How to Install and Configure K3S

Install and configure K3s, the lightweight CNCF-certified Kubernetes distribution, with its built-in local storage provider, Traefik ingress, and packaged components.

Kishore October 1, 2023 3 min read

AWX Automation Linux Containers Home Lab

K3s is a CNCF sandbox project that delivers a lightweight yet powerful certified Kubernetes distribution. K3S is a lightweight Kubernetes, easy to install, Simple but powerful “batteries-included” features have been added, such as:

local storage provider
service load balancer
helm controller
Traefik ingress controller

Install K3S on Rocky Linux 8

Prerequisites

Operating Systems:

If you are using RHEL/CentOS/Fedora, follow these steps.

Hardware:

Spec	Minumum	Recommended
CPU	1 Core	2 Cores
Memory	512 MB	1 GB

Installation:

K3s provides an installation script that is a convenient way to install it as a service on systemd based systems. This script is available at https://get.k3s.io.

Since i am using this installation method, will just run:

curl -sfL https://get.k3s.io | sh -

Login to Dev_Machine using root credentials and run the command
after few minutes the script will install the required packages and completes the installation
change the permissions of k3s config file

chmod 644 /etc/rancher/k3s/k3s.yaml

4. check the k3s service status using systemctl

5. check the status of nodes

5. check the kubectl verstion and status of Pods

Note

This is not complete Installation, configuration and Administration of Kubernetes. This just allows me to run AWX Operator

Watch the video on How to setup K3S kubernetes cluster

What K3s actually is

K3s is a CNCF-certified Kubernetes distribution packaged as a single binary of roughly 50 MB. It is not a fork of Kubernetes — it is upstream Kubernetes with the parts a single-node or edge deployment does not need removed or replaced:

etcd is optional. By default K3s uses an embedded SQLite datastore instead of etcd, which is what makes a single-node install trivial. For HA you switch to embedded etcd or an external datastore.
Ingress is built in. Traefik is deployed as part of the cluster, so a fresh install can serve HTTP/HTTPS traffic immediately.
Storage is built in. A local-path provisioner creates PersistentVolumes on the node’s disk without any extra CSI driver.
Helm is built in. A Helm controller watches HelmChart CRDs, so you can declaratively install charts by dropping a manifest in /var/lib/rancher/k3s/server/manifests/.

The trade-off is that you give up some flexibility: you cannot easily swap the datastore, ingress or storage provider, and the control plane is not highly available unless you configure it to be. For a home lab, an edge site, or a CI runner, that trade-off is usually worth it.

After the install

The kubeconfig lives at /etc/rancher/k3s/k3s.yaml and is owned by root. Either copy it to your user’s ~/.kube/config, or set KUBECONFIG=/etc/rancher/k3s/k3s.yaml. From there, kubectl get nodes and kubectl get pods -A are the two commands that tell you whether the install actually worked.

The install script also drops a token at /var/lib/rancher/k3s/server/node-token. You will need it if you later add worker nodes with K3S_URL=https://<server>:6443 K3S_TOKEN=<token> sh -.

Caveats

This is a single-node control plane. If the node dies, the cluster dies. For anything you care about, run three server nodes.
The get.k3s.io script pulls the latest stable release. Pin a version with INSTALL_K3S_VERSION=v1.x.y if you need reproducibility.
K3s defaults to the local-path storage class, which provisions hostPath volumes. That is fine for a lab and wrong for anything multi-node.

Back to list