Running

1. Docker Setup

   • Ensure you have Docker and Docker Compose installed.
   • Build and start the containers:

     docker-compose down # before restart
     docker-compose up --build

2. Access the Services

   • The Command Service will be available at http://localhost:3000.
   • Worker Services will be running on ports 3001, 3002, and 3003.

3. API Endpoints

   • Submit Code: POST /submit-code

     curl --location --request POST 'http://localhost:3000/submit-code' \
     --header 'Content-Type: application/json' \
     --data-raw '{
         "code": "print(\"Hello, World!\")"
     }'

   • Get Execution Result: GET /exec-result?jobId={jobId}

     curl --location --request GET 'http://localhost:3000/exec-result?jobId={jobId}'

4. Monitoring Queues

   • Bull Board is set up to monitor the queues.
   • Access Bull Board at http://localhost:3000/admin/queues.

5. Load Testing

   • Ensure you have artillery installed.
   • Run the testing:

      artillery run command/artillery-config.yaml

Programming Design

NodeJS Service That Executes Code, currently supported only Python code TODO: other codes execution

Overview

The system allows frontend clients to submit Python code for execution on the server-side. It consists of two microservices: the Command Service and the Worker Service. The Command Service handles code submission and job management, while the Worker Service executes the submitted code asynchronously.

Components

1. Command Service

   • Responsible for receiving code submissions from the frontend and managing job execution.
   • Stores submitted Python code and job metadata in MongoDB.
   • Enqueues job IDs into the code-save queue (Redis) for asynchronous processing by the Worker Service.
   • Provides RESTful APIs for code submission and result retrieval.

2. Worker Service

   • Listens to the code-save queue for new job IDs.
   • Saves Python code as a file and enqueues the job ID into the code-exec queue.
   • Listens to the code-exec queue for job execution.
   • Executes the Python code, captures output/errors, and stores the results in MongoDB.
   • Continuously processes incoming jobs from both queues.

3. Databases

   • MongoDB: Stores job metadata (job ID, status, etc.) and execution results (output, errors, etc.).

4. Queues

   • Redis: Manages the code-save and code-exec queues for asynchronous job processing.

Workflow

1. Code Submission Workflow

   • Frontend submits Python code to /submit-code endpoint of Command Service.
   • Command Service stores the code in MongoDB, generates a job ID, and enqueues the job ID in the code-save queue.
   • Returns the job ID to the frontend.

2. Execution Result Retrieval

   • Frontend polls /exec-result?jobId={jobId} endpoint of Command Service to check job status.
   • Command Service retrieves execution results from MongoDB once available.

3. Worker Service Workflow

   • Code Save Workflow:
     • Worker Service listens to the code-save queue for new jobs.
     • Upon receiving a job ID, saves Python code as a file.
     • Enqueues the job ID in the code-exec queue.
   • Code Exec Workflow:
     • Worker Service listens to the code-exec queue for new jobs.
     • Upon receiving a job ID, retrieves the Python code file and executes it.
     • Captures output/errors and stores results in MongoDB.

Security Considerations

• Code Execution Sandbox: Execute Python code in a secure sandbox (e.g., Docker containers) to prevent malicious activities.
• Input Validation: Validate submitted Python code to mitigate injection attacks.
• Access Controls: Restrict access to APIs and ensure secure communication (HTTPS).

Scalability and Performance

• Horizontal Scaling: Both services can be scaled independently to handle increasing code submissions and execution demands.
• Queue Management: Efficient queue handling ensures jobs are processed promptly.

Technologies Used

• Node.js (NestJS): For building scalable and maintainable backend services.
• MongoDB: NoSQL database for storing job metadata and execution results.
• Redis: Message queue for managing job queues (code-save and code-exec).
• Docker: Containerization for code execution sandboxing.

Conclusion

This architecture ensures efficient and scalable execution of Python code submitted from the frontend, providing reliable and timely results to users.