There are a few ways to set this up. I've tried a couple of them out
personally and from my past experience I would say that the next step
depends upon how you have multitenancy set up.
Typically I will use multiple django/mezzanine projects installed on the
same server, in completely separate dirs. Each project has a domain name
(and an elastic ip associated with it), which is where nginx comes into the
mix. Inside my nginx config I have several upstream servers defined. I use
one virtualhost for each of the projects plus whatever utility servers that
I might need.
example upstream setup:
upstream GUnicornSite1 {
server 127.0.0.1:8001;
}
upstream GUnicornSite2 {
server 127.0.0.1:8002;
}
upstream GUnicornSite3 {
server 127.0.0.1:8003;
}
Now that nginx knows about the webservers, its time to actually start those
webservers. Its important to note that in your gunicorn configuration, you
need to make sure that each server is listening on whichever port number
you specified in your nginx.conf definition. Aside from any static media
nonsense that may or may not arise, I don't think there is much more to
specify as far as the gUnicorn config. I made the move to uwsgi because I
needed websocket support.
The uwsgi part of my nginx.conf looks significantly different than what a
gunicorn setup needs, but here is a simple example of how I listen on a
certain port.
Server{
listen 9001;
location / {
proxy_pass http://GUnicornServer1/;
Server{
listen 9002;
location / {
proxy_pass http://GUnicornServer2/;
You can also specify that you want to listen for a certain domain/ip
address, the syntax is a little tricky so i would suggest finding a good
tutorial that meets your needs. So in the above case, if a request is sent
to port 9002, it will be forwarded to whichever website is running on the
9002 server. Again, instead of using port 9002 you could just as easily
route the request based on the ip. I use this setup with uwsgi's emperor
mode because it makes managing the vassal servers simple.
If you, instead, want to host multiple sites using multi-tenancy over a
single webserver then I would consider using middleware to route incoming
urls to the appropriate urls.py. nginx should not need any additional
configuration, it just needs to route all the requests to the one
webserver. Routing then happen when the request is processed by the
middleware.
There's probably an easier way to do this too that I'm not aware of. I've
done subdomain routing like this in the past, but it felt kind of hacky. I
feel like nginx is the more capable solution. I might be deploying a site
later today that involves this sort of setup. If i get around to it ill
post the actual code chunks.
best,
avg
--
You received this message because you are subscribed to the Google Groups
"Mezzanine Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to mezzanine-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
