Nginx max connections

[Anton Yuzhaninov]

Joe Van Dyk wrote:
> On Wed, Feb 18, 2009 at 9:24 AM, Anton Yuzhaninov <citrin at citrin.ru> wrote:
>> Swish wrote:
>>> Is there any method to tell if Nginx is hitting max connections, other
>>> than getting a request timeout? Will anything appear in the logs? I ask for
>>> tuning purposes to get the optimal number of worker_processes and
>>> worker_connections. Also, at what point is it better to increase the
>>> worker_processes vs worker_connections, how can you tell what proportion is
>>> most efficient?
>> For tuning purposes it is better to use rrd-graphs.
>> Here is example (but not best) of how to setup this:
>> http://blog.kovyrin.net/2006/04/29/monitoring-nginx-with-rrdtool/lang/en/
> 
> Well, what's the best way of doing it?  :D
> 

1. Active connections and other (reading/writing) better to show on different graphs, because 
active >> reading and writing.

2. On graph with requests per second better see 3 lines - accepts, handled, requests

See attached samples for example.

-- 
  Anton Yuzhaninov
-------------- next part --------------
A non-text attachment was scrubbed...
Name: nginx-80-nginx-active-2day.png
Type: image/png
Size: 3498 bytes
Desc: not available
URL: <http://nginx.org/pipermail/nginx/attachments/20090219/aa8824fc/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: nginx-80-nginx-connection-2day.png
Type: image/png
Size: 5198 bytes
Desc: not available
URL: <http://nginx.org/pipermail/nginx/attachments/20090219/aa8824fc/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: nginx-80-nginx-server-2day.png
Type: image/png
Size: 4357 bytes
Desc: not available
URL: <http://nginx.org/pipermail/nginx/attachments/20090219/aa8824fc/attachment-0002.png>