Poor NGINX Performance due to configurations

Aleksandar Lazic al-nginx at none.at
Fri Oct 9 22:27:57 UTC 2015


Hi.

Am 09-10-2015 09:09, schrieb Payam Chychi:
> On Thursday, October 8, 2015 at 3:13 PM, Naseem Al-Naji wrote:
> 
>> Hey there! Yes, I believe it is mainly just how we have configured it.
>> The NGINX performs much better on our staging servers even though they
>> only have 1 worker process, whereas our production has 4 worker
>> processes, where 4 is recommended.
>> 
>> On Thu, Oct 8, 2015 at 2:41 PM, Payam Chychi <pchychi at gmail.com>
>> wrote:
>> 
>> On 2015-10-08, 1:23 PM, Naseem Al-Naji wrote:
>> 
>> Hi everyone,
>> 
>> My company, Mulesoft, uses NGINX to help deliver our product. We use
>> your commercialized version, and have noticed poor performance on our
>> production servers, whereas our staging servers perform much better.
>> The load balancing on production causes jumps in CPU:
>> 
>> <Untitled>
>> 
>> I was wondering if we could arrange a call to help us diagnose this
>> problem. I've attached our nginx.conf that is running on an M3.XLarge
>> EC2 instance.
>>> 
>> Hi,
>> 
>> I assume this is not due to some system tune setting or IO issues?
> 
> Can you please share you system tune settings with us, such as output
> for ulimit -a as the user running the nginx process as well as sysctl
> and output of ethreal on your nic(s).
> 
> Output of dmesg also helps in case there are hardware issiues.
> 
> Another tool to use is dstat, it helps you monitor IO and 
> network/system
> resource usage.

and the output from 'nginx -V'.
I case you have a new version the output of -T also.

BR Aleks



More information about the nginx mailing list