一、keepalived简介

keepalived是分布式部署解决系统高可用的软件，结合lvs(LinuxVirtual Server)使用，解决单机宕机的问题。
keepalived是一个基于VRRP协议来实现IPVS的高可用的解决方案。对于LVS负载均衡来说，如果前端的调度器direct发生故障，则后端的realserver是无法接受请求并响应的。因此，保证前端direct的高可用性是非常关键的，否则后端的服务器是无法进行服务的。而我们的keepalived就可以用来解决单点故障（如LVS的前端direct故障）问题。keepalived的主要工作原理是：运行keepalived的两台服务器，其中一台为MASTER，另一台为BACKUP，正常情况下，所有的数据转换功能和ARP请求响应都是由MASTER完成的，一旦MASTER发生故障，则BACKUP会马上接管MASTER的工作，这种切换时非常迅速的。

 

二、测试环境

下面拿4台虚拟机进行环境测试,实验环境为centos6.6 x86_64,在lvs NAT模式环境下只有前端两台keepalived调度机有公网ip，后端真实机只有内网ip，具体用途和ip如下

服务器类型

公网ip

内网ip

LVS VIP1

192.168.214.70

192.168.211.254

LVS VIP2

192.168.214.71

192.168.211.253

Keepalived host1

192.168.214.76

192.168.211.76

Keepalived host2

192.168.214.77

192.168.211.77

Realserver A

192.168.211.79

       Realserver B

192.168.211.83

拓扑图如下

LVS/NAT双主 + keepalived负载均衡实现_第1张图片

三、软件安装

1、安装lvs所需包ipvsadm

yum install -y ipvsadm

ln -s /usr/src/kernels/`uname -r`  /usr/src/linux

lsmod |grep ip_vs

 

#注意Centos 6.X安装lvs，使用1.26版本。并且需要先安装yuminstall libnl* popt* -y

 

执行ipvsadm(modprobe ip_vs)把ip_vs模块加载到内核

[root@test85 ~]# ipvsadm -L -n

IP Virtual Server version 1.2.1 (size=4096)

Prot LocalAddress:Port Scheduler Flags

 -> RemoteAddress:Port          Forward Weight ActiveConn InActConn

 

#IP Virtual Server version 1.2.1 ---- ip_vs内核模块版本

 

 

2、安装keepalived

yum install -y keepalived

chkconfig keepalived on

注：在centos7系列系统中开机自动启动使用systemctl enable keepalived

 

四、开启路由转发 

在lvs/NAT模式下前端两台调度机192.168.211.76和192.168.211.77上需要开启路由转发功能

vi /etc/sysctl.conf

修改net.ipv4.ip_forward = 0为net.ipv4.ip_forward= 1

保存退出后，使用systcl –p命令让其生效

 

五、keepalived配置

先看下211.76 keepalived配置文件

根据拓扑图得知211.76对应的是vip1 192.168.214.70的主，vip2192.168.214.71的备。

具体详细的参数说明见上篇LVS/DR + keepalived负载均衡实现


[root@localhost ~]# cat /etc/keepalived/keepalived.conf 
! Configuration File for keepalived

global_defs {
   notification_email {
     [email protected]
   }
   notification_email_from [email protected]
   smtp_server mail.test.com
   smtp_connect_timeout 30
   router_id LVS_DEVEL
}
vrrp_sync_group VG1 {        #lvs vip1组
   group {
      VI_1
      VI_GATEWAY1
   }
} 
vrrp_sync_group VG2 {   #lvs vip2组
   group {
      VI_2
      VI_GATEWAY2
   }
} 

vrrp_instance VI_GATEWAY1 {      #lvs vip1的配置
    state MASTER
    interface eth1
    lvs_sync_daemon_inteface eth1
    virtual_router_id 51
    priority 150
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 1111
    }
    virtual_ipaddress {
        192.168.211.254
    }
}
vrrp_instance VI_1 {
    state MASTER
    interface eth0
    lvs_sync_daemon_inteface eth0
    virtual_router_id 52
    priority 150
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 1111
    }
    virtual_ipaddress {
    192.168.214.70  
    }
}
vrrp_instance VI_GATEWAY2 {  #lvs vip2的配置
    state BACKUP
    interface eth1
    lvs_sync_daemon_inteface eth1
    virtual_router_id 53
    priority 100
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 1111
    }
    virtual_ipaddress {
        192.168.211.253
    }
}
vrrp_instance VI_2 {
    state BACKUP
    interface eth0
    lvs_sync_daemon_inteface eth0
    virtual_router_id 54
    priority 100
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 1111
    }
    virtual_ipaddress {
    192.168.214.71
    }
}

