Cloud-init configurations for spinning up Kubernetes clusters with Calico networking on Proxmox. Because manually configuring k8s nodes gets old fast.

Pre-configured cloud-init YAML files that turn a fresh Ubuntu VM into a fully-configured Kubernetes node in minutes. Perfect for lab environments where you need to deploy/destroy clusters frequently without the headache.

• ✅ Kubernetes (v1.34.0) with kubeadm
• ✅ Calico CNI with WireGuard encryption enabled
• ✅ Prometheus Node Exporter (port 9100) - ready for your Prometheus server
• ✅ SigNoz OpenTelemetry Collector - ships logs and metrics automatically
• ✅ System optimizations - swap disabled, kernel modules loaded, sysctls tuned
• ✅ Zero-to-hero - from VM creation to running cluster in ~5 minutes

• Proxmox (or any cloud-init compatible hypervisor)
• Ubuntu 22.04+ VM template
• Your SSH public key (already configured in the files)

Use cloud-init/k8s-full-calico.yaml for your first node (control plane):

1. Create a new VM in Proxmox
2. In the cloud-init tab, paste the contents of k8s-full-calico.yaml
3. Update the SIGNOZ_ENDPOINT in the monitoring config section (or edit /etc/monitoring/config.env after boot)
4. Boot the VM and grab a coffee ☕
5. SSH in and check status: kubectl get nodes

The join command is saved to /root/join.sh for adding worker nodes.

If cloud-init isn't working or you prefer a script-based approach, use setup-k8s.sh:

One-liner to download and execute:

curl -fsSL https://your-server.com/setup-k8s.sh | sudo bash

Or download first, review, then execute:

curl -fsSL https://your-server.com/setup-k8s.sh -o setup-k8s.sh
sudo bash setup-k8s.sh

⚠️ Security Note: Always review scripts before executing, especially from remote sources.

Use cloud-init/base-ubuntu.yaml to create a base template with common packages and your SSH key. Then clone and customize from there.

Both monitoring services are pre-configured but need your server URLs:

SigNoz (logs & metrics):

# Edit on the VM after creation
sudo nano /etc/monitoring/config.env
# Set SIGNOZ_ENDPOINT="http://your-signoz-server:4318"
sudo systemctl restart otelcol

Prometheus (metrics): Just add a scrape job pointing to node-ip:9100 in your Prometheus config. The Node Exporter is already running and waiting.

See MONITORING.md for detailed setup instructions.

• cloud-init/k8s-full-calico.yaml - Full Kubernetes + Calico + monitoring setup (cloud-init)
• setup-k8s.sh - Bash script alternative (same functionality as cloud-init)
• cloud-init/base-ubuntu.yaml - Base Ubuntu configuration for templates
• MONITORING.md - Monitoring configuration guide

The configs are designed to be modified. Common tweaks:

• Kubernetes version: Change kubernetesVersion in the kubeadm config
• Pod CIDR: Update podSubnet (currently 192.168.0.0/16)
• Calico MTU: Adjust in calico-ip-pool.yaml if needed
• User/SSH keys: Update the users section

• This is optimized for lab environments - not production-hardened
• Control plane taint is removed (workloads can run on master)
• Swap is disabled (Kubernetes requirement)
• WireGuard is enabled for encrypted pod-to-pod communication
• All services auto-start on boot

Cluster not initializing?

• Check journalctl -u kubelet for errors
• Verify containerd is running: systemctl status containerd
• Ensure swap is off: swapoff -a

Monitoring not working?

• Verify services: systemctl status node_exporter otelcol
• Check SigNoz endpoint in /etc/monitoring/config.env
• See MONITORING.md for detailed troubleshooting

See LICENSE file.

Pro tip: Create a VM template from the base config, then clone it for faster node provisioning. Your future self will thank you. 🙏