- Set ALLOWED_ORIGINS to production domain (fixes CORS for the web app)
- Add LOG_LEVEL=INFO to backend
- Add AutoKuma monitoring labels for Uptime Kuma auto-discovery
- Add Homepage dashboard labels (Productivity group)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Replace .gitlab-ci.yml with .gitea/workflows/deploy.yml
- Fix Dockerfile.backend: add scan_processor.py and searcher_v3.py to
COPY command (they were missing, would cause runtime ImportError)
- Update docker-compose.yml comment to reference Gitea workflow
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
On every push to main: builds both Docker images on the server via
docker compose up --build -d, prunes dangling images, and prints the
running container list. No registry required — shell executor runner
on the deployment server is all that's needed.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Single-container supervisord approach added unnecessary complexity.
Two containers is simpler and more standard:
- Dockerfile.backend: python:3.11-slim, uvicorn on port 8000
- Dockerfile.frontend: node build → nginx:alpine on port 80
- nginx.conf: proxy_pass restored to http://backend:8000
- docker-compose.yml: two services with depends_on
- Removed combined Dockerfile and supervisord.conf
Each container does one thing; logs are separate; either can be
restarted independently.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace two-container setup (separate backend + nginx frontend) with a
single image that runs both via supervisord:
- New Dockerfile: Node stage builds React, Python+nginx stage is the runtime
- supervisord.conf: manages uvicorn (api_server.py) + nginx as sibling procs
- nginx.conf: proxy_pass updated to localhost:8000 (same container)
- docker-compose.yml: simplified to one service on port 80
Deploy:
docker-compose up -d # or
docker build -t flight-radar . && docker run -p 80:80 flight-radar
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