virtual_server 192.168.214.70 80 {     #vip1 
    delay_loop 6
    lb_algo rr
    lb_kind NAT
    #nat_mask 255.255.255.0
    persistence_timeout 10
    protocol TCP

    real_server 192.168.211.79 80 {       #vip1对应的后端真实机
        weight 100
        TCP_CHECK {
            connect_timeout 3
            connect_port 80
            nb_get_retry 3
            delay_before_retry 3
        }
    }
}
virtual_server 192.168.214.71 8080 {     #vip2
    delay_loop 6
    lb_algo rr
    lb_kind NAT
   #nat_mask 255.255.255.0
    persistence_timeout 10
    protocol TCP

   real_server 192.168.211.83 8080 {   #vip2对应的后端真实机
        weight 100
        TCP_CHECK {
            connect_timeout 3
            connect_port 8080
            nb_get_retry 3
            delay_before_retry 3
        }
    }
}


再看下211.77 keepalived配置文件

根据拓扑图得知211.77对应的是vip1 192.168.214.70的备，vip2192.168.214.71的主。


[root@localhost ~]# cat /etc/keepalived/keepalived.conf
! Configuration File for keepalived

global_defs {
  notification_email {
    [email protected]
   }
  notification_email_from [email protected]
  #smtp_server 192.168.200.1
  smtp_connect_timeout 30
  router_id LVS_DEVEL
}
vrrp_sync_group VG1 {
  group {
     VI_1
     VI_GATEWAY1
   }
} 
vrrp_sync_group VG2 {
  group {
     VI_2
     VI_GATEWAY2
   }
} 

vrrp_instance VI_GATEWAY1 {
   state BACKUP
   interface eth1
   lvs_sync_daemon_inteface eth1
    virtual_router_id51
   priority 100
   advert_int 1
   authentication {
       auth_type PASS
       auth_pass 1111
    }
   virtual_ipaddress {
       192.168.211.254
    }
}
vrrp_instance VI_1 {
   state BACKUP
   interface eth0
   lvs_sync_daemon_inteface eth0
   virtual_router_id 52
   priority 100
   advert_int 1
   authentication {
       auth_type PASS
       auth_pass 1111
    }
   virtual_ipaddress {
   192.168.214.70     
    }
}
vrrp_instance VI_GATEWAY2 {
   state MASTER
    interfaceeth1
   lvs_sync_daemon_inteface eth1
   virtual_router_id 53
   priority 150
   advert_int 1
   authentication {
       auth_type PASS
       auth_pass 1111
    }
   virtual_ipaddress {
       192.168.211.253
    }
}
vrrp_instance VI_2 {
   state MASTER
   interface eth0
   lvs_sync_daemon_inteface eth0
   virtual_router_id 54
   priority 150
   advert_int 1
   authentication {
       auth_type PASS
       auth_pass 1111
    }
   virtual_ipaddress {
   192.168.214.71
    }
}

virtual_server 192.168.214.70 80 {
   delay_loop 6
   lb_algo rr
   lb_kind NAT
   #nat_mask 255.255.255.0
   persistence_timeout 10
   protocol TCP

   real_server 192.168.211.79 80 {
       weight 100
       TCP_CHECK {
           connect_timeout 3
           connect_port 80
           nb_get_retry 3
           delay_before_retry 3
       }
    }
}
virtual_server 192.168.214.71 8080 {
   delay_loop 6
   lb_algo rr
   lb_kind NAT
  #nat_mask 255.255.255.0
   persistence_timeout 10
   protocol TCP

  real_server 192.168.211.83 8080 {
       weight 100
       TCP_CHECK {
           connect_timeout 3
           connect_port 8080
           nb_get_retry 3
           delay_before_retry 3
       }
    }
}


