well basically we use that on gcloud k8s in the future. but we still have a bare metal cluster. we only used ignition + vmware to provision it in the first plac

well basically we use that on gcloud k8s in the future. but we still have a bare metal cluster. we only used ignition + vmware to provision it in the first place and it was a breeze.. etcd was configured on the fly in a HA setup, k8s was made ready to bootstrap (just needed to run kubeadm init --config...) and we had basically a working cluster. highly available, with as many nodes we wanted, just copy the worker.yaml and edit what's needed. we actually play around with tectonic at the moment.