Ok, after upgrading to the latest stable release and building/deploying the debug version (with a few small modifications to the debug logging to make it easier to track the time in msecs of various log statements), I determined that the issue is the lingering close. Apparently, the $request_time variable that i had relied upon to know how long nginx took to process the request is misleading, becby rich l - Migration from Other Servers
As I've now done some more investigation, i wanted to post the additional information i've collected. First, i want to note that this happens on a server without significant load and that is definitely not swapping. Second, i've modified the logging to collect the $upstream_response_time as well as the $request_time. I see numerous cases where the upstream_response time is approximatelby rich l - Migration from Other Servers
Hi, I'm a big fan of nginx and have migrated some of my appservers to use nginx, while leaving some on apache 2.2 for comparison. I'm running ruby/rails on mongrels behind the nginx/apache frontend webserver. (Note: i'm not using passenger on either.) I have one problem to resolve before I can switch over to nginx completely. Hopefully someone here can help! I've noticed that the averageby rich l - Migration from Other Servers