a question about limit_zone module

[Igor Sysoev]

On Wed, May 23, 2007 at 05:15:35PM +0200, Manlio Perillo wrote:

> Igor Sysoev ha scritto:
> >On Wed, May 23, 2007 at 03:36:14PM +0200, Manlio Perillo wrote:
> >
> >>Igor Sysoev ha scritto:
> >>>On Wed, May 23, 2007 at 02:09:26PM +0200, Manlio Perillo wrote:
> >>>
> >>>>In the wiki, the documentation about limit_conn says:
> >>>>
> >>>>Directive assigns the maximum number of simultaneous connections for 
> >>>>one session
> >>>>
> >>>>
> >>>>What means "simultaneus"?
> >>>If session == $remote_addr, then it means that client could not open more
> >>>than specified concurrent connections.
> >>But usually concurrent means n sessions in t seconds.
> >>What is the value of t, if any?
> >
> >Here concurrent means at the some moment.
> >
> 
> Ok, thanks.
> 
> I would like to have a more general module, where I can limit, as an 
> example, more that 3 connections in 1 seconds from the same address, or 
> with the same value of an HTTP (including upstream) header.

No, nginx currently does not support this.
I plan to extend limit_zone capabilites to limit say
no more than 5 request per second with 10 requests burst (using
leaky bucket algorithm).


-- 
Igor Sysoev
http://sysoev.ru/en/