Ubuntu with multipass and jq installed.

sudo apt install jq
snap install multipass

Create at least 2 VMs with multipass using the following commands.

multipass launch --cpus 2 --mem 2G --disk 10G --cloud-init server-cloud-config.yaml --name server-node &
multipass launch --cpus 2 --mem 2G --disk 10G --cloud-init agent-cloud-config.yaml --name agent-node1 &
...
multipass launch --cpus 2 --mem 2G --disk 10G --cloud-init agent-cloud-config.yaml --name agent-node9 &

You may change the options to better fit your host system and use case. Remove the & if you don't want to run them in parallel.

The cloud-init yaml file will provision all VMs with Docker and k3s on the server node.

Since agent nodes need information from the server node we need to manually install k3s on them.

1. First we need to get the server node IP and a join token. This is the information we need to use when installing k3s.

   The commands below will export this information into a local file.

    echo "export server_ip=$(multipass info server-node --format json | jq -r '.info[].ipv4[0]')" > env
    echo "export server_token=$(multipass exec server-node -- sudo cat /var/lib/rancher/k3s/server/node-token)" >> env
   

2. Copy this file to the agent nodes.

    multipass transfer env agent-node1:env
    ...
    multipass transfer env agent-node9:env
    rm env
   

3. Now we need to open a shell to the node itself.

    multipass shell agent-node1
   

4. Finally we can run the commands needed to actually install k3s on the node and have it join the cluster.

   The first command puts the exported information into the shell's environment, then we use it in the installation command.

    eval $(cat env)
    curl -sfL https://get.k3s.io | K3S_URL=https://${server_ip}:6443 K3S_TOKEN=${server_token} sh -s - --docker
    rm env
    exit
   

Now repeat steps 3 and 4 for all agent nodes.

When you have installed k3s on all nodes we should check the status of the cluster.

multipass exec server-node -- sudo kubectl get nodes

NAME          STATUS   ROLES                  AGE     VERSION
server-node   Ready    control-plane,master   2m16s   v1.20.5+k3s1
agent-node1   Ready    <none>                 97s     v1.20.5+k3s1
agent-node2   Ready    <none>                 57s     v1.20.5+k3s1
...

If all nodes have status Ready all is well.

If anything goes wrong you can always just delete the VM and start from scratch. Do this with the following commands.

multipass delete <name>
multipass purge

Then just start over from the creation of the node.

• https://multipass.run/
• https://cloudinit.readthedocs.io/en/latest/
• https://rancher.com/docs/k3s/latest/en/