六、后端真实机配置

后端web应用机器192.168.211.79需要配置默认网关为vip1内网ip地址192.168.211.254，web应用机器192.168.211.83需要配置默认网关为vip2内网ip地址192.168.211.253

注意，线上环境如果更改默认网关为vip内网地址后，可能会造成连接不上服务器，需要提前添加相关静态路由。


七、启动keepalived服务及查看相关信息

在211.76和211.77上分别启动keepalived服务

在211.76主机上查看信息

通过ip addr可以看到vip1 地址已经绑定在eth0和eth1网口上

[root@localhost ~]# ip addr

1: lo: mtu65536 qdisc noqueue state UNKNOWN

   link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00

   inet 127.0.0.1/8 scope host lo

   inet6 ::1/128 scope host

      valid_lft forever preferred_lft forever

2: eth0: mtu 1500 qdisc pfifo_fast state UP qlen1000

   link/ether 52:54:00:55:b2:d4 brd ff:ff:ff:ff:ff:ff

   inet 192.168.214.76/24 brd 192.168.214.255 scope global eth0

    inet 192.168.214.70/32 scope global eth0

   inet6 fe80::5054:ff:fe55:b2d4/64 scope link

      valid_lft forever preferred_lft forever

3: eth1: mtu 1500 qdisc pfifo_fast state UP qlen1000

   link/ether 52:54:00:85:11:95 brd ff:ff:ff:ff:ff:ff

   inet 192.168.211.76/24 brd 192.168.211.255 scope global eth1

    inet192.168.211.254/32 scope global eth1

   inet6 fe80::5054:ff:fe85:1195/64 scope link

      valid_lft forever preferred_lft forever


在211.76上查看日志信息，看到已成功进入keepalived vip1组的主机模式，vip2组的备机模式。

LVS/NAT双主 + keepalived负载均衡实现_第2张图片

在211.77主机上查看信息

通过ip addr可以看到vip2 地址已经绑定在eth0和eth1网口上

[root@localhost ~]# ip addr

1: lo: mtu65536 qdisc noqueue state UNKNOWN

   link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00

   inet 127.0.0.1/8 scope host lo

   inet6 ::1/128 scope host

      valid_lft forever preferred_lft forever

2: eth0: mtu 1500 qdisc pfifo_fast state UP qlen1000

   link/ether 52:54:00:1b:a2:11 brd ff:ff:ff:ff:ff:ff

   inet 192.168.214.77/24 brd 192.168.214.255 scope global eth0

    inet 192.168.214.71/32 scope global eth0

   inet6 fe80::5054:ff:fe1b:a211/64 scope link

      valid_lft forever preferred_lft forever

3: eth1: mtu 1500 qdisc pfifo_fast state UP qlen1000

   link/ether 52:54:00:64:47:7d brd ff:ff:ff:ff:ff:ff

   inet 192.168.211.77/24 brd 192.168.211.255 scope global eth1

    inet192.168.211.253/32 scope global eth1

   inet6 fe80::5054:ff:fe64:477d/64 scope link

      valid_lft forever preferred_lft forever

 

在211.77上查看日志信息，看到已成功进入keepalived vip1组的备机模式，vip2组的主机模式。

LVS/NAT双主 + keepalived负载均衡实现_第3张图片

通过ipvsadm -L –n 查看相应lvs连接信息

211.76上查看

[root@localhost ~]# ipvsadm -L -n

IP Virtual Server version 1.2.1 (size=4096)

Prot LocalAddress:Port Scheduler Flags

 -> RemoteAddress:Port          Forward Weight ActiveConn InActConn

TCP 192.168.214.70:80 rr persistent 10

 -> 192.168.211.79:80           Masq    100    3         5        

TCP 192.168.214.71:8080 rr persistent 10

 -> 192.168.211.83:8080         Masq    100    0         0

 

211.77上查看

[root@localhost ~]# ipvsadm -L -n

IP Virtual Server version 1.2.1 (size=4096)

