Installation on a "bare metal" single Node

There are quite a lot of options how to install a kubernetes cluster.

Rancher

Since version 2.0 Rancher supports out of the box "production ready" Kubernetes cluster functionality. The setup is straight forward and is completly dockerized. Just deploy a recent rancher via docker, login to the web UI and from their deploy your kubernetes cluster. Under the hood a original not modified Kubernetes is running. You can either control it via Ranchers UI or with kubectl.

Kubeadm

kubeadm is a toolkit for easy installation of Kubernetes. Follow the Installation Instructions and the Usage Instructions.

For the Canal pod network use --pod-network-cidr=10.244.0.0/16

After initialization execute

kubectl apply -f https://raw.githubusercontent.com/projectcalico/canal/master/k8s-install/1.7/rbac.yaml
kubectl apply -f https://raw.githubusercontent.com/projectcalico/canal/master/k8s-install/1.7/canal.yaml

Troubleshooting:

After kube init the process blocks at This might take a minute or longer if the control plane images have to be pulled. Further investigations resulted in the etcd container not starting because of tcdmain: listen tcp xxx.xxx.xxx.xxx:xxxxx bind: cannot assign requested address. To solve this issue create a file kubeadm.yml:

apiVersion: kubeadm.k8s.io/v1alpha1
kind: MasterConfiguration
etcd:
  extraArgs:
    'listen-peer-urls': 'http://127.0.0.1:2380'

and start initialization again with kubeadm init --config kubeadm.yml. (see Github for details and origin). A previous attempt must be reverted with kubeadm reset.

Check the status with kubectl get pods --all-namespaces. If your canal pod is in the status 'CrashLoopBackOff' check the following:

• kubectl logs canal-zhzph --namespace=kube-system (adjust pod name)
• check docker ps -a for any exiting "flannel" containers and its log with docker logs CONTAINERID

Maybe you see messages linke Error from server (BadRequest): a container name must be specified for pod canal-zhzph, choose one of: [calico-node install-cni kube-flannel] or Error registering network: failed to acquire lease: node "HOSTNAME" pod cidr not assigned than edit /etc/kubernetes/manifests/kube-controller-manager.yaml and add to the command section

    - --allocate-node-cidrs=true
    - --cluster-cidr=10.244.0.0/16

Then restart kubelet with systemctl restart kubelet (original hint on Github

DNS warnings can be resolved by editing /etc/hosts

Minikube

Minikube is a tool to provision a Kubernetes cluster on a dev box by deploying a Linux virtual machine with e.g. VirtualBox or VMware.

Kubectl

Get logs of pod

kubectl logs POD --namespace=NAME

List all pods

kubectl get pods --all-namespaces

List all deployments

kubectl get deployment --all-namespaces

List all services

kubectl get svc --all-namespaces

List all roles

kubectl get clusterroles

Deploy Dashboard

Deploy dashboard

kubectl apply -f https://raw.githubusercontent.com/kubernetes/dashboard/master/src/deploy/recommended/kubernetes-dashboard.yaml

Start proxy

kubectl proxy [-p 8080]

Access dashboard http://localhost:8001/api/v1/namespaces/kube-system/services/https:kubernetes-dashboard:/proxy/ (adjust port if necessary)

Get login token

kubectl -n kube-system get secret # list service accounts
kubectl -n kube-system describe secret deployment-controller-token-****

Alternative grant dashboard admin rights: Create `dashboard-admin.yml' with following content:

apiVersion: rbac.authorization.k8s.io/v1beta1
kind: ClusterRoleBinding
metadata:
  name: kubernetes-dashboard
  labels:
    k8s-app: kubernetes-dashboard
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: cluster-admin
subjects:
- kind: ServiceAccount
  name: kubernetes-dashboard
  namespace: kube-system

Deploy to your cluster kubectl create -f dashboard-admin.yml. Then you skip authentication in the login screen

helm

Powerful "package manager" for Kubernetes. Install client for example with your system's package manager. With an already configured kubectl run helm init to install on your Kubernetes the server side component named tiller. Update helm's repo information with help repo update. Inspect a package with helm inspect.

When you encounter the error Error: release jenkins failed: namespaces "dev" is forbidden: User "system:serviceaccount:kube-system:default" cannot get namespaces in the namespace "dev" during installation of a package execute kubectl create clusterrolebinding add-on-cluster-admin --clusterrole=cluster-admin --serviceaccount=kube-system:default in order to add the role cluster-admin to the serviceaccount kube-system:default.

Create namespace

cat <<EOF | kubectl create -f -
{
  "kind": "Namespace",
  "apiVersion": "v1",
  "metadata": {
    "name": "dev",
    "labels": {
      "name": "dev"
    }
  }
}
EOF