Cloud Solutions/Infrastructure/DMI Capstone Project

PetClinic CloudOps — AWS EKS GitOps Platform

11Services Orchestrated
EnabledGitOps Self-Healing
FullObservability Stack

The Challenge

Deploying an 11-service Spring Boot microservices architecture on AWS required production-grade Kubernetes infrastructure with automated GitOps deployments, controlled startup sequencing to prevent crash-loop failures, full observability, and zero-touch secrets management — without any credentials ever touching the repository.

The Engineering Solution

Designed the complete Kubernetes layer on AWS EKS using Helm umbrella charts packaging all 11 services as a single deployable unit. Implemented ArgoCD GitOps with automated pruning and self-healing (selfHeal: true), controlled startup sequencing via init containers (config-server → discovery-server → all services), Nginx Ingress routing exclusively through the api-gateway, and Azure OpenAI secrets managed via Kubernetes Secrets. Delivered a fully auto-provisioned observability stack (Prometheus, Grafana with preloaded dashboards, Zipkin) and a one-shot bootstrap script automating all 8 cluster setup steps.

Technologies Used

AWS EKSKubernetes v1.30HelmArgoCDTerraformPrometheusGrafanaZipkinNginx IngressGitHub ActionsSpring Boot

Interactive System Mockup & Analytics

CLUSTER ARCHITECTURE

PetClinic CloudOps — AWS EKS Cluster Architecture

GitOps Layer
ArgoCD — Dev Env
ArgoCD — Production
selfHeal: true
Auto-pruning
Kubernetes Layer
11 Helm sub-charts
Umbrella chart
Init containers
Nginx Ingress
Observability
Prometheus scraping
Grafana dashboards
Zipkin tracing
Health endpoints

System State Comparison

Before State

Manual service deployments with no unified GitOps strategy, no startup sequencing, and no observability

After Integration

Fully automated EKS GitOps cluster with ArgoCD self-healing, init-container sequencing, and full observability

A remarkably clean Kubernetes architecture. The umbrella chart approach made deployments, upgrades, and rollbacks trivially simple — and the observability stack was production-ready from day one.

Technical ReviewerDevOps Micro Internship Programme