Hello everyone,
As you can see in configuration, we have nginx servers (there are 2 of them) as reverse proxy which keep requests balanced for 6 gunicorn application(web.py framework) nodes.
In python part, we calculate execution time and we know that application part completes in ~0.02 sec per request but it takes 60-300 milliseconds to see it on browser in the beginning which is acceptable for us.
In time (~20-30 minutes), active connection number increases to 1000+ from 300+. Our python application still completes in ~0.02 sec but it takes sometimes even 30 seconds to see it on browser.
/nginx_status
Active connections: 361
server accepts handled requests
71644 71644 195385
Reading: 0 Writing: 13 Waiting: 348
Do you have any idea?
On the other hand; is that feasible way to deliver a web.py application?
Thanks.