Zeus — Event-Driven E-commerce on EKS (GitOps)

This repository contains both the platform (an Amazon EKS cluster provisioned with Terraform and bootstrapped with ArgoCD for GitOps) and the e-commerce application it runs: a set of event-driven microservices plus a TanStack Start storefront.

---

🛍️ E-commerce Application (apps/)

A polyglot, event-driven shop. Services communicate over a shared RabbitMQ topic exchange (basebandit.events) and coordinate an order saga (currently event-choreographed; see apps/orders/SAGA_PATTERN.md, with a planned migration to Temporal).

Service	Stack	Port	Responsibility
web	TypeScript / TanStack Start	3000	Storefront UI + BFF (calls the gateway only)
gateway	Go / Gin	8081	Single public edge: JWT validation, identity forwarding, routing
auth	Java / Spring Boot	8084	Registration, login, JWT issuance; publishes user.registered
orders	TypeScript / NestJS	8080	Cart, order lifecycle, saga orchestrator
inventory	Go / Gin	8082	Product catalog, stock reservation
payments	Python / FastAPI	8083	Payment processing
shipping	Python / FastAPI	8085	Fulfillment; owns shipment.* events
notifications	Java / Spring Boot	8086	Email on order/user events (via MailHog locally)

Browser ─► web (SSR + BFF) ─► gateway (JWT edge) ─► auth │ orders │ inventory │ payments │ shipping
                                                     └──── RabbitMQ basebandit.events ────► notifications

Order flow: order.created → inventory.reserved → payment.completed → order.confirmed → shipment.shipped → order.shipped → (delivery) → order.delivered, with compensation on inventory/payment failure → order.cancelled.

Run the whole stack locally

docker compose up --build          # all services + RabbitMQ + MailHog + per-service Postgres
./apps/inventory/seed-products.sh        # seed a 20-item demo catalog (with images)

Then open the storefront at http://localhost:3000 (RabbitMQ UI :15672, MailHog :8025). Each service also has its own docker-compose.yml and README.md for isolated development.

---

☁️ Platform (Terraform + ArgoCD)

This project provisions a development platform on Amazon EKS cluster using Terraform. The setup includes VPC networking, EKS control plane, worker nodes, and bootstrapping ArgoCD for GitOps workflows.

---

✅ What This Project Does

• Provisions a secure and scalable VPC with public/private subnets
• Deploys an Amazon EKS Cluster with worker nodes (Node Groups)
• Installs ArgoCD and ArgoCD Image Updater via Helm using Terraform
• Uses EKS Pod Identity Agent (instead of IRSA) for managing IAM roles for service accounts

---

🧰 Prerequisites

• Terraform CLI ≥ 1.3
• AWS CLI configured with a profile or IAM credentials
• kubectl configured (will be needed to interact with EKS)
• Helm CLI (for optional local management)

---

📁 Project Structure

.
├── charts
│   ├── argocd
│   │   └── argocd-server
│   │       ├── argocd-config
│   │       ├── staging-argocd-image-updater-values.yaml
│   │       └── staging-argocd-values.yaml
│   └── rbac
├── infrastructure
│   └── staging
│       ├── argo-image-updater.tf
│       ├── argocd.tf
│       ├── eks.tf
│       ├── igw.tf
│       ├── locals.tf
│       ├── nat.tf
│       ├── nodes.tf
│       ├── outputs.tf
│       ├── pod-identity-addon.tf
│       ├── providers.tf
│       ├── routes.tf
│       ├── subnets.tf
│       ├── terraform.tfstate
│       ├── variables.tf
│       └── vpc.tf
└── README.md

---

🚀 Step-by-Step Setup

✅ Step 1: Provision the VPC (tag: v1.0.0)

git checkout v1.0.0
cd infrastructure/staging
terraform init
terraform apply

This sets up the VPC, subnets, NAT Gateway, Internet Gateway, and route tables.

---

✅ Step 2: Provision the EKS Cluster (tag: v2.0.0)

git checkout v2.0.0
cd infrastructure/staging
terraform init
terraform apply

This step provisions the EKS control plane, node groups (e.g. t3.medium), IAM roles for nodes, and enables the EKS Pod Identity Agent instead of traditional IRSA.

---

✅ Step 3: Install and Configure ArgoCD via Helm (tag: v3.0.0)

git checkout v3.0.0
cd infrastructure/staging
terraform init
terraform apply

This step deploys:

• ArgoCD Helm chart
• ArgoCD Image Updater
• Custom values from charts/argocd/argocd-server/*.yaml
• ArgoCD is deployed in the argocd namespace and bootstrapped for GitOps
• Uses Pod Identity Agent for access to AWS (e.g., fetching ECR credentials)

---

🧹 Clean Up

To tear down all infrastructure:

terraform destroy

To destroy only ArgoCD resources:

terraform destroy -target=helm_release.argocd -target=helm_release.argocd_image_updater

---

🖥️ ArgoCD UI

Below are screenshots from the deployed ArgoCD instance showing the GitOps application dashboards:

🔹 ArgoCD Dashboard (App of Apps)

---

🔹 Staging Payments Application View

---

📝 Notes

• Make sure your AWS vCPU limits can support the desired number of nodes (e.g. t3.medium).
• Default ArgoCD configs are set to run with minimal replicas and resources. Adjust accordingly based on cluster capacity.
• Pod scheduling issues may arise on underpowered nodes (e.g. t3.micro) due to ArgoCD controller resource limits.

---

Built for educational purposes, you will need a lot more configuring for a production-ready environment.