Kubeadm Ansible Playbook

Build a Kubernetes cluster using Ansible with kubeadm. The goal is easily install a Kubernetes cluster on machines running:

Ubuntu 16.04
CentOS 7
Debian 9

System requirements:

Deployment environment must have Ansible 2.4.0+
Master and nodes must have passwordless SSH access

Usage

Add the system information gathered above into a file called hosts.ini. For example:

[master]
192.16.35.12

[node]
192.16.35.[10:11]

[kube-cluster:children]
master
node

If you're working with ubuntu, add the following properties to each host ansible_python_interpreter='python3':

[master]
192.16.35.12 ansible_python_interpreter='python3'

[node]
192.16.35.[10:11] ansible_python_interpreter='python3'

[kube-cluster:children]
master
node

Before continuing, edit group_vars/all.yml to your specified configuration.

For example, I choose to run flannel instead of calico, and thus:

# Network implementation('flannel', 'calico')
network: flannel

Note: Depending on your setup, you may need to modify cni_opts to an available network interface. By default, kubeadm-ansible uses eth1. Your default interface may be eth0.

After going through the setup, run the site.yaml playbook:

$ ansible-playbook site.yaml
...
==> master1: TASK [addon : Create Kubernetes dashboard deployment] **************************
==> master1: changed: [192.16.35.12 -> 192.16.35.12]
==> master1:
==> master1: PLAY RECAP *********************************************************************
==> master1: 192.16.35.10               : ok=18   changed=14   unreachable=0    failed=0
==> master1: 192.16.35.11               : ok=18   changed=14   unreachable=0    failed=0
==> master1: 192.16.35.12               : ok=34   changed=29   unreachable=0    failed=0

The playbook will download /etc/kubernetes/admin.conf file to $HOME/admin.conf.

If it doesn't work download the admin.conf from the master node:

$ scp k8s@k8s-master:/etc/kubernetes/admin.conf .

Verify cluster is fully running using kubectl:

$ export KUBECONFIG=~/admin.conf
$ kubectl get node
NAME      STATUS    AGE       VERSION
master1   Ready     22m       v1.6.3
node1     Ready     20m       v1.6.3
node2     Ready     20m       v1.6.3

$ kubectl get po -n kube-system
NAME                                    READY     STATUS    RESTARTS   AGE
etcd-master1                            1/1       Running   0          23m
...

Resetting the environment

Finally, reset all kubeadm installed state using reset-site.yaml playbook:

$ ansible-playbook reset-site.yaml

Additional features

These are features that you could want to install to make your life easier.

Enable/disable these features in group_vars/all.yml (all disabled by default):

# Additional feature to install
additional_features:
  helm: false
  metallb: false
  healthcheck: false

Helm

This will install helm in your cluster (https://helm.sh/) so you can deploy charts.

MetalLB

This will install MetalLB (https://metallb.universe.tf/), very useful if you deploy the cluster locally and you need a load balancer to access the services.

Healthcheck

This will install k8s-healthcheck (https://github.com/emrekenci/k8s-healthcheck), a small application to report cluster status.

Utils

Collection of scripts/utilities

Vagrantfile

This Vagrantfile is taken from https://github.com/ecomm-integration-ballerina/kubernetes-cluster and slightly modified to copy ssh keys inside the cluster (install https://github.com/dotless-de/vagrant-vbguest is highly recommended)

Tips & Tricks

Specify user for Ansible

If you use vagrant or your remote user is root, add this to hosts.ini

[master]
192.16.35.12 ansible_user='root'

[node]
192.16.35.[10:11] ansible_user='root'

Access Kubernetes Dashboard

As of release 1.7 Dashboard no longer has full admin privileges granted by default, so you need to create a token to access the resources:

$ kubectl -n kube-system create sa dashboard
$ kubectl create clusterrolebinding dashboard --clusterrole cluster-admin --serviceaccount=kube-system:dashboard
$ kubectl -n kube-system get sa dashboard -o yaml
apiVersion: v1
kind: ServiceAccount
metadata:
  creationTimestamp: 2017-11-27T17:06:41Z
  name: dashboard
  namespace: kube-system
  resourceVersion: "69076"
  selfLink: /api/v1/namespaces/kube-system/serviceaccounts/dashboard
  uid: 56b880bf-d395-11e7-9528-448a5ba4bd34
secrets:
- name: dashboard-token-vg52j

$ kubectl -n kube-system describe secrets dashboard-token-vg52j
...
token:      eyJhbGciOiJSUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJrdWJlcm5ldGVzL3NlcnZpY2VhY2NvdW50Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9uYW1lc3BhY2UiOiJrdWJlLXN5c3RlbSIsImt1YmVybmV0ZXMuaW8vc2VydmljZWFjY291bnQvc2VjcmV0Lm5hbWUiOiJkYXNoYm9hcmQtdG9rZW4tdmc1MmoiLCJrdWJlcm5ldGVzLmlvL3NlcnZpY2VhY2NvdW50L3NlcnZpY2UtYWNjb3VudC5uYW1lIjoiZGFzaGJvYXJkIiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9zZXJ2aWNlLWFjY291bnQudWlkIjoiNTZiODgwYmYtZDM5NS0xMWU3LTk1MjgtNDQ4YTViYTRiZDM0Iiwic3ViIjoic3lzdGVtOnNlcnZpY2VhY2NvdW50Omt1YmUtc3lzdGVtOmRhc2hib2FyZCJ9.bVRECfNS4NDmWAFWxGbAi1n9SfQ-TMNafPtF70pbp9Kun9RbC3BNR5NjTEuKjwt8nqZ6k3r09UKJ4dpo2lHtr2RTNAfEsoEGtoMlW8X9lg70ccPB0M1KJiz3c7-gpDUaQRIMNwz42db7Q1dN7HLieD6I4lFsHgk9NPUIVKqJ0p6PNTp99pBwvpvnKX72NIiIvgRwC2cnFr3R6WdUEsuVfuWGdF-jXyc6lS7_kOiXp2yh6Ym_YYIr3SsjYK7XUIPHrBqWjF-KXO_AL3J8J_UebtWSGomYvuXXbbAUefbOK4qopqQ6FzRXQs00KrKa8sfqrKMm_x71Kyqq6RbFECsHPA

$ kubectl proxy

Copy and paste the token from above to dashboard.

jonny-levi/kubeadm-ansible