Hello everyone,<div><br></div><div>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. </div><div><br>
</div><div>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. </div>
<div><br></div><div>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.</div><div>
<br></div><div><b>/nginx_status</b></div><div><div>Active connections: 361 </div><div>server accepts handled requests</div><div> 71644 71644 195385 </div><div>Reading: 0 Writing: 13 Waiting: 348 </div>
</div><div><br></div><div><b>Gunicorn: </b><a href="http://pastebin.com/CY8m922F" target="_blank">http://pastebin.com/CY8m922F</a></div><div><b>Nginx:</b> <a href="http://pastebin.com/7bx5H33Q" target="_blank">http://pastebin.com/7bx5H33Q</a></div>
<div>
<br></div><div>Do you have any idea?</div><div><br></div><div>On the other hand; is that feasible way to deliver a web.py application?</div><div><br></div><div>Thanks.</div><div><div>-Umut</div></div>