Skip to content

Running k0s in Docker#

We publish a k0s container image with every release. By default, we run both controller and worker in the same container to provide an easy local testing "cluster".

The containers are published both on Docker Hub and GitHub. The examples in this page show Docker Hub, because it's more simple to use. Using GitHub requires a separate authentication (not covered here). Alternative links:


You can run your own k0s-in-docker easily with:

docker run -d --name k0s --hostname k0s --privileged -v /var/lib/k0s -p 6443:6443
Just grab the kubeconfig file with docker exec k0s cat /var/lib/k0s/pki/admin.conf and paste e.g. into Lens.

Running workers#

If you want to attach multiple workers nodes into the cluster you can run separate containers for each worker.

First, we need a join token for the worker:

token=$(docker exec -t -i k0s k0s token create --role=worker)

Then join a new worker by running the container with:

docker run -d --name k0s-worker1 --hostname k0s-worker1 --privileged -v /var/lib/k0s k0s worker $token

Repeat for as many workers you need, and have resources for. :)

Docker Compose#

You can also run k0s with Docker Compose:

version: "3.9"
    container_name: k0s
    command: k0s controller --config=/etc/k0s/config.yaml --enable-worker
    hostname: k0s
    privileged: true
      - "/var/lib/k0s"
      - /run
      - /var/run
      - "6443:6443"
    network_mode: "bridge"
      K0S_CONFIG: |-
        kind: Cluster
          name: k0s
        # Any additional configuration goes here ...

Known limitations#

No custom Docker networks#

Currently, we cannot run k0s nodes if the containers are configured to use custom networks e.g. with --net my-net. This is caused by the fact that Docker sets up a custom DNS service within the network and that messes up CoreDNS. We know that there are some workarounds possible, but they are bit hackish. And on the other hand, running k0s cluster(s) in bridge network should not cause issues.