Trouble with somewhat complex proxy config

[Maxim Dounin]

Hello!

On Wed, Dec 14, 2011 at 05:42:35PM -0800, Daniel Jabbour wrote:

> Hi,
> 
> So basically here is what I'm trying to accomplish:
> 
> 1. I have several different rails apps running inside passenger 
> standalone instances. I'd use the module, but I need to have 
> separate ruby versions / parameters for each, so they're happily 
> on standalone instances.
> 
> 2. The standalone instances run fine and I can connect to them.
> 
> 3. Based on the presence of an argument "needle" in the GET 
> params, I want to proxy pass to different locations. If the 
> $needle is present, I want to proxy to $needle.somehost.com. 
> Otherwise I want to proxy to localhost:3000. I know this sounds 
> a little crazy, and I'd rather not get into "why," suffice it to 
> say it's a pretty unique setup.
> 
> Here is my config, though it isn't quite working out. I get 
> internal errors from nginx after a long timeout.
> 
> server {
>   listen host.ip.address:443 ssl;
>   server_name  host.server.com;
>   root /var/www;
>  
>   ssl_certificate /my.crt;
>   ssl_certificate_key /my.key; 
>   
>   error_page 500 502 503 504	/50x.html;
>   location = /50x.html {
>     root   html;
>   }
>   
>   location / {
>     resolver dns.ip.address;
>     if ($args ~* needle=([a-z0-9\-]+)\&? ) {
>       set          $needle         $1;
>       proxy_pass   https://$needle.somehost.com;
>       break;
>     }
>     proxy_pass http://127.0.0.1:3000;
>     proxy_set_header Host $host;
>   }
> }
> 
> Any assistance would be greatly appreciated.

What's in error log?

Maxim Dounin