嗯,1.6.7 的老集群。早上反应数据库实例创建不出来。发现接口 port 的 tcp 不可达,查看 kube-proxy 的日志如下
1 2 3 4 5 6
[root@HB1-xxxxx-S03 kubernetes]# tailf kube-proxy.INFO E0708 09:44:03.437284 394350 proxier.go:1062] can't open "nodePort for cloudify/cloudae-cfy-nginx:port1" (:16081/tcp), skipping this nodePort: listen tcp :16081: bind: address already in use E0708 09:44:03.439285 394350 proxier.go:1062] can't open "nodePort for default/cloudae-app-grafana-svc:cloudae-app-grafana-svc" (:16101/tcp), skipping this nodePort: listen tcp :16101: bind: address already in use E0708 09:44:04.485556 394350 proxier.go:1062] can't open "nodePort for cloudify/cloudae-cfy-nginx:port2" (:16123/tcp), skipping this nodePort: listen tcp :16123: bind: address already in use E0708 09:44:04.487749 394350 proxier.go:1062] can't open "nodePort for cloudify/cloudae-cfy-influxdb:web" (:16083/tcp), skipping this nodePort: listen tcp :16083: bind: address already in use E0708 09:44:04.493137 394350 proxier.go:1062] can't open "nodePort for cloudify/cloudae-cfy-nginx:port1" (:16081/tcp), skipping this nodePort: listen tcp :16081: bind: address already in use
# TYPE node_sockstat_TCP_alloc gauge node_sockstat_TCP_alloc 33276 # HELP node_sockstat_TCP_inuse Number of TCP sockets in state inuse. # TYPE node_sockstat_TCP_inuse gauge node_sockstat_TCP_inuse 33059