Graphite Message Lab

An intentionally small, interactive application for exercising Project Graphite's complete deployment path.

Visitors send a message privately or ask for it to be published. Django validates it through a JSON API, PostgreSQL persists it, Redis delivers a background job, and a Celery worker counts its words. A message appears on the public wall only when its sender asked for publication and an administrator publishes it in /admin/; the browser polls the API to refresh the wall.

Stack

• Frontend: server-rendered HTML, Tailwind CSS, and a small JavaScript client
• Backend: Django and Gunicorn
• Database: PostgreSQL
• Queue: Redis
• Worker: Celery
• Local packaging: Docker and Docker Compose
• Delivery: GitHub Actions, GitHub Container Registry, Coolify, and Docker Compose

Run locally

Docker with Compose v2 is the only prerequisite.

cp .env.example .env
docker compose build
docker compose up
docker compose exec web python manage.py migrate

Open http://localhost:8000, choose "Public after review" and submit a message. It stays private until you publish it from /admin/ (create an account with ./run manage createsuperuser), and then it appears on the public wall within 15 seconds.

Useful checks:

./run manage test
./run lint
curl --fail http://localhost:8000/up/databases

API

GET /api/messages/ returns the latest 25 published messages. POST /api/messages/ accepts:

{
  "display_name": "Ada",
  "body": "Hello Graphite",
  "requested_visibility": "public"
}

requested_visibility defaults to "private".

Messages are stored and read by the administrator. Do not enter private or sensitive information.

Deployment

Production metadata is registered in projects/graphite-message-lab.yml in the private platform repository. Coolify deploys compose.production.yaml, which pulls the prebuilt GHCR images, limits memory and CPU, persists PostgreSQL and Redis in named volumes, runs database migrations before Gunicorn starts, and exposes only the web service through Coolify's proxy. VPS setup, secret entry, deployment, rollback, backup, and cleanup are documented in the organisation's private operations runbook.

Attribution

The Docker foundation was adapted from Nick Janetakis' MIT-licensed docker-django-example. See UPSTREAM.md and LICENSE.