Kubernetes on Proxmox: Provisioning Control Plane and Worker Nodes

Overview

With our base VM template complete, the next step is provisioning the actual virtual machines that will form our Kubernetes cluster.

In this post, we’ll focus on the infrastructure side of Kubernetes rather than Kubernetes itself. The goal is to end with a set of clean, reachable Linux nodes that are ready for kubeadm in the next post.

For this setup, we’ll provision:

• 1 control plane node
• 2 worker nodes

This is a practical homelab topology that focuses on learning Kubernetes features while keeping resource requirements reasonable. For production environments, you'd typically run 3+ control plane nodes for high availability.

Why Separate Provisioning from Bootstrap?

It’s tempting to jump straight into Kubernetes installation, but separating these steps has real benefits:

• Clear separation between infrastructure and cluster logic
• Easier troubleshooting when something goes wrong
• Faster rebuilds when experimenting or iterating
• A more production-like mental model

At the end of this post, Kubernetes is not installed yet — and that’s intentional.

Node Layout and Sizing

Control Plane Node

The control plane node is responsible for:

• API server
• Scheduler
• Controller manager
• etcd

Recommended specs:

• Count: 1
• CPU: 2 vCPUs
• Memory: 4 GB RAM
• Disk: 40-60 GB

Worker Nodes

Worker nodes run application workloads.

Recommended specs:

• Count: 2
• CPU: 2-4 vCPUs
• Memory: 4-6 GB RAM
• Disk: 40-60 GB

Cloning the VM Template

All nodes will be created by cloning the base template built in the previous post.

Naming Conventions

Using predictable names makes debugging and automation much easier.

Control plane node:

• k8s-cp-1

Worker nodes:

• k8s-w-1
• k8s-w-2

Clone Steps (Proxmox UI)

For each node:

1. Right-click the base VM template
2. Select Clone
3. Choose Full Clone
4. Set the VM name (from the list above)
5. Select the appropriate target node and storage
6. Complete the clone

Repeat until all three VMs are created (1 control plane + 2 workers).

Adjusting VM Resources

After cloning, update each VM’s resources based on its role.

Control Plane VM

• Set CPU to 2 cores
• Set memory to 4 GB
• Leave disk size unchanged

Worker VMs

• Set CPU to 2-4 cores (depending on your available resources)
• Set memory to 4-6 GB
• Leave disk size unchanged

SSH Access

sudo ssh-keygen -A
sudo systemctl restart ssh

Networking Configuration

All nodes are attached to the Kubernetes VLAN (vmbr30).

IP Addressing Strategy

For now, we’ll rely on DHCP to assign IP addresses:

• Simplifies initial setup
• Easy to change later to static or reservations
• Works well in homelab environments

Once the VMs boot, verify that each node receives an IP address in the expected subnet.

ip a

Take note of each node’s IP — you’ll need them for SSH access.

Initial Node Verification

Before moving on, verify that every node is reachable and healthy.

SSH Access

From your workstation:

ssh k8sadmin@<node-ip>

Confirm:

• Login works
• Hostnames are correct
• Network connectivity is stable

Hostname Validation

On each node:

hostnamectl

If needed, set the hostname manually:

Replacing the k8s-cp-1 with each name respectively per virtual machine being configured.

sudo hostnamectl set-hostname k8s-cp-1

(Reboot after changing the hostname.)

Disk Visibility (Important for Longhorn Later)

Before installing Kubernetes, ensure worker nodes see their disks correctly.

On worker nodes:

lsblk

You should see:

• Root disk
• Sufficient free space for future storage workloads

We’ll rely on this later when configuring Longhorn.

Baseline Readiness Checklist

At this point, you should have:

• 1 reachable control plane VM
• 2 reachable worker VMs
• SSH access to all nodes
• Correct hostnames set
• Networking verified

No Kubernetes components should be installed yet.

What’s Next

With the infrastructure in place, the next step is bootstrapping Kubernetes itself.

In the next post, we’ll:

• Install Kubernetes components
• Initialize the control plane with kubeadm
• Join worker nodes
• Install Calico for cluster networking

➡️ Next: Kubernetes on Proxmox – Bootstrapping the Cluster with kubeadm