Issue with auto subdomain nd trailing slash

Hi All,

I am running into a weird issue, after having configured nginx 0.7.67
(comes
with debian 6) for serving PHP pages, with automatic subdomains.
It works fine for the most part, but when requesting a url that
correspond
to an actual folder, but without a trailing slash, the subdomain
recognition
fails.

Here is the start of the config:

server {

listen 80;

# base domain
set $root_domain XXXX.net;

server_name XXXX.net *.XXXX.net;

if ($host ~* ^www\.XXXX\.net$) {
  rewrite ^(.*) $scheme://XXXX.net$1 permanent;
  break;
}

set $sub_domain "";
if ($host ~* ^(.*)\.XXXX\.net$) {
  set $sub_domain $1;
}

if ($sub_domain) {

debugging instruction: figure out what’s the value of $host

rewrite ^ Google permanent;
break;
set $root_path /var/www/$root_domain/$sub_domain/public;
}

if ($sub_domain = "") {

debugging instruction: figure out what’s the value of $host

rewrite ^ Google permanent;
break;
set $root_path /var/www/$root_domain/www/public;
}

#if the directory doesn’t exist, prevent access
if (!-d $root_path) {
return 403;
}

#if we have made it here, set the root to the computed directory
root $root_path;

# logging
access_log  /var/log/nginx/$sub_domain.$root_domain.access.log;
error_log   /var/log/nginx/$sub_domain.$root_domain.error.log;

location / {
    index  index.html index.php;

    # serve static files that exist without running other rewrite 

tests
if (-f $request_filename) {
expires 30d;
break;
}

    # send all non-existing file or directory requests to index.php
    if (!-e $request_filename) {
        rewrite  ^/(.*)  /index.php  last;
    }
}

# pass the PHP scripts to FastCGI server listening on 127.0.0.1:9000
try_files $uri =404;
location ~ \.php$ {
    include /etc/nginx/fastcgi_params;
    fastcgi_pass   127.0.0.1:9000;
    fastcgi_index  index.php;
    fastcgi_param  SCRIPT_FILENAME  $root_path$fastcgi_script_name;
}


# optional authentication
# would require to be put in its own server block
# if only valid for some sub-domais
#location / {
#      auth_basic            "Authentication Required";
#      auth_basic_user_file  $root_path/private/authfile;
#}


location ~ /\.ht{
    deny all;
}

}

With this setup, you can access XXXX.net and any subdomain is served
based
on the corresponding folder under web root. All of that works very fine
except in one case:

If you request: xxxx.net - This website is for sale! - xxxx Resources and Information., and /admin/ is a folder
under
the subD1 web site, everything goes well, the index file (index.php) is
served.
Now if one requests xxxx.net - This website is for sale! - xxxx Resources and Information. (without trailing
slash),
the sub domain is not recognized anymore, and the root path is set to
the
path to XXXX.net instead of that associated with subD1.

Now the real weird thing is that it is the $host variable that’s wrong.
In
the first case (request for a folder with trailing slash), $host
contains
subD1.XXXX.net, and the regexp matches, the sub domain is properly
calculated.
In the second case (no trailing slash), $host is wrong: it contains
XXXX.net
instead of subD1.XXXX.net, and so the regexp doesn’t match and the sub
domain is not recognized.

Please note that my question is not about trying to redirect urls with
or
without trailing slash to its counterpoint. My issue is with the
subdomain
regexp not matching because the $host value is wrong.

Any pointer as to what’s going on? Am I doing something wrong?

Thanks for any pointer and best regards

Posted at Nginx Forum:

On Fri, Feb 22, 2013 at 06:27:27AM -0500, shumisha wrote:

I am running into a weird issue, after having configured nginx 0.7.67 (comes
with debian 6) for serving PHP pages, with automatic subdomains.
It works fine for the most part, but when requesting a url that correspond
to an actual folder, but without a trailing slash, the subdomain recognition
fails.

What is probably happening is that when you ask for /dir, you get a http
redirect to /dir/. That redirect must include the full url, including
the hostname. And nginx has to choose which hostname to use. And it
doesn’t choose the one that you want it to choose.

What is the output of

curl -i xxxx.net - This website is for sale! - xxxx Resources and Information.

? I guess it will show xxxx.net - This website is for sale! - xxxx Resources and Information.. You then separately
request xxxx.net - This website is for sale! - xxxx Resources and Information., and therefore shouldn’t be surprised
that the subdomain recognition fails, because in this request there is
no subdomain.

Possibly setting server_name_in_redirect
(Module ngx_http_core_module) to “off” will work for
you? The docs say that default is off, but I think that may have been
different in the 0.7 series.

f

Francis D. [email protected]

Hi Francis,

Thanks a lot for your message, I think I got things under control thanks
to
you!

You’re right, there’s indeed a 301 thrown by nginx from
xxxx.net - This website is for sale! - xxxx Resources and Information. to xxxx.net - This website is for sale! - xxxx Resources and Information.
(as shown by the curl output).

I had tried server_name_in_redirect to off before, to no avail, but I
tried
again, just in case. I looked up the documentation again, and I
realized:

  • it’s probably not active on my version, as nginx still uses the first
    server in the server_name directive regardless of the directive value.
    However, it’s a valid config value
  • but that got me thinking and I simply switched the orders of server,
    so
    that it reads:

server_name *.xxxx.net xxxx.net;

So now the automatic redirect still happens (I think it should be
configurabe, maybe it is) but by placing the *.xxxx.net server first in
the
server_name directive, the sub domain is preserved and the redirect now
happens with the correct host.

Thanks again for your help :slight_smile:

Rgds

Posted at Nginx Forum: