Module 4

CI/CD on AWS

Build a full deployment pipeline from GitHub to EC2 using CodePipeline, CodeBuild, and CodeDeploy.

CodePipelineCodeBuildCodeDeployAppSpec

What You'll Learn

Automate deployments from GitHub to EC2 using AWS's CI/CD trio: CodePipeline, CodeBuild, and CodeDeploy. Push code → tests run → app deploys automatically.

Azure	AWS	Notes
`Azure DevOps Pipeline`	`CodePipeline`	Orchestrates the full flow
`Build Agent`	`CodeBuild`	Managed build environment
`Release Pipeline`	`CodeDeploy`	Deploys to EC2/ASG
`Service Connection`	`CodeStar Connection`	Links to GitHub
`azure-pipelines.yml`	`buildspec.yml`	Build definition file

☁️ Azure Parallel

Key difference: Azure DevOps bundles build + release into one service. AWS separates them into 3 distinct services (CodePipeline + CodeBuild + CodeDeploy). More modular, but more to configure.

The Pipeline Flow

text

GitHub Push
    │
    ▼
┌─────────────────────────┐
│  Stage 1: SOURCE        │  CodeStar Connection detects push
│  (CodePipeline)         │  Downloads source code
└───────────┬─────────────┘
            ▼
┌─────────────────────────┐
│  Stage 2: BUILD         │  Uses buildspec.yml
│  (CodeBuild)            │  npm install → npm test → package
└───────────┬─────────────┘
            ▼
┌─────────────────────────┐
│  Stage 3: DEPLOY        │  Uses appspec.yml
│  (CodeDeploy)           │  Stop → Install → Start → Validate
└─────────────────────────┘

Step 1: buildspec.yml — Build Definition

This file tells CodeBuild what to do. It's like anazure-pipelines.yml but with AWS-specific phases.

yamlbuildspec.yml

version: 0.2

phases:
  install:
    runtime-versions:
      nodejs: 18
    commands:
      - cd app && npm ci

  pre_build:
    commands:
      - cd app && npm test

  build:
    commands:
      - echo "Build succeeded on $(date)"

artifacts:
  files:
    - "**/*"
  discard-paths: no

📘 Key Concept

Phases explained: install sets up the environment,pre_build runs quality checks, build does the main build, and the artifacts section defines what gets passed to CodeDeploy.

Step 2: appspec.yml — Deployment Definition

This file tells CodeDeploy how to deploy to each EC2 instance. It defines lifecycle hooks that run scripts at specific stages.

yamlappspec.yml

version: 0.0
os: linux

files:
  - source: /app
    destination: /home/ec2-user/aws-sandbox-app

hooks:
  ApplicationStop:
    - location: scripts/stop_server.sh       # Stop old version
  BeforeInstall:
    - location: scripts/install_dependencies.sh  # Install Node.js
  AfterInstall:
    - location: scripts/after_install.sh     # npm install, run migrations
  ApplicationStart:
    - location: scripts/start_server.sh      # pm2 start
  ValidateService:
    - location: scripts/validate_service.sh  # curl /health

Deployment Strategies

🔄 In-Place (OneAtATime)

Updates one instance at a time. Traffic is rerouted while each instance updates. Simple and safe.

🔵🟢 Blue/Green

Launches new instances, switches ALB traffic all at once. Zero-downtime, easy rollback.

Step 3: CodeStar Connection to GitHub

Before deploying the pipeline, create a GitHub connection:

Go to CodePipeline Console → Settings → Connections
Click Create connection → GitHub
Authorize AWS in your GitHub account
Copy the Connection ARN — you'll need it for the CloudFormation stack

🧪

Hands-On: Deploy Pipeline and Trigger a Build

bash

# Deploy the CI/CD stack
aws cloudformation create-stack \
  --stack-name aws-sandbox-cicd \
  --template-body file://cloudformation/07-cicd-pipeline.yaml \
  --capabilities CAPABILITY_NAMED_IAM \
  --parameters \
    ParameterKey=GitHubConnectionArn,ParameterValue=YOUR_ARN \
    ParameterKey=GitHubOwner,ParameterValue=YOUR_USERNAME \
    ParameterKey=GitHubRepo,ParameterValue=AWS_sandbox

# Trigger the pipeline with a code push
echo "# Deploy test $(date)" >> README.md
git add . && git commit -m "test: trigger pipeline" && git push

# Watch the pipeline status
aws codepipeline get-pipeline-state --name aws-sandbox-pipeline \
  --query 'stageStates[*].{Stage:stageName,Status:latestExecution.status}'

Watch the pipeline progress through Source → Build → Deploy. Check the CodePipeline console for a visual view of each stage.

Key Takeaways

CodePipeline orchestrates, CodeBuild builds, CodeDeploy deploys
buildspec.yml = what to build, appspec.yml = how to deploy
CodeDeploy lifecycle hooks give you full control over the deployment process
In-Place deployment is simpler; Blue/Green is safer for production