[svc][op]LVS+keepalived

lvs是一种负载均衡技术.注意区分负载均衡和高可用的区别.

keepalive是lvs的管理工具

ipvsadm也是lvs的管理工具

keepalive借助ipvsadm管理lvs.所以通常说lvs+keepalive

lvs支持多种负载方式.

理论及配置说明:

http://www.cnblogs.com/edisonchou/p/4281978.html

lvs:DR模式wrr负载:

拓扑

　　　　KA1    LAMP-1

pc------|

　　　　KA2　 LAMP-2

VIP:192.168.14.100

KA1:192.168.14.134

KA1:192.168.14.135

LAMP-1:192.168.141

LAMP-2:192.168.142

安装软件:

lvs:

yum install ipvsadm -y
rpm -qa ipvsadm

KA:

yum install keepalived -y

lamp:配置

RS绑定VIP：
ip addr add 192.168.14.100/ dev lo label lo:
route add -host 192.168.14.100 dev lo
抑制ARP：
echo "" >/proc/sys/net/ipv4/conf/lo/arp_ignore
echo "" >/proc/sys/net/ipv4/conf/lo/arp_announce
echo "" >/proc/sys/net/ipv4/conf/all/arp_ignore
echo "" >/proc/sys/net/ipv4/conf/all/arp_announce

前期准备:

KA的配置文件:主KA

/etc/keepalived/keepalived.conf

! Configuration File for keepalived

global_defs {
   router_id LVS_19
   vrrp_mcast_group4 224.0.1.18
}

vrrp_instance VI_1 {
    state MASTER
    interface eth0
    virtual_router_id
    priority
    advert_int
    authentication {
        auth_type PASS
        auth_pass
    }
    virtual_ipaddress {
      192.168.14.100/ dev eth0 label eth0:
    }
}

#ipvsadm -A -t 192.168.14.100: -s rr -p
virtual_server 192.168.14.100  {
    delay_loop
    lb_algo wrr
    lb_kind DR
    nat_mask 255.255.255.0
    persistence_timeout
    protocol TCP

#ipvsadm -a -t 192.168.14.100: -r 10.0.1.6: -g
    real_server 192.168.14.141  {
        weight
        TCP_CHECK {
        connect_timeout
        nb_get_retry
        delay_before_retry
        connect_port
        }
    }

    real_server 192.168.14.142  {
        weight
        TCP_CHECK {
        connect_timeout
        nb_get_retry
        delay_before_retry
        connect_port
        }
    }
}

备:KA:

! Configuration File for keepalived

global_defs {
   router_id LVS_19
   vrrp_mcast_group4 224.0.1.18
}

vrrp_instance VI_1 {
    state BACKUP
    interface eth0
    virtual_router_id
    priority
    advert_int
    authentication {
        auth_type PASS
        auth_pass
    }
    virtual_ipaddress {
      192.168.14.100/ dev eth0 label eth0:
    }
}

#ipvsadm -A -t 192.168.14.100: -s rr -p
virtual_server 192.168.14.100  {
    delay_loop
    lb_algo wrr
    lb_kind DR
    nat_mask 255.255.255.0
    persistence_timeout
    protocol TCP

#ipvsadm -a -t 192.168.14.100: -r 10.0.1.6: -g
    real_server 192.168.14.141  {
        weight
        TCP_CHECK {
        connect_timeout
        nb_get_retry
        delay_before_retry
        connect_port
        }
    }

    real_server 192.168.14.142  {
        weight
        TCP_CHECK {
        connect_timeout
        nb_get_retry
        delay_before_retry
        connect_port
        }
    }
}

检查:如果一台down了,另一台通过ifconfig就会出现绑定的VIP了

ip add |egrep  "134|135"

理论笔记:

集群--一组服务器
lvs harproxy nginx,keepalive heartbeat
负载均衡 高可用

负载模型:
nginx 7层负载
lvs 4层负载
lvs---nginx---webserver

负载:
1,7*24h服务
2,一台机器扛不住

lvs底层原理底层原理:
ipvs
ipvsadm自带
keepalive

名词介绍:
CIP(Client ip address)
DIP(Director IP address)
VIP(virtual ip address)
RIP(Real Server Ip address)=RS

负载的3种模式:
NAT
DR(direct route)
TUN(ip tunnel)
fullnat

DR
RIP
1,绑定VIP
2,抑制arp回应
主要:
LB通过修改目的mac,因此无法改写端口
2000W pv
请求10000一下都可以用haproxy/nginx

NAT模式:
LB要打开转发:
net.ipv4_forward=1