DEV Community

Cover image for 🚀 NestJS Microservices with RabbitMQ, Retries & DLQ — Production Setup That Actually Scales.
codingKrills
codingKrills

Posted on

🚀 NestJS Microservices with RabbitMQ, Retries & DLQ — Production Setup That Actually Scales.

#architecture #backend #microservices #tutorial

🚀 NestJS Microservices with RabbitMQ, Retries & DLQ — Production Setup That Actually Scales

❌ Stop building fragile microservices that silently fail.
✅ Start building systems that retry, recover, and scale automatically.

If you're using NestJS and not handling failures, retries, and dead-letter queues (DLQ) properly… your system will break in production. Period.

In this guide, I’ll show you a battle-tested architecture using NestJS + RabbitMQ with:

  • 🔁 Retries
  • 💀 Dead Letter Queues (DLQ)
  • ⚡ Event-driven communication
  • 🧱 Scalable microservices

🧠 What You’ll Build

A real-world system with services like:

  • authentication
  • rbac
  • quotation
  • packages
  • generate-quotation-pdf
  • generate-quotation-ai
  • subscription-plans
  • user-subscriptions
  • notifications
  • send-otp

🏗️ Architecture Overview 

API Gateway → RabbitMQ → Microservices (Consumers)
Enter fullscreen mode Exit fullscreen mode

👉 No direct service-to-service calls
👉 Fully async & decoupled

📁 Monorepo Structure 

nestjs-microservices/
│
├── apps/
│   ├── api-gateway/
│   ├── authentication/
│   ├── rbac/
│   ├── quotation/
│   ├── packages/
│   ├── generate-quotation-pdf/
│   ├── generate-quotation-ai/
│   ├── subscription-plans/
│   ├── user-subscriptions/
│   ├── notifications/
│   └── send-otp/
│
├── libs/
│   ├── rabbitmq/
│   ├── common/
│   ├── dto/
│   ├── config/
│   └── utils/
Enter fullscreen mode Exit fullscreen mode

👉 Clean separation = easy scaling

📡 Messaging Setup (Core Idea)

Instead of this ❌

await this.authService.validateUser();
Enter fullscreen mode Exit fullscreen mode

We do this ✅

this.client.send('auth.validate', payload);
Enter fullscreen mode Exit fullscreen mode

👉 Async, scalable, resilient

⚙️ RabbitMQ Configuration (DLQ + Retry) 

ClientsModule.register([
  {
    name: 'RABBITMQ_SERVICE',
    transport: Transport.RMQ,
    options: {
      urls: [process.env.RABBITMQ_URL],
      queue: 'main_queue',
      queueOptions: {
        durable: true,
        deadLetterExchange: 'dlx',
        deadLetterRoutingKey: 'dlq',
      },
    },
  },
]);
Enter fullscreen mode Exit fullscreen mode

🔁 Retry Mechanism (The Secret Sauce) 

Main Queue → Retry Queue (TTL) → Main Queue → DLQ
Enter fullscreen mode Exit fullscreen mode

Retry Queue Example: 

{
  queue: 'quotation_retry',
  queueOptions: {
    messageTtl: 5000,
    deadLetterRoutingKey: 'quotation_queue',
  },
}
Enter fullscreen mode Exit fullscreen mode

👉 If a job fails → wait 5s → retry automatically

💀 Dead Letter Queue (DLQ)

When retries fail:

@EventPattern('quotation.dlq')
handleFailed(data: any) {
  console.error('DLQ Message:', data);
}
Enter fullscreen mode Exit fullscreen mode

👉 No silent failures anymore 🔥

📤 Producer Example 

@Injectable()
export class QuotationProducer {
  constructor(@Inject('RABBITMQ_SERVICE') private client: ClientProxy) {}

  createQuotation(data: any) {
    return this.client.emit('quotation.create', data);
  }
}
Enter fullscreen mode Exit fullscreen mode

📥 Consumer Example 

@EventPattern('quotation.create')
async handle(data: any, context: RmqContext) {
  const channel = context.getChannelRef();
  const message = context.getMessage();

  try {
    await this.service.process(data);
    channel.ack(message);
  } catch (e) {
    channel.nack(message, false, false); // goes to DLQ
  }
}
Enter fullscreen mode Exit fullscreen mode

⚡ Event-Driven Wins

Traditional API Event-Driven
Tight coupling ❌ Loose coupling ✅
Blocking ❌ Async ✅
Hard to scale ❌ Horizontally scalable ✅

🔥 Real Use Cases

  • 📩 send-otp → async SMS
  • 📄 generate-quotation-pdf → heavy processing
  • 🤖 generate-quotation-ai → AI calls
  • 🔔 notifications → emails, push

👉 Your API stays FAST ⚡

🧠 Production Best Practices

  • ✅ Use idempotency keys
  • ✅ Add correlation IDs
  • ✅ Monitor DLQ actively
  • ✅ Scale consumers independently
  • ✅ Separate queues per service
  • ✅ Validate DTOs strictly

🐳 Bonus: Run Everything with Docker 

rabbitmq:
  image: rabbitmq:3-management
  ports:
    - "5672:5672"
    - "15672:15672"
Enter fullscreen mode Exit fullscreen mode

👉 Access dashboard → http://localhost:15672

💣 Hard Truth

If your system doesn’t handle failure… it’s already broken.

🚀 Final Takeaway

Build microservices that don’t just “work”… but recover, retry, and scale automatically.

🧠 TL;DR

  • Use NestJS + RabbitMQ
  • Prefer event-driven architecture
  • Implement retries + DLQ
  • Scale consumers independently
  • Never ignore failures

🔥 Want More?

If this helped you, I can next show:

  • ☸️ Kubernetes deployment (auto-scaling microservices)
  • 📊 Monitoring with Prometheus + Grafana
  • 🔐 Full auth + API Gateway setup

Just comment: “PART 2” 👇

Top comments (0)