Code to Cloud: Deploying a Full-Stack App with Docker, Terraform & CI/CD
# From Code to Cloud: Deploying a Full-Stack App with Docker, Terraform & CI/CD
Building applications is one thing.
**Deploying them reliably in production is a completely different game.**
In this blog, I’ll walk through how I built and deployed a **Hello World full-stack application** using **Django + React**, containerized it with **Docker**, provisioned infrastructure using **Terraform**, and automated everything using **GitHub Actions CI/CD** — along with the **real problems I faced and how I fixed them**.
This is not a toy tutorial. These are real DevOps lessons.
---
## 🧩 Project Overview
### Tech Stack
- **Backend**: Django (REST API)
- **Frontend**: React + Vite + TypeScript
- **Web Server**: Nginx
- **Containerization**: Docker & Docker Compose
- **Infrastructure**: AWS EC2 + Application Load Balancer (ALB)
- **IaC**: Terraform
- **CI/CD**: GitHub Actions
### High-Level Architecture
User → ALB → EC2 → Docker Compose
├── Frontend (Nginx :80 → 3000)
└── Backend (Gunicorn :8000)
---
## 🐳 Dockerizing the Application (The Right Way)
### Backend: Django with Multi-Stage Docker Build
To keep images small and secure, I used **multi-stage Docker builds**.
#### Why Multi-Stage Builds?
- Smaller image size
- Faster deployments
- No build tools in production
- Better security
### Backend Dockerfile
```dockerfile
FROM python:3.12-slim AS builder
WORKDIR /app
RUN apt-get update && apt-get install -y \
gcc libpq-dev \
&& rm -rf /var/lib/apt/lists/*
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
FROM python:3.13-slim
WORKDIR /app
RUN useradd -m appuser && \
chown -R appuser /app /home/appuser
COPY --from=builder /usr/local /usr/local
COPY . .
COPY entrypoint.sh .
RUN chmod +x /app/entrypoint.sh
USER appuser
ENV PYTHONDONTWRITEBYTECODE=1
ENV PYTHONUNBUFFERED=1
EXPOSE 8000
CMD ["/app/entrypoint.sh"]
⚙️ Backend Entrypoint Script
The entrypoint ensures the backend is production-ready on startup.
#!/bin/sh
set -e
echo "Starting backend..."
python manage.py migrate --noinput || true
python manage.py collectstatic --noinput || true
exec gunicorn config.wsgi:application --bind 0.0.0.0:8000
Why this matters
Containers self-initialize
No manual DB steps
Works consistently across environments
🎨 Frontend: React + Vite + Nginx
The frontend uses a build once, serve fast approach.
Frontend Dockerfile
FROM node:20 AS builder
WORKDIR /app
COPY package*.json .
RUN npm ci
COPY . .
RUN npm run build
FROM nginx:alpine
COPY --from=builder /app/dist /usr/share/nginx/html
COPY nginx.conf /etc/nginx/conf.d/default.conf
EXPOSE 80
🔗 Docker Compose – Service Orchestration
services:
backend:
image: krishnahogale/devops-assessment-backend:latest
ports:
- "8000:8000"
environment:
- DEBUG=false
- ALLOWED_HOSTS=*
frontend:
image: krishnahogale/devops-assessment-frontend:latest
ports:
- "3000:80"
depends_on:
- backend
Key Decision
Use pre-built images, not
build:on EC2Ensures consistency between CI and production
☁️ Infrastructure Provisioning with Terraform
Terraform provisions:
EC2 instance
Application Load Balancer
Target Groups
Security Groups
Outputs
ec2_public_ipalb_dns_name
These outputs are consumed by the CI/CD pipeline.
🔄 CI/CD Pipeline – Terraform + Docker + EC2
Pipeline Name
CI/CD Pipeline - Terraform + Docker + EC2
Pipeline Flow
Provision Infrastructure using Terraform
Build & Push Docker Images to Docker Hub
Deploy to EC2 using SSH & Docker Compose
Destroy Infrastructure on Failure
This pipeline ensures:
Fully automated deployments
Zero manual EC2 setup
Cost safety using auto-destroy
🚨 Troubleshooting: Real Issues I Faced
❌ Issue 1: Backend Container Exiting Immediately
Cause
- Used
pythonmanage.pyrunserverin container
Fix
Switched to Gunicorn
Added
entrypoint.sh
❌ Issue 2: Database Migrations Not Applied
Cause
- Migrations not executed inside container
Fix
python manage.py migrate --noinput || true
❌ Issue 3: Docker Image Size Too Large
Cause
- Build tools present in runtime image
Fix
- Implemented multi-stage Docker builds
❌ Issue 4: Frontend Not Loading via Nginx
Cause
- Incorrect build artifact copy
Fix
COPY --from=builder /app/dist /usr/share/nginx/html
❌ Issue 5: Docker Compose Rebuilding Images on EC2
Cause
- Used
build:instead ofimage:
Fix
- Switched to Docker Hub images
❌ Issue 6: Permission Denied for Entrypoint Script
Cause
- Script not executable
Fix
chmod +x entrypoint.sh
❌ Issue 7: ALB Returning 503 Service Temporarily Unavailable
Symptoms
http://98.xx.xx.xx:3000 ✅ Works
http://alb-dns.amazonaws.com ❌ 503
Root Causes
Target group port mismatch
Incorrect health check path
Missing ALB → EC2 security group rule
Fix
Target group port →
3000Health check path →
/Updated security groups
❌ Issue 8: Docker Not Installed on Fresh EC2
Fix
- Installed Docker automatically during deployment
❌ Issue 9: CI/CD Failure Left AWS Resources Running
Fix
- Added
destroy-on-failurejob in GitHub Actions
🧠 Key DevOps Learnings
ALB
503almost always means unhealthy targetsNever use Django
runserverin productionMulti-stage Docker builds are essential
CI/CD pipelines must handle failure paths
Automate cleanup to avoid AWS cost leaks
About the author
- I am an Information Technology Analyst at NTT DATA North America, where I specialize in DevOps, automation testing, and cloud-native deployments. Over the past 3 years, I’ve built CI/CD pipelines, test automation frameworks, and Generative AI solutions that help organizations streamline their workflows and improve efficiency. My Background - I studied at the Indian Institute of Technology, Patna. - Since July 2022, I’ve been contributing to projects at NTT DATA North America. My Skills - Automation Testing: REST API and UI automation using Java, Selenium, REST Assured, TestNG. - DevOps Tools: GitHub Actions, Jenkins, Docker, Terraform, Kubernetes, Azure DevOps, AWS. - Programming: Proficient in Python and Java. - Cloud & Microservices: Deploying Python-based applications on AWS EC2, integrating with MySQL and OpenAI APIs. - Monitoring & Performance: Experienced with Datadog, Grafana, JMeter, Gatling. My Achievements - Improved test efficiency by 40% through automation frameworks. - Led Generative AI initiatives, including prompt engineering, LangChain workflows, and retrieval-augmented generation (RAG) systems. - Enhanced application performance by 10% through API call caching and load testing. My Professional Presence I am active on LinkedIn, where I have over 4,000 followers and 500+ connections
Comments
Join the discussionNo comments yet. Be the first to comment.