Prot LocalAddress:Port Scheduler Flags

 -> RemoteAddress:Port          Forward Weight ActiveConn InActConn

TCP 192.168.214.70:80 rr persistent 10

 -> 192.168.211.79:80           Masq    100    0         0        

TCP 192.168.214.71:8080 rr persistent 10

 -> 192.168.211.83:8080         Masq    100    3         4 


八、keepalived测试

使用vip1地址192.168.214.70访问后端web192.168.211.79的页面

LVS/NAT双主 + keepalived负载均衡实现_第4张图片

使用vip2地址192.168.214.71访问后端web192.168.211.83的页面

LVS/NAT双主 + keepalived负载均衡实现_第5张图片

正常访问没问题后，我们来模拟lvs集群故障，把前端lvs调度机主机211.76宕机，看211.77能否把vip1的地址漂移过来

通过ip addr命令查看发现vip1的地址都已经漂移过来了

[root@localhost ~]# ip addr

1: lo: mtu65536 qdisc noqueue state UNKNOWN

   link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00

   inet 127.0.0.1/8 scope host lo

   inet6 ::1/128 scope host

      valid_lft forever preferred_lft forever

2: eth0: mtu 1500 qdisc pfifo_fast state UP qlen1000

   link/ether 52:54:00:1b:a2:11 brd ff:ff:ff:ff:ff:ff

    inet 192.168.214.77/24 brd 192.168.214.255scope global eth0

   inet 192.168.214.71/32 scope global eth0

    inet 192.168.214.70/32scope global eth0

   inet6 fe80::5054:ff:fe1b:a211/64 scope link

      valid_lft forever preferred_lft forever

3: eth1: mtu 1500 qdisc pfifo_fast state UP qlen1000

   link/ether 52:54:00:64:47:7d brd ff:ff:ff:ff:ff:ff

   inet 192.168.211.77/24 brd 192.168.211.255 scope global eth1

   inet 192.168.211.253/32 scope global eth1

    inet192.168.211.254/32 scope global eth1

   inet6 fe80::5054:ff:fe64:477d/64 scope link

      valid_lft forever preferred_lft forever

 

再211.77上通过日志查看，看到vip1的地址已漂移了过来，变成了主机状态

May 15 14:38:26 localhostKeepalived_vrrp[11898]: VRRP_Instance(VI_1) Transitionto MASTER STATE

May 15 14:38:26 localhostKeepalived_vrrp[11898]: VRRP_Group(VG1) Syncing instances to MASTER state

May 15 14:38:26 localhostKeepalived_vrrp[11898]: VRRP_Instance(VI_GATEWAY1) Transitionto MASTER STATE

May 15 14:38:27 localhostKeepalived_vrrp[11898]: VRRP_Instance(VI_GATEWAY1)Entering MASTER STATE

May 15 14:38:27 localhostKeepalived_vrrp[11898]: VRRP_Instance(VI_GATEWAY1) setting protocol VIPs.

May 15 14:38:27 localhostKeepalived_vrrp[11898]: VRRP_Instance(VI_GATEWAY1) Sending gratuitous ARPs oneth1 for 192.168.211.254

May 15 14:38:27 localhostKeepalived_healthcheckers[11897]: Netlink reflector reports IP 192.168.211.254added

May 15 14:38:27 localhostKeepalived_vrrp[11898]: VRRP_Instance(VI_1) EnteringMASTER STATE

May 15 14:38:27 localhostKeepalived_vrrp[11898]: VRRP_Instance(VI_1) setting protocol VIPs.

May 15 14:38:27 localhostKeepalived_vrrp[11898]: VRRP_Instance(VI_1) Sending gratuitous ARPs on eth0 for192.168.214.70

May 15 14:38:27 localhostKeepalived_healthcheckers[11897]: Netlink reflector reports IP 192.168.214.70added

May 15 14:38:32 localhostKeepalived_vrrp[11898]: VRRP_Instance(VI_GATEWAY1) Sending gratuitous ARPs oneth1 for 192.168.211.254

May 15 14:38:32 localhostKeepalived_vrrp[11898]: VRRP_Instance(VI_1) Sending gratuitous ARPs on eth0 for192.168.214.70


如果想了解更多，请关注我们的公众号
公众号ID:opdevos
扫码关注

LVS/NAT双主 + keepalived负载均衡实现_第6张图片