LVS+keepalived+Nginx实现负载均衡时前后端故障全自动化处理

最近帮朋友公司配了一下LVS,实现全自动化处理故障问题,基本环境如下
VIP : 192.168.101.100
LVS1: 192.168.101.103
LVS2: 192.168.101.104
RS1: 192.168.101.105
RS2: 192.168.101.108

后端还有一些服务器IP就不一一写出来了


首先各个服务器上配置好各种需要的环境,所需要依赖包,然后在master 和 backup上装好keepalived 和 LVS,直接yum安装 .不启动LVS,启动keepalived就好,下面直接说配置 全部在keepalived上

[root@localhost shell]# cat /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
   router_id LVS_DEVEL
}
vrrp_sync_group VGM {     ###定义一个vrrp组
group {
VI_1
}
notify_master "/data/shell/arp.sh"
notify_backup "/data/shell/arp.sh"
}
vrrp_script chk_nginx {    #### 定义脚本
script "/data/shell/check_nginx.sh"
interval 1
weight -2
}
vrrp_script chk_nginx2 {      #### 定义脚本
script "/data/shell/check_nginx2.sh"
interval 1
weight -2
}
vrrp_instance VI_1 {
    state MASTER    ###backup上配置相同,只需要改下相关权重等
    interface eth1
    virtual_router_id 51
    priority 101
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 1111
    }
track_script {    #### 执行脚本检测
chk_nginx
chk_nginx2
}
    virtual_ipaddress {
        192.168.101.100
 }
}
virtual_server 192.168.101.100 80 {
    delay_loop 2
    lb_algo wrr
    lb_kind DR
    persistence_timeout 60
    protocol TCP
    real_server 192.168.101.105 80 {
       weight 3
        HTTP_GET {
        url {
        path /test.html
        digest 9a8acfd3c07a45099bcde1a97a9fb7c5
        }
        connect_timeout 5
        nb_get_retry 3
        delay_before_retry 3
        connect_port 80
}
}
 real_server 192.168.101.108 80 {
        weight 3
        HTTP_GET {
        url {
        path /test2.html
        digest 9649358e69985102d0da5ff240e1f2a4
        }
        connect_timeout 5
        nb_get_retry 3
        delay_before_retry 3
        connect_port 80
}
}
}

[root@localhost shell]# cat /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
   router_id LVS_DEVEL
}
vrrp_sync_group VGM {     ###定义一个vrrp组
group {
VI_1
}
notify_master "/data/shell/arp.sh"
notify_backup "/data/shell/arp.sh"
}
vrrp_script chk_nginx {    #### 定义脚本
script "/data/shell/check_nginx.sh"
interval 1
weight -2
}
vrrp_script chk_nginx2 {      #### 定义脚本
script "/data/shell/check_nginx2.sh"
interval 1
weight -2
}
vrrp_instance VI_1 {
    state MASTER    ###backup上配置相同,只需要改下相关权重等
    interface eth1
    virtual_router_id 51
    priority 101
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 1111
    }
track_script {    #### 执行脚本检测
chk_nginx
chk_nginx2
}
    virtual_ipaddress {
        192.168.101.100  
}
}
virtual_server 192.168.101.100 80 {
    delay_loop 2
    lb_algo wrr
    lb_kind DR
    persistence_timeout 60
    protocol TCP
    real_server 192.168.101.105 80 {
       weight 3
    HTTP_GET {
        url {
        path /test.html
        digest 9a8acfd3c07a45099bcde1a97a9fb7c5
        }
    connect_timeout 5
    nb_get_retry 3
    delay_before_retry 3
    connect_port 80
}
}
real_server 192.168.101.108 80 {
    weight 3
    HTTP_GET {
        url {
        path /test2.html
        digest 9649358e69985102d0da5ff240e1f2a4
        }
    connect_timeout 5
    nb_get_retry 3
    delay_before_retry 3
    connect_port 80
}
}    
}


如果你前端开启了防火墙或者selinux,请关闭防火墙测试或者selinux测试,否则你会出现两个master,哈哈,两个vip地址同时出现,或者起不来服务,如果不关闭iptables,请加入以下策略

###允许80端口对外提供服务
-A INPUT -m state --state NEW -m tcp -p tcp --dport 80 -j ACCEPT
###DR模式，当用户请求LVS-DR VIP时，只有DR响应客户端的ARP广播包，允许vrrp虚拟路由器冗余协议-A   INPUT   -d   224.0.0.0/8   -j   ACCEPT
-A   INPUT    -p   vrrp   -j   ACCEPT

###允许80端口对外提供服务  
-A INPUT -m state --state NEW -m tcp -p tcp --dport 80 -j ACCEPT  
###DR模式，当用户请求LVS-DR VIP时，只有DR响应客户端的ARP广播包，允许vrrp虚拟路由器冗余协议-A   INPUT   -d   224.0.0.0/8   -j   ACCEPT  
-A   INPUT    -p   vrrp   -j   ACCEPT

下面是几个在master 和backup上执行的高效脚本,检测后端nginx的
[root@localhost shell]# cat arp.sh
#!/bin/bash
VIP=192.168.101.100
GATEWAY=192.168.101.1
/sbin/arping -I eth1 -c 5 -s $VIP $GATEWAY &>/dev/null
########################################################
[root@localhost shell]# cat check_nginx.sh
!/bin/bash
url="http://192.168.101.105"      ####realserver web地址检测
status=$(/usr/bin/curl -s --head "$url" | awk '/HTTP/ {print $2}')
if [ "$status" != "200" ]; then
/data/shell/arp.sh start
fi
[root@localhost shell]# cat check_nginx2.sh
!/bin/bash
url="http://192.168.101.108"      ####realserver web地址检测
status=$(/usr/bin/curl -s --head "$url" | awk '/HTTP/ {print $2}')
if [ "$status" != "200" ]; then
/data/shell/arp.sh start
fi

[root@localhost shell]# cat arp.sh
#!/bin/bash
VIP=192.168.101.100
GATEWAY=192.168.101.1
/sbin/arping -I eth1 -c 5 -s $VIP $GATEWAY &>/dev/null
########################################################
[root@localhost shell]# cat check_nginx.sh
!/bin/bash
url="http://192.168.101.105"      ####realserver web地址检测
status=$(/usr/bin/curl -s --head "$url" | awk '/HTTP/ {print $2}')
if [ "$status" != "200" ]; then
/data/shell/arp.sh start
fi
[root@localhost shell]# cat check_nginx2.sh
!/bin/bash
url="http://192.168.101.108"      ####realserver web地址检测
status=$(/usr/bin/curl -s --head "$url" | awk '/HTTP/ {print $2}')
if [ "$status" != "200" ]; then
/data/shell/arp.sh start
fi

 arp缓存清理shell  {arp.sh}
#!/bin/bash
VIP=192.168.101.100
GATEWAY=192.168.101.1
/sbin/arping -I eth1 -c 5 -s 192.168.101.100 192.168.101.1 &>/dev/null

#!/bin/bash
VIP=192.168.101.100
GATEWAY=192.168.101.1
/sbin/arping -I eth1 -c 5 -s 192.168.101.100 192.168.101.1 &>/dev/null

 这个配置会实现 无论前端调度或者后端nginx服务任意一台中断都会在最短的时间响应到正常的请求,经测试,lvs前端两台故障几乎是0秒切换,后端nginx的web服务故障也就是几秒钟的自动切换,如果是DNS解析,请做好解析定时缓存清理