Currently we have a single Nginx web sever connecting to a pool of PHP
app servers. The response from the application layer (PHP / DB) is
excellent, but our site load times are not great. We’ve run some
analytical tools and the network / web server response time is where a
significant bottleneck exists.
In working with our managed hosting company on this issue, we ran the
following shell commands. I was wondering if someone could give me an
idea if these numbers are indicative of a configuration issue with OS
and / or Nginx OR will more CPU power help with this problem OR maybe we
need to load balance the traffic to another web server.
Should Nginx be able to handle this kind of traffic in general? Is there
something that can be done in terms of Nginx configuration or OS
configuration that can improve this issue (i.e. > 7,000 connections
waiting)?
root@server1 [~]# netstat -anp | grep ‘:80’ | grep ESTABLISHED -c
4785
Over 4000 current connections.
root@server1 [~]# netstat -anp | grep ‘:80’ | grep -i WAIT -c
7982
7000 Connections in waiting for data state.
root@server1 [~]# netstat -anp | grep ‘:80’ | grep -i SYN -c
280
Posted at Nginx Forum: