I run it on a dedicated machine (ryzen 3700x (8 cores, 16 threads) with 16G of ram and jbod
The idea behind it to have a quick way of creating disposable kubernetes clusters with full access to nodes to control and debug any situation imaginable
my setup has non-dhcp range (192.168.0.200-) assigned to the control-plane (master) and nodes, this way I get sequential ips, easy ssh access from both hw node and my laptop.
packer build debian12.pkg.hcl
takes just a little over 9 minutes to build an image
terraform init && terraform plan -out cluster && terraform apply cluster
not more than a minute to raise the machines (todo: remove grub timeout for speed)
ansible-playbook -i hosts kubernetes.yml
takes some time, though depends on your network speed and hardware
kubectl get nodes ; kubectl get po -A -o wide
you have a fully working cluster, now install helm and get wild