Cloud-Native CI/CD Platform with Safe Deployment & Auto Failure Detection

Overview

This project demonstrates a production-grade, cloud-native CI/CD system built entirely on AWS. It automates the full lifecycle from code commit → build → test → deploy → validate → serve, ensuring that only healthy and verified deployments are exposed to users.

The system incorporates infrastructure-level validation, container orchestration, and load-balanced traffic routing to simulate real-world DevOps environments.

---

Objectives

• Build a fully automated CI/CD pipeline
• Ensure zero-downtime deployments
• Implement health-based traffic routing
• Detect and prevent faulty deployments
• Gain hands-on experience with AWS DevOps ecosystem

---

System Architecture

Developer → CloudShell → CodeCommit → CodePipeline → CodeBuild → ECR → ECS (Fargate) → ALB → Users
                          ↓
                    CloudWatch Logs & Metrics
                          ↓
                         IAM (Access Control Layer)

---

End-to-End Workflow

1. Developer pushes code (CloudShell → CodeCommit)
2. CodePipeline triggers automatically
3. CodeBuild: → Builds Docker image → Runs tests → Pushes image to ECR
4. ECS: → Pulls new image → Starts new task
5. ALB: → Sends /health request → Validates container
6. If healthy: → Traffic routed to container
7. If unhealthy: → Task terminated (auto failure handling)

Tech Stack & Services Used

Development & Access

• CloudShell

  • Used for Git operations, AWS CLI, debugging, and pipeline triggering
  • Eliminates local environment dependency

---

Identity & Security

• IAM (Identity and Access Management)

  • Managed permissions for:

    • CodeBuild (ECR access)
    • CodePipeline (ECS deploy)
    • ECS task execution roles

  • Ensures least-privilege access control

---

Source Control

• AWS CodeCommit

  • Private Git repository
  • Triggers pipeline on code push

---

CI/CD Orchestration

• AWS CodePipeline

  • Automates:

    • Source → Build → Deploy stages

  • Event-driven pipeline execution

---

Build System

• AWS CodeBuild

  • Runs in isolated container environment

  • Executes:

    • Docker build
    • Test scripts
    • Push to ECR

  • Uses buildspec.yml

---

Container Registry

• Amazon ECR (Elastic Container Registry)

  • Stores versioned Docker images
  • Secure image retrieval for ECS

---

Compute Layer

• Amazon ECS (Fargate)

  • Serverless container orchestration

  • No EC2 management required

  • Handles:

    • Task scheduling
    • Scaling
    • Deployment updates

---

Networking & Traffic Routing

• Application Load Balancer (ALB)

  • Public entry point
  • Routes HTTP traffic to ECS tasks
  • Performs health checks

• Target Group

  • Registers ECS tasks
  • Tracks health status
  • Enables intelligent routing

---

Infrastructure (Underlying)

• Amazon EC2 (Implicit via ALB & Networking)

  • Provides VPC, subnets, and networking backbone
  • Hosts ALB infrastructure

---

Monitoring & Observability

• Amazon CloudWatch

  • Logs:

    • ECS container logs
    • CodeBuild logs

  • Metrics:

    • CPU / Memory
    • Task failures

  • Used for debugging deployment issues

---

End-to-End Workflow

1. Developer pushes code (CloudShell → CodeCommit)
2. CodePipeline triggers automatically
3. CodeBuild:
   → Builds Docker image
   → Runs tests
   → Pushes image to ECR
4. ECS:
   → Pulls new image
   → Starts new task
5. ALB:
   → Sends /health request
   → Validates container
6. If healthy:
   → Traffic routed to container
7. If unhealthy:
   → Task terminated (auto failure handling)

---

Health Check Mechanism

GET /health → 200 OK

• Lightweight endpoint
• Independent of DB
• Used by ALB for validation

---

Real-World Challenges & Debugging

Issue Encountered:

503 Service Unavailable
0 healthy targets

Root Causes:

• Application not responding to /health
• Double app.listen() causing crash
• DB dependency blocking startup
• Health check timeout too aggressive
• ECS tasks restarting continuously

---

Fixes Implemented

• Added /health endpoint
• Fixed server binding (0.0.0.0:3000)
• Removed duplicate server start
• Increased health check grace period
• Corrected ECS ↔ ALB target mapping
• Simplified startup (removed blocking DB init)

---

Final Outcome

Metric	Before	After
Deployment	Failed	Successful
Target Health	0	1 Healthy
Availability	503 Error	Working Endpoint
Stability	Unstable	Stable

---

Live Endpoint

http://<ALB-DNS>/health

Expected Response:

OK

---

IAM Role Design

• CodeBuild Role:

  • ECR access
  • CloudWatch logs

• CodePipeline Role:

  • ECS update permissions
  • Task definition registration

• ECS Task Role:

  • Pull image from ECR
  • Logging permissions

---

Key DevOps Learnings

• Container running ≠ application ready
• Health checks are critical for production systems
• Startup dependencies must not block readiness
• Observability is essential for debugging
• Proper IAM design is required for secure pipelines

---

Future Enhancements

• Blue-Green deployment (CodeDeploy)
• Prometheus + Grafana monitoring
• Terraform (Infrastructure as Code)
• Auto-scaling based on metrics
• Canary deployments

---

Resume Highlight

Designed and implemented a cloud-native CI/CD pipeline using AWS services with automated container deployment, health-based validation, and failure detection to ensure reliable production releases.

---

Author

Shreya Baranwal

---

Support

If you found this useful, give it a ⭐ and share!