Webchat GitOps (Kind + ArgoCD + OpenTofu)

A real-time communication platform utilizing a Python (FastAPI) backend and a static/Nginx frontend, orchestrated following GitOps practices via ArgoCD and OpenTofu in a Kind-based local Kubernetes environment.

This project demonstrates a fully automated workflow:

• Build → Push → GitOps → Deploy
• Fully local (Kind + local registry)
• Dynamic multi-app deployment using ArgoCD ApplicationSet
• Kustomize-based or helm apps
• Real-time WebSocket chat backed by Redis Pub/Sub

---

Features

• Multi-application GitOps using ApplicationSet with directory generator
• Local registry integration with Kind
• FastAPI WebSocket backend + Redis Pub/Sub
• Nginx frontend serving static assets
• Taskfile-based automation (build, push, apply)
• Argo CD Image Updater integration:
  • Updates container images in-place without Git write-back
  • Fully reproducible for local development
  • Works with Kustomize-based apps
  • Configured to always pick the newest build

---

Architecture

graph TD
    %% External Source
    GitHub[(GitHub Repo)]

    subgraph Host [Local Environment]
        Taskfile[Taskfile.dev]
        Tofu[OpenTofu / Helm]
        Registry[Docker Registry]
    end

    subgraph Kind [Kind Kubernetes Cluster]

        subgraph ArgoStack [ArgoCD Setup]
            ArgoSuite[ArgoCD & AppSet Controller]
            ImageUpdater[Argo Image Updater]
            AppConfigs[AppSet / Project CRDs]
            ArgoApps[ArgoCD Applications]
        end

        subgraph Workloads [App Namespaces]
            direction TB
            Web[Web Frontend]
            Backend[Backend API]
            Redis[Redis]
        end

    end

    %% Flow 1: Setup
    Taskfile --> Tofu
    Tofu --> ArgoStack

    %% Flow 2: GitOps Sync
    ArgoSuite -- "1. Reads" --> GitHub
    ArgoSuite -- "2. Reconciles" --> AppConfigs
    ArgoSuite -- "3. Generates" --> ArgoApps
    ArgoApps -- "4. Deploys" --> Workloads

    %% Flow 3: Image Update Loop
    Taskfile -- "A. Pushes Image" --> Registry
    ImageUpdater -- "B. Polls" --> Registry
    ImageUpdater -- "C. Updates Tag" --> ArgoApps

Loading

---

Prerequisites

• go-task
• Docker
• kind
• kubectl
• OpenTofu (open source Terraform alternative)

---

Setup

Setup everything

task setup

This will:

• Create a Kind cluster and local registry
• Install ArgoCD and Argo CD Image Updater
• Apply GitOps configuration
• Build & push backend and frontend images

---

Destroy everything

task teardown

---

Application

• WebSocket endpoint: /chat/ws
• Real-time chat messages broadcast via Pub/Sub
• Frontend served on Nginx
• Backend served with FastAPI

---

Development Workflow

# Make changes to backend or frontend
task backend:push
# or
task web:push
# or
task push-all

# ArgoCD ApplicationSet auto-syncs
# Image Updater automatically updates Deployment images

No manual kubectl apply required. The system handles namespace creation, image updates, and application deployment automatically.

---

Considerations

• In a production scenario, the git writeback method for argo image updater should be chosen for visibility and more reliable GitOps setup. The argo method was chosen to enable local development, without the need to push anything to git
• This project is structered as a monorepo. The application code, IaC and GitOps setup is all contained here. In a production environment these components should be decoupled for beter maintainability and more flexible deployment process