ERROR: Invalid HTTP_HOST header: '/webapps/../gunicorn.sock'

Question

After I deployed my Django App last night I got tons of strange Emails saying:

ERROR: Invalid HTTP_HOST header: '/webapps/example_com/run/gunicorn.sock

I'm sure this is somehow related to the following nginx config:

upstream example_app_server {
  server unix:/webapps/example_com/run/gunicorn.sock fail_timeout=0;
}

server {

    listen   80;
    server_name example.com;

    client_max_body_size 4G;

    access_log /webapps/example_com/logs/nginx-access.log;
    error_log /webapps/example_com/logs/nginx-error.log;


    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://example_app_server;
            break;
        }
    }
}

Answer 1

I found the answer to my my question in a django bug report.

    proxy_set_header Host $http_host;

has to be replaced with:

    proxy_set_header Host $host;

to make nginx pass the correct headers from that on instead of the gunicorn socket the requested page was in the django alerts.

Answer 2

This person explains a bit more what is going on based on this very same post. Here's his/her explanation:

...when a request is made to the server and the HTTP Host is empty, nginx sets the HTTP host to the gunicorn sock.

I can generate this error using curl:

curl -H "HOST:" MY_DOMAIN_NAME -0 -v

This sends a request without a HTTP Host. The -0 causes curl to use HTTP version 1.0. If you do not set this, the request will use HTTP version 1.1, which will cause the request to be rejected immediately and not generate the error.

The solution is to replace $http_host with $host (as pointed out on Stackoverflow). When HTTP Host is missing, $host will take on the value of the “server_name” directive. This is a valid domain name and is the one that should be used.