Documentation: System Requirements for additional nodes

Right now the system requirements added in https://github.com/replicatedhq/kURL/pull/14 (visible at https://kurl.sh/docs/install-with-kurl/system-requirements) suggest that a kURL node requires the following:

• 4 CPUs or equivalent per machine
• 8 GB of RAM per machine
• 30 GB of Disk Space per machine

My understanding is that this is the requirement for a single-node single-master cluster, where all the default kURL addons get stacked on the single node, including:

• prometheus
• grafana
• weave
• contour
• kotsadm stack
• k8s control plane
• rook

Many of these do not need to be replicated for each additional node, be it a master or a worker. It would be great to get an understanding of what the requirements are for both additional master nodes and for additional worker nodes. Kubeadm recommends 2/2:

• 2 GB or more of RAM per machine (any less will leave little room for your apps)
• 2 CPUs or more

But I'm guessing that at rest this might be more like 1/1, and that worker nodes would need even less.

Ask

Can we document the differences here? Right now it seems that every node in the cluster requires at least 8GB/4CPU which seems like a very high overhead given what's running.

I think some statement about what is actually used by a kURL stack and what is recommended to leave room for applications is warranted as well.

Start with the linked system-requirements page and compare its per-node requirements with the listed kURL addons and Kubernetes control-plane components. Document separate guidance for additional master and worker nodes, distinguishing resource usage from recommended application headroom; done means the page clearly explains these differences.