What happened:
I've bootstrapped a cluster with kubeadm on a banana pi, it is up and running. Upon restarting, some process is slow that kube-apiserver does not boot up, exits with last lines:

I1218 19:19:04.831034       1 log.go:172] http: TLS handshake error from 10.0.0.39:43582: EOF
I1218 19:19:04.867159       1 log.go:172] http: TLS handshake error from 10.0.0.39:43598: EOF
I1218 19:19:04.932248       1 log.go:172] http: TLS handshake error from 10.0.0.39:43552: EOF
I1218 19:19:05.169874       1 log.go:172] http: TLS handshake error from 10.0.0.39:43578: EOF
W1218 19:19:05.534620       1 storage_scheduling.go:95] unable to get PriorityClass system-node-critical: Get https://[::1]:6443/apis/scheduling.k8s.io/v1beta1/priorityclasses/system-node-critical: net/http: TLS handshake timeout. Retrying...

What you expected to happen:
Expected the kube-apiserver to start up.

How to reproduce it (as minimally and precisely as possible):
On a banana pi (A20) bootstrap a cluster with kubeadm, add kube-router cni, and then restart kubelet.

Anything else we need to know?:
The version below contains a proposed fix, that is why the GitVersion is dirty.

Environment:

• Kubernetes version
  Client Version: version.Info{Major:"1", Minor:"13", GitVersion:"v1.13.1", GitCommit:"eec55b9ba98609a46fee712359c7b5b365bdd920", GitTreeState:"clean", BuildDate:"2018-12-13T10:39:04Z", GoVersion:"go1.11.2", Compiler:"gc", Platform:"linux/arm"}
  Server Version: version.Info{Major:"1", Minor:"13+", GitVersion:"v1.13.2-beta.0.1+598a01989dcd06-dirty", GitCommit:"598a01989dcd06ec776248506fe6eb32f499bd38", GitTreeState:"dirty", BuildDate:"2018-12-18T19:10:37Z", GoVersion:"go1.11.2", Compiler:"gc", Platform:"linux/arm"}

• Cloud provider or hardware configuration:
  Banana pi m1 plus

• OS (e.g. from /etc/os-release):
  PRETTY_NAME="Debian GNU/Linux 9 (stretch)"
  NAME="Debian GNU/Linux"
  VERSION_ID="9"
  VERSION="9 (stretch)"
  ID=debian
  HOME_URL="https://www.debian.org/"
  SUPPORT_URL="https://www.debian.org/support"
  BUG_REPORT_URL="https://bugs.debian.org/"

• Kernel (e.g. uname -a):
  Linux k8s-master 4.9.0-8-armmp Unit test coverage in Kubelet is lousy. (~30%) #1 SMP Debian 4.9.130-2 (2018-10-27) armv7l GNU/Linux

• Install tools:
  kubeadm

• Others:
  Seems that in pkg/registry/scheduling/rest/storage_scheduling.go AddSystemPriorityClasses() adds a function which uses wait.Poll. On line 95 it logs 'Retrying...', howewer the return statement tells wait.Poll() to stop polling.

/kind bug