About terraform-talos-gitops-cluster

Kubernetes cluster with fully automated infrastructure using Terraform and Talos on Proxmox. GitOps with Argo CD, Cilium networking, and full observability powered by OpenTelemetry.

s

Published by

serhii-myronets

Visit View Profile

README.md

View on GitHub

Talos Kubernetes Cluster on Proxmox with Terraform

This repository contains infrastructure-as-code configurations for deploying a minimal, production-grade Kubernetes cluster using Talos Linux and Terraform on Proxmox VE. This repository provides a fully declarative, script-free setup: Talos is configured and installed automatically during VM provisioning via Terraform.

Overview

This project is designed for enthusiasts, students, or professionals who want to gain hands-on experience with a production-grade GitOps Kubernetes cluster using modest hardware — such as an old laptop or mini PC. It offers a fully automated deployment pipeline without requiring cloud resources or expensive infrastructure.

The Kubernetes cluster is composed of multiple control plane and worker nodes provisioned on a Proxmox host using Terraform. Talos Linux is injected and configured automatically as part of the VM provisioning step. The configuration supports high availability (HA) and uses a virtual IP for the control plane endpoint. The deployment includes core platform components (ingress, certificate management, GitOps), observability stack (metrics, logs, traces), and demo microservices applications for testing.

Architecture

The Kubernetes cluster operates in an isolated subnet (192.168.100.0/24) with virtual machines provisioned directly on a Proxmox VE host. A dedicated NAT bridge (vmbr1) is used to provide connectivity. Each node is assigned a static IP from this subnet. The control plane nodes are configured in high availability (HA) mode and share a virtual IP (192.168.100.50) for the Kubernetes API.

Proxmox VE (192.168.1.100)
  └─ vmbr1: 192.168.100.1 (NAT Gateway)
       ├─ controlplane-1: 192.168.100.60
       ├─ controlplane-2: 192.168.100.61
       ├─ worker-1:      192.168.100.70
       ├─ worker-2:      192.168.100.71
       └─ cluster VIP:   192.168.100.50 (Kubernetes API)

A static route to 192.168.100.0/24 must be configured on the developer workstation via the Proxmox host.

Features

Support for high availability across control-plane nodes
Fully declarative setup (no shell scripts)
Talos Linux installed and configured via Terraform
Proxmox-native VM provisioning
GitOps with Argo CD and Helmfile
Cilium CNI with kube-proxy disabled
Longhorn for persistent volumes
Full observability stack with OpenTelemetry Collector (metrics, logs, traces via Tempo, Loki, Prometheus, Grafana) Demo microservices instrumented for end-to-end tracing and performance metrics collection

Directory Structure

Path	Description
`00-prerequisite/`	Environment preparation: hardware requirements, dependencies, Proxmox and networking setup
`01-infrastructure/`	Terraform configurations for Proxmox VM provisioning and Talos injection
`02-bootstrap/`	Installs base components (cert-manager, ingress, Argo CD, Longhorn, etc.) using Helmfile
`03-gitops/`	Deploys applications via Argo CD, including observability stack and demo workloads