-
Notifications
You must be signed in to change notification settings - Fork 6
Installation without docker
- Python (any version matching the ones specified here
grep python-version .github/workflows/django.yml). - Virtualenv (e.g. python3-venv package from your distro's repositories).
- PostgreSQL database (Django framework will administer tables inside it).
- E-mail server, Open port, TLS certificates, DNS routing, Reverse Proxy, Backup strategy, etc.
Execute these commands as the user that will be running/ exposing the application.
Note that the default branch is 'develop'. If you wanted to get one of the stable releases, please do so by adding --branch 1.0.0 to git-clone. Do so with an appropriate version number from releases.
git clone --depth=1 git@github.com:maxplanck-ie/parkour2.git
python3 -m venv env_app
source ./env_app/bin/activate
pip install -r parkour_app/requirements/prod.txt
Now the core installation is done, you may move forward.
We assume some environmental variables are set (e.g. LANG, LC_TIME, TZ, etc.) ... Next, you'll need to set further environmental variables (use export so that these are inherited by downstream processes). Take a look at misc/parkour.env to get an idea of what you'll need (e.g. DATABASE_URL is read by Django to authenticate with your PostgreSQL instance). Most of these variables may be skipped, we have added some sane defaults into the loading instructions at parkour_app/settings/wui/base.py (e.g. using SQLite if DATABASE_URL is unset).
Next, change directory into the parkour_app/ subfolder. From now on, all commands will assume this positioning.
At last, run the application (in this example, using 4 workers and binding it to port 8000 of all available IP addresses, to be taken care afterward by a web server running as a reverse proxy). See Green Unicorn's Official Documentation for further details.
gunicorn wui.wsgi:application -t 600 -w 4 -b :8000
At this point, the database is empty. Before populating it, we'd need to define the schema (e.g. creating empty tables, defining relation between columns/ foreign keys, etc.) For this, run python manage.py migrate --traceback. The command will run all the django-generated code matching the Model definitions of our application (e.g. files matching parkour_app/**/migrations/*.py)
Optinally, pre-load some data: python manage.py load_initial_data. You may edit fixtures manually (e.g. files matching parkour_app/**/fixtures/*.json). Please note that since the primary keys (pk) are defined in such files; every time you re-load these JSON Fixtures, the database is gracefully overwritten on these items. You might prefer saving customizations with pg_dump.
Optionally, create an admin account with the following command: python manage.py createsuperuser. You'll be asked for an email account, and a password (twice). Login, remove some of the default users that were pre-loaded and create new normal and staff users as you see fit... These default credentials may come in handy,
-
parkour-staff@parkour-demo.ie-freiburg.mpg.dewith passwordparkour-staff. -
parkour-admin@parkour-demo.ie-freiburg.mpg.dewith passwordparkour-admin.
Finally, further populate your database according to your needs (see the user manual for details, or check Django's loaddata administrative command). Use the cog icon in the top-right corner to access the 'Site Admin' panel.
Add any required cronjobs for a proper backup strategy. In there, you should also add: python manage.py clearsessions (mind the paths to proper python interpreter/ venv, and your application's manage.py location).
Take a look at other administrative commands, like django-admin dumpdata and/ or the database pg_dump & pg_restore. Even if you're not using docker, running make help for checking the essentials is highly recommended. And, upon the rules listed there, you may always execute in dry-run mode (e.g. make -n save-postgres) to see the sequence of instructions.
Of course, make shall be run from the repository root, one level up from your current work directory, parkour_app.