Redeploy Django projects using management commands
Using Fabric to run a management command on each remote instance to redeploy itself.
The remote command will pull the source code from Git, run all necessary Django management commands, and restart the server.
It's intended for small to medium scale apps.
It's faster. When the you run everything remotely it only takes about 10 seconds to deploy an instance.
git pull
pip install -r requirements.txt
syncdb
migrate
collectstatic
restart server (webserver dependent)
- A deployed instance running either:
- Apache with mod_WSGI
- GUnicorn
- Fabric
- A virtualenv dedicated to the project.
- A Git hosted repository for the project
Add the app to your INSTALLED_APPS
setting:
INSTALLED_APPS = ( #... 'sitepush', )
Add the server configuration to your settings file:
DEPLOYS = { 'default': { 'HOST': '192.168.0.1', # IP or domain name 'USER': 'user', # User to login with 'BRANCH': 'master', # Git branch to pull updates from 'SETTINGS': 'settings', # Settings file to use (optional) 'DIR': '/var/www/myproject/', 'ENV': 'myvirtualenv', # virtualenv used in the project 'WEBSERVER': 'gunicorn', # webserver - 'apache' or 'gunicorn' 'PID_FILE': '/tmp/gunicorn.pid', # Gunicorn pid file path 'WSGI_FILE': 'deploy/wsgi.py', # Apache wsgi file path }, }
The PID_FILE
/ WSGI_FILE
settings should be set depending on the web
server you're using.
BRANCH
is optional and defaults to master
python manage.py deploy_remote default # Or python manage.py deploy_remote srv1 srv2 #...' # Don't install requirements: python manage.py deploy_remote default --noreqs # Deploy all remote instances: python manage.py deploy_remote --all