unhold the binaries on the worker nodes during upgrade (#14708)

Author: RA489

content/en/docs/tasks/administer-cluster/kubeadm/kubeadm-upgrade-1-12.md

@@ -233,8 +233,10 @@ This page explains how to upgrade a Kubernetes cluster created with `kubeadm` fr
     {{< tabs name="k8s_upgrade" >}}
     {{% tab name="Ubuntu, Debian or HypriotOS" %}}
     # replace "x" with the latest patch version
+    apt-mark unhold kubelet kubeadm
     apt-get update
     apt-get upgrade -y kubelet=1.12.x-00 kubeadm=1.12.x-00
+    apt-mark hold kubelet kubeadm
     {{% /tab %}}
     {{% tab name="CentOS, RHEL or Fedora" %}}
     # replace "x" with the latest patch version
@@ -268,6 +270,7 @@ This page explains how to upgrade a Kubernetes cluster created with `kubeadm` fr
     kubectl uncordon $NODE
     ```
 
+
 1.  After the kubelet is upgraded on all nodes, verify that all nodes are available again by running the following command from anywhere kubectl can access the cluster:
 
     ```shell

content/en/docs/tasks/administer-cluster/kubeadm/kubeadm-upgrade-1-13.md

@@ -285,8 +285,10 @@ A possible workaround is described [here](https://github.com/kubernetes/kubeadm/
     {{< tabs name="k8s_kubelet_and_kubeadm" >}}
     {{% tab name="Ubuntu, Debian or HypriotOS" %}}
     # replace x in 1.13.x-00 with the latest patch version
+    apt-mark unhold kubelet kubeadm
     apt-get update
     apt-get install -y kubelet=1.13.x-00 kubeadm=1.13.x-00
+    apt-mark hold kubelet kubeadm
     {{% /tab %}}
     {{% tab name="CentOS, RHEL or Fedora" %}}
     # replace x in 1.13.x-0 with the latest patch version