Quick installation

Quick installation#

This is the fastest way to get the Grader platform running locally from source. It uses the repository’s Makefile targets, which wrap uv and the workspace packages so you don’t have to run commands manually in each package.

Warning

This setup is intended for local development and testing purposes only. It is not suitable for production use. The local autograding executors run student code with full access to the service database and filesystem, so a malicious submission could read or alter grades and other students’ work. Use the Kubernetes installation with KubeAutogradeExecutor for any deployment that grades untrusted student code.

Prerequisites#

  • Python 3.9+

  • uv (Python package manager)

  • Node.js 20+ (only needed when rebuilding the labextension frontend)

  • Git

1. Clone the repository#

git clone https://github.com/TU-Wien-dataLAB/grader.git
cd grader

The repository is a monorepo containing both the grader-service backend (packages/service) and the grader-labextension frontend (packages/labextension). You no longer need to clone them separately.

2. Install dependencies#

make sync

This runs uv sync --all-packages --all-groups, which creates a single virtual environment with both packages installed in editable mode plus all development, test, and documentation dependencies.

3. Create the database#

The repository ships example configuration files in dev/local/token/. Create the database schema by running the migration against the service config:

uv run grader-service-migrate -f dev/local/token/grader_service_config.py

The default config uses an embedded SQLite database and runs autograding tasks eagerly, so no external database or message broker is required for local testing.

4. Start Grader Service and JupyterHub#

Start the Grader Service:

make run-service

Grader Service runs at http://127.0.0.1:4010.

In a separate terminal, start JupyterHub:

make run-hub

JupyterHub will be running at http://127.0.0.1:8080.

Note

Start the Grader Service before JupyterHub.

Full stack with Docker Compose#

If you prefer a fully containerized environment (including RabbitMQ, a Celery worker, and hot-reload), use:

make dev-up

See the Docker installation guide for details.

Cleanup#

To remove build artifacts and caches:

make clean

This does not delete the virtual environment. To recreate it from scratch, run make sync again.