/hetzner-kube

A CLI tool for provisioning kubernetes clusters on Hetzner Cloud

Primary LanguageGoApache License 2.0Apache-2.0

Build Status Go Report Card Maintainability Gitter chat

hetzner-kube: fast and easy setup of kubernetes clusters on Hetzner Cloud

This project contains a CLI tool to easily provision kubernetes clusters on Hetzner Cloud.

This is my very first tool written in Go.

How to install

Binary releases

Get the binary from releases page

From source

hetzner-kube is written in Go. To install Go please follow the instructions on its homepage.

To get and build hetzner-kube from source run this command:

$ go get -u github.com/xetys/hetzner-kube

The project source will now be in your $GOPATH directory ($GOPATH/src/github.com/xetys/hetzner-kube) and the binary will be in $GOPATH/bin.

If you want to build it yourself later, you can change into the source directory and run go build or go install.

Code completion

BASH

To load completion run

source <(hetzner-kube completion bash)

To configure your bash shell to load completions for each session add to your "~/.bashrc" file

# ~/.bashrc or ~/.profile
echo 'source <(hetzner-kube completion bash)\n' >> ~/.bashrc

Or you can add it to your bash_completition.d folder:

# On linux
hetzner-kube completion bash > /etc/bash_completion.d/hetzner-kube
# On OSX with completion installed via brew (`brew install bash-completion`)
hetzner-kube completion bash > /usr/local/etc/bash_completion.d/hetzner-kube

ZSH

To configure your zsh shell to load completions run following commands:

# On linux
hetzner-kube completion zsh | sudo tee /usr/share/zsh/vendor-completions/_hetzner-kube
# On OSX
hetzner-kube completion zsh | sudo tee /usr/share/zsh/site-functions/_hetzner-kube

Than rebuild autocomplete function with:

compinits

Usage

In your Hetzner Console generate an API token and

$ hetzner-kube context add my-project
Token: <PASTE-TOKEN-HERE>

Then you need to add an SSH key:

$ hetzner-kube ssh-key add -n my-key

This assumes, you already have a SSH keypair ~/.ssh/id_rsa and ~/.ssh/id_rsa.pub

And finally, you can create a cluster by running:

$ hetzner-kube cluster create --name my-cluster --ssh-key my-key

This will provision a brand new kubernetes cluster in latest version!

To access the cluster via kubectl, create a config file:

# Write ~/.kube/config (you'll be asked to overwrite if file already exists)
hetzner-kube cluster kubeconfig my-cluster
# Alternatively, create a separate file and point kubectl to it:
#hetzner-kube cluster kubeconfig --print my-cluster > ~/.kube/config-my-cluster
#export KUBECONFIG=~/.kube/config-my-cluster

For a full list of options that can be passed to the cluster create command, see the Cluster Create Guide for more information.

HA-clusters

You can build high available clusters with hetzner-kube. Read the High availability Guide for further information.

Custom Options

addons

You can install some addons to your cluster using the cluster addon sub-command. Get a list of addons using:

$ hetzner-kube cluster addon list

contributing new addons

You want to add some cool stuff to hetzner-kube? It's quite easy! Learn how to add new addons in the Developing Addons documentation.

backup & restore (for HA-clusters)

If you are running an external etcd cluster, you can use the etcd management to backup the etcd cluster using

$ hetzner-kube cluster etcd backup my-cluster --snapshot-name my-snapshot

(--snapshot-name is optional)

and restore it using

$ hetzner-kube cluster etcd restore my-cluster --snapshot-name my-snapshot

If you place a different snapshot (with .db file extension) in /root/etcd-snapshots of the first etcd node, you can use the restore command for migration of kubernetes clusters.

phases

After a cluster is once created (nodes creation is enough), you can perform the steps from hetzner-kube cluster create separately.

You can use

$ hetzner-kube cluster phase -h

to only list all available phases.

Cluster creation is a chain of these phases:

  • provision
  • network-setup
  • etcd (if HA)
  • install-masters
  • setup-ha (if HA)
  • install-workers

Some of theses phase have additional options, to run the actual phase differently from the usual cluster creation.

examples for phases

To simply run a phase from cluster creation, you cam run a phase, eg. the etcd phase using:

$ hetzner-kube cluster phase etcd my-cluster

In order to upgrade or migrate a cluster, you might want to keep the etcd data, and run

$ hetzner-kube cluster phase etcd my-cluster --keep-data

and preserve the existing certificates using

$ hetzner-kube cluster phase install-master my-cluster --keep-all-certs

The latter command can be also useful during cluster migration, if you place the existing certs in /etc/kubernetes/pki before running the install-masters phase.

cloud-init

If you like to run some scripts or install some additional packages while provisioning new servers, you can use cloud-init

$ hetzner-kube cluster create --name my-cluster --nodes 3 --ssh-key my-key --cloud-init <PATH-TO-FILE>

An example file to make all nodes ansible ready. The comment on the first line is important:

#cloud-config
package_update: true
packages:
 - python

Full tutorial

This article guides through a full cluster setup.