Performance: the response times of Django's WSGI applications soar under a heavy load

We have a Django application that runs with mod_wsgi and WSGIDaemonProcess. We are now receiving a lot of traffic and response times have skyrocketed, but CPU usage on our servers is low, around 5-10%.

I can not find out where the bottleneck is. After a short time we get this error in the apache error log:

Timeout when reading the response headers of the daemon process & # 39; wsgi & # 39;

Happy to provide more records, this is the configuration of the Daemon process:

WSGIDaemonProcess wsgi processes = 100 threads = 10 maximum requests = 500 display name =% {GROUP}