Kubectl exec exits unexpectedly

Deployment: single instance deploy, version 2.0.8
kubectl config file is downloaded from rancher ui.
I exec into the pod using "kubectl exec -it /bin/sh -n ". If commands were typed continually, everything went well. But if I hang for about 10 secs then typed enter, kubectl was shut down.
If tracing tail of rancher server container log, we found this:
E1025 05:50:46.299072 7 upgradeaware.go:310] Error proxying data from client to backend: EOF

Looks communication error between rancher server and kubenetes api server. Any suggestion?

thanks.

Looks like restart of rancher server resolves the issue. But still trying don’t know what is causing it.