httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Enrico Weigelt <weig...@metux.de>
Subject Re: RFC for a Perchild-like-MPM
Date Sun, 23 Jan 2005 13:58:08 GMT
* Nick Maynard <nick.maynard@alumni.doc.ic.ac.uk> wrote:

> Sorry to resurrect this thread, but IMHO (as someone who's an appalling 
> C developer), we really really need perchild or something like it to 
> work.  Having tried MetuxMPM (and got annoyed with its inability to deal 
> with SSL), I believe some serious work needs to be done here.

The problem is: SSL is *NOT* usable for virtual hosting. You need an 
separate socket for each SSL vhost, so you'll probably prefere 
several independent httpd's - maybe then stripped down w/o any vhost support.

> >- What are the other design considerations to doing 'perchild' the 
> >'right' way?
> My only consideration is this.  We are in the process of implementing a 
> per-UID bandwidth accounting setup on our hosts.  I think the process of 
> using the frontend to handle the socket would mean that all traffic 
> would go through this frontend process (and thus the UID owning it). 
> This would preclude our traffic measures from having any meaning.  Does 
> anyone have any thoughts on this?

Aehm, what exactly do you want to do ?

Measure users's consumed traffic ?  (-> logfile ?)
Measure users's consumed bandwidth ? (eh, does this make sense at all? )
Limit user's bandwidth ? (very complicated w/o proper kernel support)


cu
-- 
---------------------------------------------------------------------
 Enrico Weigelt    ==   metux IT service

  phone:     +49 36207 519931         www:       http://www.metux.de/
  fax:       +49 36207 519932         email:     contact@metux.de
  cellphone: +49 174 7066481
---------------------------------------------------------------------
 -- DSL ab 0 Euro. -- statische IP -- UUCP -- Hosting -- Webshops --
---------------------------------------------------------------------

Mime
View raw message