k0s — Zero Friction Kubernetes Distribution

What it is

k0s is a Kubernetes distribution by Mirantis that ships as a single binary with etcd, containerd, and CNI networking included. There is nothing to install separately. It runs on any Linux host, from a Raspberry Pi edge node to a multi-node production cluster with high availability. k0s is CNCF certified and supports FIPS-validated deployments.

k0s is designed for platform teams, edge computing deployments, and anyone who wants a minimal-footprint Kubernetes cluster without the complexity of kubeadm or managed services.

How it saves time or tokens

Setting up Kubernetes with kubeadm requires installing container runtimes, configuring etcd, setting up networking, and managing certificates separately. k0s eliminates all of that. A single curl | sh installs everything, and k0s install controller --single gives you a working cluster in under a minute. For edge deployments where resources are limited, k0s's minimal footprint means you can run Kubernetes on hardware that would struggle with other distributions.

How to use

1. Install k0s on the first control plane node:

curl -sSLf https://get.k0s.sh | sudo sh
sudo k0s install controller --single
sudo k0s start

1. Get the kubeconfig:

sudo k0s kubeconfig admin > ~/.kube/config
kubectl get nodes

1. For a multi-node cluster, generate a join token and add workers:

sudo k0s token create --role worker > worker-token
# On worker node:
sudo k0s install worker --token-file worker-token
sudo k0s start

Example

Deploying a simple application on k0s:

# After k0s is running
kubectl create deployment nginx --image=nginx:alpine --replicas=3
kubectl expose deployment nginx --port=80 --type=NodePort
kubectl get svc nginx
# Access via NodePort on any cluster node

k0s works with standard kubectl and all Kubernetes resources. There is no k0s-specific API to learn.

Related on TokRepo

• DevOps tools — Browse Kubernetes and infrastructure tools
• Self-hosted tools — Explore self-hosted platform solutions

Common pitfalls

• Using --single mode in production. The single-node mode is convenient for development but provides no high availability. Production clusters should use separate controller and worker nodes.
• Not opening required firewall ports between nodes. k0s needs port 6443 (API), 4789 (VXLAN), and 10250 (kubelet). Blocked ports cause nodes to fail joining.
• Forgetting that k0s manages its own containerd instance. If you already have containerd installed, k0s uses its bundled version, not your system installation. This avoids conflicts but can be confusing.
• Starting with an overly complex configuration instead of defaults. Begin with the minimal setup, verify it works, then customize incrementally. This approach catches configuration errors early and keeps troubleshooting straightforward.