When you say it's not running on your server, what's the output from:
ps aux | grep nginx (any processes running?)
[root@msweb01 ~] $ ps aux | grep "^nginx"
nginx 24214 0.0 0.2 1037044 40580 ? S Mar07 0:00 php-fpm: pool default-webnode-check
...
nginx 30379 0.0 0.0 60336 4524 ? S 11:18 0:04 nginx: worker process
lsof -Pi | grep LISTEN (any processes listening on 80/443 or your web port?)
[root@msweb01 ~] $ lsof -Pi | grep LISTEN
nginx 1046 root 13u IPv4 35851146 0t0 TCP msweb01.dev:81 (LISTEN)
php-fpm 24212 root 7u IPv4 31168887 0t0 TCP localhost:9001 (LISTEN)
nginx 30377 nginx 13u IPv4 35851146 0t0 TCP msweb01.dev:81 (LISTEN)
/etc/init.d/nginx status (depending on your systems initd script)
[root@msweb01 ~] $ /etc/init.d/nginx status
nginx (pid 1046) is running...
As Michael Hampton noted, I'd first try:
/etc/init.d/nginx restart to properly bump that nginx server, then depending on your logging options set in nginx.conf and related configs, I'd tail them and work out where your web request is going.
I'm probably completely wrong here, but I had something similar to this recently when testing (and thus swapping) between apache and nginx, in which Apache/PHP had written template config cache to the file system, then starting nginx (which worked fine and worked as expected) later threw an error when hitting the dame template as the nginx user didn't have access to access/update the template cache. Might be something to look at anyway.