This playbook will build an HA Kubernetes cluster on Flatcar hosts with k3s, kube-vip via ansible.
This project tries to be unoppiniated so it just install the bare-minimum required for having an HA cluster that can handle your kubectl commands. This means that an installation produced from this project will lack important things like a CNI, an IngressController and a LoadBalancer, and yes the ones included with k3s are disabled.
This is based on the work from this fork which is based on this other fork ] which is based k3s-io/k3s-ansible. It uses kube-vip to create a load balancer for control plane.
Build a Kubernetes cluster using Ansible with k3s. The goal is easily install a HA Kubernetes cluster on machines running:
- Flatcar
- Fedora CoreOS (not tested)
on processor architecture:
- x64
- arm64
- Deployment environment must have Ansible 2.4.0+. If you need a quick primer on Ansible you can check out TechnoTim's docs and setting up Ansible.
serverandagentnodes should have passwordless SSH access, if not you can supply arguments to provide credentialsñg-ask-pass --ask-become-passto ach command.
First create a new directory based on the sample directory within the inventory directory:
cp -R inventory/sample inventory/my-clusterSecond, edit inventory/my-cluster/hosts.ini to match the system information gathered above.
For example:
[master]
192.168.30.38
192.168.30.39
192.168.30.40
[node]
192.168.30.41
192.168.30.42
[k3s_cluster:children]
master
nodeIf multiple hosts are in the master group, the playbook will automatically set up k3s in HA mode with etcd.
This requires at least k3s version 1.19.1 however the version is configurable by using the k3s_version variable.
If needed, you can also edit inventory/my-cluster/group_vars/all.yml to match your environment.
For most of it operations Ansible requires Python installed in the remote host, this is an issue as Flatcar does not include Python. Fortunately we can still use the parts of ansible that do not require Python to install pypy (a minimal python distribution) in the Flatcar nodes.
ansible-playbook bootstrap-python.yml -i inventory/my-cluster/hosts.iniStart provisioning of the cluster using the following command:
ansible-playbook site.yml -i inventory/my-cluster/hosts.iniAfter deployment control plane will be accessible via virtual ip-address which is defined in inventory/group_vars/all.yml as apiserver_endpoint
Rerunning this playbook will have no effect in the cluster if no variables or template files are modified.
- Joining new nodes/masters. Simply add them to the hosts.ini and rerun (
⚠️ This operations causes the restart of the k3s daemon and it may produce downtime). - Modifying the systemd service (including the process arguments) and k3s manifests. Edit them/the variable they use and rerun.
- Upgrading k3s. This is best done via the Rancher’s system-upgrade-controller.
- Removing nodes. This is best done manually or with an specialized tool. Removing a node from the inventory will just case that future runs of the playbook wodn't know of the node existance.
Be aware that the ability of removing the cluster has low priority in this project and it's highly advised to reinstall the flatcar hosts instead.
ansible-playbook reset.yml -i inventory/my-cluster/hosts.iniYou should also reboot these nodes due to the VIP not being destroyed
To copy your kube config locally so that you can access your Kubernetes cluster run:
scp debian@master_ip:~/.kube/config ~/.kube/configSee the commands here.
You may want to kickstart your k3s cluster by using Vagrant to quickly build you all needed VMs with one command.
Head to the vagrant subfolder and type vagrant up to get your environment setup.
After the VMs have got build, deploy k3s using the Ansible playbook site.yml by the
vagrant provision --provision-with ansible command.
This repo is really standing on the shoulders of giants. To all those who have contributed.
Thanks to these repos for code and ideas: