通过Gunicorn / Nginx使用Django的站点框架运行多个站点

Question

I have a Django based CMS that uses Django's sites framework and Nginx/Apache/mod_wsgi virtual hosts to run a number of websites on different domains. We're assessing other options for a Django stack and have the CMS running with a single site on a new server with Nginx proxying to Gunicorn (gunicorn_django, specifically).

Although this works great for a single site, I'm not sure how to configure Gunicorn for multiple sites. The problem is that with Apache/mod_wsgi, we could set the DJANGO_SETTINGS_MODULE for mod_wsgi to the appropriate site's settings.py

import os, sys

def inflight(filename):
    """
    Calculate absolute path to the folder containing "myfile.wsgi", then
    append to the PYTHONPATH.
    """
    ROOT = ('/').join(os.path.abspath(os.path.dirname(filename)).split('/')[0:-1])
    sys.path.append(ROOT)
    sys.path.append(os.path.join(ROOT, 'website'))

    sys.stdout = sys.stderr
    # Each website should have a settings file: /www/mysite.com/website/settings.py
    os.environ['DJANGO_SETTINGS_MODULE'] = 'website.settings'
    import django.core.handlers.wsgi
    return django.core.handlers.wsgi.WSGIHandler()

At the moment I'm thinking that I have to have a different instance of Gunicorn for each virtual host site we run but that seems overkill for the traffic we get to most of our sites.

Does anyone run Gunicorn with Django's sites framework and can give a hint to how it's configured?

Answer 1

I had the same problem and stumbled upon this question in search of the same answer.

I know the question is old and you've probably figured it out by now, but since it might be useful for someone else, here's how I solved it:

• You do need to run separate gunicorn processes to make django sites framework work because you can only point a gunicorn instance at one settings.py file. If you're sites don't get much traffic, I'd only create 1 or 2 gunicorn workers per site. (I know, still probably overkill).

• Ideally you would want to manage these different processes with something like supervisord to make it easier to manage starting/stoping/restarting your different sites, but I couldn't get it working.

First, start up your gunicorn servers on local host at different ports using the command line. Ex:

gunicorn_django -w 2 -b 127.0.0.1:1000 /path/to/my/django/project/site1_settings.py --daemon

gunicorn_django -w 2 -b 127.0.0.1:1001 /path/to/my/django/project/site2_settings.py --daemon

You now have 2 django sites running on localhost at ports 1000 and 1001 (you can use whatever ports suite you).

Now you need to create two separate nginx server configuration files to point each domain name at it's respective django site. Ex:

server {
    listen 80;
    server_name website1.com;
    client_max_body_size 4G;

    keepalive_timeout 4;

    location /media {
        root /path/to/my/django/project/media; #this servers your static files
    }

    location / {
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header HOST $http_host;
        proxy_redirect off;

        if (!-f $request_filename){
            proxy_pass http://127.0.0.1:1000; #point to the django site already running on local host
            break;
        }
    }

    #add in any additional nginx configuration such as support for 500 errors or proxy apache to server php files etc.
}

Then create a duplicate of the nginx configuration for your 2nd site, but change the server name and the proxy_pass to the values for site 2.

Make sure your server configuration files are included in the main nginx.conf file.

Reload nginx and you should be good to go.

If anyone has an easier/better way to go about this please post it.