Files

T

Unit Tests / test (push) Successful in 10s

Details

Dockerize Django app with dev/beta/prod env config and uv (#6 )

## Summary

- Containerize the Django app with Docker and docker-compose (dev + production)
- Refactor settings into `dev` / `beta` / `prod` environments driven by environment variables
- Connect to PostgreSQL via `DATABASE_URL` or `DB_*` vars
- Migrate package management from pip to uv (`pyproject.toml`, `uv.lock`)
- Split Gitea workflows: PRs run unit tests only; pushes to `master` run tests, Docker validation, and deploy
- Update deploy script to rsync code, preserve server `.env`, validate config, and run Docker compose

Closes #4

## Test plan

- [x] `uv run python manage.py test` passes locally (10/10)
- [x] `DJANGO_ENV=beta` and `DJANGO_ENV=prod` load with correct logging levels
- [x] `scripts/validate-env.sh` rejects missing production variables
- [ ] `docker compose up --build` starts app + Postgres locally
- [ ] Containerized unit tests pass in CI Docker job
- [ ] Server `.env` created from `.env.prod.example` before first production deploy
- [ ] CI workflow runs on this PR (tests only, no deploy)

Reviewed-on: #6

2026-07-07 11:23:50 -07:00

1.7 KiB

Raw Blame History

company_site

Django site for AIML Operations.

Local development (uv)

uv sync
cp .env.example .env
cd company_site
DJANGO_ENV=dev uv run python manage.py migrate
DJANGO_ENV=dev uv run python manage.py runserver

Docker (dev + Postgres)

cp .env.example .env
docker compose up --build

App: http://localhost:8000

Environments

Set DJANGO_ENV to one of:

Value	DEBUG default	Logging level
`dev`	true	DEBUG
`beta`	false	INFO
`prod`	false	WARNING

Secrets and service config come from environment variables. See .env.example.

CI / deploy workflows

Workflow	Trigger	What runs
`.gitea/workflows/ci.yml`	Pull requests to `master`	Unit tests only
`.gitea/workflows/deploy.yml`	Push to `master`	Unit tests → Docker build/test → deploy

Deploy never runs on pull requests. Uses separate workflow files (not job if conditions) so Gitea runners handle it reliably.

Production deploy

Server keeps its own .env at the live site path. Deploy rsyncs code but never overwrites .env.

On the server, copy .env.prod.example to .env and fill in production values.
Run bash scripts/validate-env.sh /path/to/.env to verify required variables.
Push to master — the deploy workflow runs scripts/deploy.sh, which:
- rsyncs checkout to live site (preserving .env)
- validates environment variables
- docker compose -f docker-compose.prod.yml build
- docker compose up -d
- runs migrations in the web container

Legacy venv/systemd deploy: DEPLOY_MODE=legacy bash scripts/deploy.sh <checkout>.

1.7 KiB Raw Blame History