CCE使用node节点使用VIP

背景：想在节点上使用VIP，将nodeport服务做到高可用。启动VIP后发现访问失败

部署

! Configuration File for keepalived
global_defs {
router_id master-node
}vrrp_instance VI_1 {state BACKUPinterface eth0mcast_src_ip 10.1.0.60virtual_router_id 51priority 100advert_int 1authentication {auth_type PASSauth_pass 1111}unicast_src_ip 10.1.0.60virtual_ipaddress {10.1.0.88/24}
}

! Configuration File for keepalived
global_defs {
router_id master-node
}vrrp_instance VI_1 {state MASTERinterface eth0mcast_src_ip 10.1.0.175virtual_router_id 51priority 101advert_int 1authentication {auth_type PASSauth_pass 1111}unicast_src_ip 10.1.0.175virtual_ipaddress {10.1.0.88}
}

[root@cce-node3-dev ~]# ip a show eth0
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000link/ether fa:16:3e:46:43:df brd ff:ff:ff:ff:ff:ffinet 10.1.0.60/24 brd 10.3.0.255 scope global dynamic noprefixroute eth0valid_lft 100122990sec preferred_lft 100122990secinet 10.1.0.88/32 scope global eth0valid_lft forever preferred_lft foreverinet6 fe80::f816:3eff:fe46:43df/64 scope link valid_lft forever preferred_lft forever

启动服务

-- 30007集群nodeport服务，8080 NGINX，任意一个节点启动# 测试结果如下
-- RIP
10.1.0.60:80 OK
10.1.0.60:30007 OK-- VIP
10.1.0.88:80 OK
10.1.0.88:30007 NOT OK

检查发现node节点并没有监听30007端口，而是通过ipvs添加了被负载均衡的虚拟地址。

[root@recovery-test-28141 ~]# ipvsadm -Ln | grep 30007 -A3
TCP  10.1.0.60:30007   rr-> 10.244.0.89:80               Masq        1      0