🚀 From Frustration to Production: Deploying a Full-Stack App with Terraform & Ansible on AWS

When I started this project, I thought it would be a straightforward deployment.

It wasn’t.

From database connection errors to broken Terraform configs and missing Ansible templates, this project pushed me to think like a real DevOps engineer — not just follow tutorials.

In this article, I’ll walk you through how I deployed a Node.js application (EpicBook) using:

Terraform → Infrastructure provisioning
Ansible → Configuration & deployment
AWS (EC2 + RDS) → Hosting
PM2 + Nginx → Production runtime

And most importantly…
👉 I’ll show you the exact commands, mistakes, and fixes so you can replicate it yourself.

🏗️ Architecture Overview

Here’s what we’re building:

EC2 (Ubuntu) → Runs the app
RDS (MySQL) → Stores data (private subnet)
Nginx → Reverse proxy
PM2 → Keeps Node.js running
⚙️ Step 1: Provision Infrastructure with Terraform

First, I navigated into my Terraform directory:

cd terraform/aws

Then initialized Terraform:

terraform init

Validated config:

terraform validate

And applied the infrastructure:

terraform apply
⚠️ First Major Problem
Error: No configuration files

💡 Fix: I was in the wrong directory. Always ensure you're inside the folder containing .tf files.

⚠️ Second Problem (Very Important)
DBSubnetGroupDoesNotCoverEnoughAZs

💡 Fix:
RDS requires at least 2 Availability Zones.

I updated my Terraform to include:

Multiple private subnets
Different AZs
✅ Output

After success:

ec2_public_ip = "13.x.x.x"
rds_endpoint = "epicbook-db.xxxxx.amazonaws.com:3306"
🔐 Step 2: Connect to EC2
ssh -i ~/.ssh/key.pem ubuntu@
⚙️ Step 3: Install Ansible
sudo apt update
sudo apt install ansible-core -y

Verify:

ansible --version
⚠️ Error I Hit
ansible-playbook: command not found

💡 Fix: Install Ansible (it’s not pre-installed).

🤖 Step 4: Run Ansible Playbook
cd ansible
ansible-playbook -i inventory.ini site.yml

This automated:

Nginx setup
App deployment
DB configuration
📦 Step 5: Application Setup

Ansible handled:

git clone
npm install
sudo apt install nodejs npm mysql-client -y
🛢️ Step 6: Database Setup (RDS)

I created the database:

mysql -h -u admin -p -e "CREATE DATABASE bookstore;"
⚠️ Big Issue #1
ECONNREFUSED 127.0.0.1:3306

💡 Cause:
App was trying to connect to localhost

💡 Fix:

Passed environment variables:
DB_HOST → RDS endpoint
DB_NAME → bookstore
⚙️ Step 7: Configure App

Using Ansible template:

template:
src: config.json.j2
dest: /var/www/epicbook/config/config.json
⚠️ Error
config.json.j2 not found

💡 Fix:
Create the file here:

roles/epicbook/templates/config.json.j2
🚀 Step 8: Run App with PM2
npm install -g pm2
pm2 start server.js --name epicbook
pm2 save
⚠️ Issue
pm2 delete epicbook → not found

💡 Fix:

ignore_errors: true
🗄️ Step 9: Database Schema & Seeding

This was the trickiest part.

Run schema:

mysql -h -u admin -p bookstore < BuyTheBook_Schema.sql

Seed data:

mysql -h -u admin -p bookstore < author_seed.sql
mysql -h -u admin -p bookstore < books_seed.sql
⚠️ Issue #1
Table 'books' doesn't exist

💡 Fix:
Schema must run before seeding

⚠️ Issue #2
Unknown database 'bookstore'

💡 Fix:

Standardised DB name across:
Terraform
Ansible
SQL files
⚠️ Issue #3
Table already exists

💡 Fix:
Make tasks idempotent:

ignore_errors: yes
🎉 Final Result

I opened the app:

http://

✅ And finally…
📚 Books were displaying from the database

That moment? Worth every error.

🧠 What This Project Taught Me
Terraform is for infrastructure, not configuration
Ansible eliminates manual setup (when done right)

Order matters:

DB → Config → App → Seed
Debugging is a core DevOps skill
Small misconfigurations (like DB name) can break everything
💡 What I’d Improve Next
Use Ansible MySQL modules instead of shell
Add Load Balancer (ALB)
Implement Auto Scaling
Store secrets in AWS Secrets Manager
Add CI/CD pipeline
🚀 Final Thoughts

This wasn’t just a deployment project.

It was a real-world DevOps experience:

Broken configs
Debugging under pressure
Fixing issues step by step

And in the end… building something that actually works in production.

🔗 If you're learning DevOps

Don’t just follow tutorials.

👉 Break things
👉 Fix them
👉 Understand why

That’s how you grow.