同步漂移风险

现象：

VIP漂移不会检查从库状态，即无论从库数据是否同步，VIP都将完成漂移，数据不一致风险较大。

优化方案：

****在从节点增加主从状态检查脚本，判断主从同步进程是否正常，同时判断是否有数据延迟，如果有延迟或同步错误则不能完成切换，直接关闭keepalived服务

2、优化后配置如下

具体优化配置和脚本如下：

主库配置：

备库配置：

脑裂检查脚本：

3、防火墙端口开放

VRRP是用IP多播的方式（默认多播地址224.0.0.18）实现高可用对之间通信的，如果开启防火墙记得开放这个vrrp协议对应的策略