Imported from shipshitdev/skills (skills/nestjs-queue-architect/SKILL.md). Install upstream with npx skills add shipshitdev/skills --skill nestjs-queue-architect. Copyright stays with the author.

NestJS Queue Architect - BullMQ Expert

Design resilient, scalable BullMQ + NestJS job processing systems for media-heavy workflows.

Technology Stack

• BullMQ: 5.61.0 (Redis-backed job queue)
• @nestjs/bullmq: 11.0.4
• @bull-board/nestjs: 6.13.1 (Queue monitoring UI)

Project Context Discovery

Before implementing:

1. Check .agents/memory/ for any architecture files describing queue patterns
2. Review existing queue services and constants
3. Look for [project]-queue-architect skill

Core Patterns

Queue Constants

export const QUEUE_NAMES = {
  VIDEO_PROCESSING: 'video-processing',
  IMAGE_PROCESSING: 'image-processing',
} as const;

export const JOB_PRIORITY = {
  HIGH: 1,    // User-facing
  NORMAL: 5,  // Standard
  LOW: 10,    // Background
} as const;

Queue Service

@Injectable()
export class VideoQueueService {
  constructor(@InjectQueue(QUEUE_NAMES.VIDEO) private queue: Queue) {}

  async addJob(data: VideoJobData) {
    return this.queue.add(JOB_TYPES.RESIZE, data, {
      priority: JOB_PRIORITY.NORMAL,
      attempts: 3,
      backoff: { type: 'exponential', delay: 2000 },
    });
  }
}

Processor (WorkerHost)

@Processor(QUEUE_NAMES.VIDEO)
export class VideoProcessor extends WorkerHost {
  async process(job: Job<VideoJobData>) {
    switch (job.name) {
      case JOB_TYPES.RESIZE: return this.handleResize(job);
      case JOB_TYPES.MERGE: return this.handleMerge(job);
      default: throw new Error(`Unknown job: ${job.name}`);
    }
  }
}

Key Principles

1. One service per queue type - Encapsulate job options
2. Switch-based routing - Route by job.name
3. Structured error handling - Log, emit WebSocket, publish Redis, re-throw
4. Always cleanup - Temp files in try/finally
5. Idempotent handlers - Safe to retry

Queue Configuration

BullModule.registerQueue({
  name: QUEUE_NAMES.VIDEO,
  defaultJobOptions: {
    attempts: 3,
    backoff: { type: 'exponential', delay: 2000 },
    removeOnComplete: 100,  // Prevent Redis bloat
    removeOnFail: 50,
  },
});

Retry Strategy

Common Pitfalls

• Memory leaks: Always set removeOnComplete/Fail
• Timeouts: Set appropriate timeout for heavy jobs
• Race conditions: Make handlers idempotent

For complete processor examples, testing patterns, Bull Board setup, and Redis pub/sub integration, see: references/full-guide.md