高可用k8s集群搭建：部署负载均衡器

三台机器都安装keepalived

sudo yum install -y ipvsadm keepalived

配置keepalived

三个 master 节点使用 MASTER-BACKUP-BACKUP 模式，192.168.17.101 为 主 master。
修改 192.168.17.101 上 /etc/keepalived/keepalived.conf 内容如下

global_defs {
   router_id LVS_k8s
}

vrrp_script CheckK8sMaster {
    script "curl -k https://127.0.0.1:6443/api"
    interval 3
    weight -10
    fall 2
    rise 2
}

vrrp_instance VI_1 {
    virtual_router_id 66
    advert_int 1
    state MASTER ## 状态
    priority 100 ## 优先级
    interface eth0
    mcast_src_ip 192.168.17.101 ##这里需要改成当前机器ip
    authentication {
        auth_type PASS
        auth_pass 4743
    }
    unicast_peer {
        192.168.17.101
        192.168.17.102
        192.168.17.103
    }
    virtual_ipaddress {
        192.168.17.100 ## vip地址，不需要自己创建ip了，keepalived会自动帮我们创建
    }
    track_script {
        CheckK8sMaster
    }
}

virtual_server 192.168.17.100 6443 {
    lb_algo rr
    lb_kind DR
    persistence_timeout 0
    delay_loop 20
    protocol TCP

    real_server 192.168.17.101 6443 {
        weight 10
        TCP_CHECK {
            connect_timeout 10
        }
    }

    real_server 192.168.17.102 6443 {
        weight 10
        TCP_CHECK {
            connect_timeout 10
        }
    }

    real_server 192.168.17.103 6443 {
        weight 10
        TCP_CHECK {
            connect_timeout 10
        }
    }
}

另外两台节点配置与第一台基本相同，仅需要修改以下几处：

• state 改为 BACKUP
• priority 改为比100小的数
• mcast_src_ip 改为当前节点ip
  修改内核参数，将以下内容追加到 /etc/sysctl.conf 文件中并执行sudo sysctl -p

  net.ipv4.conf.lo.arp_ignore = 1
  net.ipv4.conf.lo.arp_announce = 2
  net.ipv4.conf.all.arp_ignore = 1
  net.ipv4.conf.all.arp_announce = 2
  net.ipv4.ip_forward = 1
  net.ipv4.nf_conntrack_max = 2048000
  net.netfilter.nf_conntrack_max = 2048000

  启动 keepalived 并配置开机启动

  systemctl start keepalived
  systemctl enable keepalived

  验证

• 在三台机器上分别执行kubectl get no均工作正常
• 重启主 master 节点 162.168.17.101，在其他两台机器上查看节点状态，发现 162.168.17.101 很快变为 NotReady ，稍后又变为 Ready