双主模型的Nginx高可用负载均衡集群

实验目的：使用keepalived实现Nginx的双主高可用负载均衡集群

实验环境：两台Nginx proxy(双主Nginx，各需要两块网卡，eth0连接内网，eth1连接外网)、两台web server(请求的负载均衡)、一台client用于验证结果。

注意：为了不影响实验结果，在实验开始前先关闭iptables和selinux

操作步骤：

一、配置IP

1.配置A主机的IP

# ip addr add dev eth0 192.168.10.2/24

2.配置B主机的IP

# ip addr add dev eth0 192.168.10.23/24

3.配置C主机的IP

# ip addr add dev eth0 192.168.10.3/24

4.配置D主机的IP

# ip addr add dev eth0 192.168.10.33/24

二、配置web服务(C和D主机都做同样配置，只需修改默认主页中的IP地址为本机的IP即可，以示区别)

1.安装apache

# yum -y install apache

2.创建默认主页

# vim /var/www/html/index.html

<h1>192.168.10.3</h1>

3.启动apache

# service httpd start

三、配置sorry_server(此服务配置于Nginx proxy主机上，两台Nginx proxy都做同样配置，只需修改默认主页中的IP地址为本机的IP即可，以示区别)

1.安装apache

# yum -y install apache

2.创建默认主页

# vim /var/www/html/index.html

<h1>sorry_server:192.168.10.2</h1>

3.修改监听端口为8080，以免与nginx所监听的端口冲突

# vim /etc/httpd/conf/httpd.conf

Listen 8080

4.启动apache服务

四、配置代理(两台Nginx proxy都做同样配置)

1.安装nginx

# yum -y install nginx

2.定义upstream集群组，在http{}段中定义；

# vim /etc/nginx/nginx.conf

        http {

            upstream websrvs {

                server 192.168.10.3:80;

                server 192.168.10.33:80;

                server 127.0.0.1:8080 backup;

            }

        }

3.调用定义的集群组，在server{}段的location{}段中调用；

# vim /etc/nginx/conf.d/default.conf

        server {

            location / {

                proxy_pass http://wersrvs;

                index index.html;

            }

        }

4.启动服务

# service nginx start

五、配置keepalived

A主机上操作

1.安装keepalived

# yum -y install keepalived

2.编辑A主机的配置文件/etc/keepalived/keepalived.conf，作如下配置：

! Configuration File for keepalived

    global_defs {

    notification_email {

        root@localhost

    }

    notification_email_from keepalived@localhost

    smtp_server 127.0.0.1

    smtp_connect_timeout 30

    router_id centos6

    vrrp_mcast_group4 224.0.100.39

    }

    vrrp_script chk_down {

        script “[[ -f /etc/keepalived/down ]] && exit 1 || exit 0”

        interval 1

        weight -5

    }

    vrrp_script chk_nginx {

        script “killall -0 nginx && exit 0 || exit 1”

        interval 1

        weight -5

        fall 2

        rise 1

    }

    vrrp_instance ngx {

        state MASTER

        interface eth1

        virtual_router_id 14

        priority 100

        advert_int 1

        authentication {

            auth_type PASS

            auth_pass MDQ41fTp

        }

        virtual_ipaddress {

            192.168.20.100/24 dev eth1

        }

        track_script {

            chk_down

            chk_nginx

        }

    }

    vrrp_instance ngx2 {

        state BACKUP

        interface eth1

        virtual_router_id 15

        priority 98

        advert_int 1

        authentication {

            auth_type PASS

            auth_pass XYZ41fTp

        }

        virtual_ipaddress {

            192.168.20.200/24 dev eth1

        }

        track_script {

            chk_down

        chk_nginx

        }

    }

B主机也作同样配置，稍作修改即可，需要修改的地方如下：

    vrrp_instance ngx {

        state BACKUP

        priority 98

    }

    vrrp_instance ngx2 {

        state MASTER

        priority 100

    }

六、模拟故障，验证结果

1.启动两台Nginx proxy的keepalived服务

# service keepalived start

2.访问192.168.20.100，结果应是后端的web server轮询响应请求

3.访问192.168.20.200，结果应是后端的web server轮询响应请求

4.将后端的web server关闭一台，访问192.168.20.100或192.168.20.200，响应请求的将只是另一台正常运行web server的主机

5.将后端的web server都关闭，此时访问192.168.20.100或192.168.20.200，响应请求的将只是Nginx proxy中定义的主server中的sorry_server

6.关闭一台Nginx proxy 的nginx服务，备server将把IP地址添加到本机，继续提供服务，此时访问192.168.20.100或192.168.20.200并不会有任何察觉