Micha <[EMAIL PROTECTED]> writes:

> I just started fetching and 'abo order' of
> http://www.wto.org/english/thewto_e recursivly, 5 levels deep, same
> host. When it started, i wanted to look at the status but accidently
> hit on 'fetch'. There was an order list in pipe, and i don't do
> automatuc fetching. So this was launched in parallel. Then i noticed
> the ppp transfer shows often paunes for 2 -3 seconds.
> The graphical log shows a very uneven and low utilization of my
> (admittedly small) bandwidth, in both dimensions (x y). 
> 
> So i wonder,
> 
> 1. Is there any priority when, while wwwioffle already is fetching,
> a user requests a new page ?

There are two options in the configuration file that can set the
maximum number of fetch processes and the maximum number of total
processes.  The only priority is that fetching cannot block real-time
requesting of new pages.  There is no bandwidth priority because this
is not necessarily possible to do.

> 2. What priority applies when i start fetching a list of orders
> while the abonnement order is already processed ?

I am not sure that I understand the question.  You are either fetching
or not fetching, you cannot fetch twice or influence the rate of
fetching when it has started.

> 3. Could my 'fetch' action have had any effect ?

I don't see how.

> 4.  Is wwwoffled doing any kind of scaling bandwith ? Like eg
> apt-proxy-v2, which runs 'twisted' factories which scale by parallel
> threading (or so). 

No, it isn't that clever.

> I'm still on a 56kbit modem (but only for some more weeks now :)
> While fetching, a typical WWWOFFLE Server Status:
> Version      : 2.8e
> State        : online
> Fetch        : active (wwwoffle -fetch)
> Purge        : inactive
> Last-Online  : Sun, 06 Mar 2005 16:51:18 CET
> Last-Offline : unknown
> Total-Servers: 5
> Fetch-Servers: 3
> Server-PIDs  :  8553 8548 8555 3092 8558 (configured max 6)
> Fetch-PIDs   :  8553 8548 8555
> 
> The same time, outgoing shows 1271 orders (growing).
> 
> /var/cache/wwwoffle is a seperate ext3 partition, with hash-trees
> (tune2fs -O dir_index) enabled, size after last purge (yesterday)
> 970 MB, on a fast HD.

I don't think that the handling of the outgoing directory is what is
causing the problem.  More likely it is because the host that you are
accessing is slow and making 3 parallel requests to it will not make
it go any faster.

-- 
Andrew.
----------------------------------------------------------------------
Andrew M. Bishop                             [EMAIL PROTECTED]
                                      http://www.gedanken.demon.co.uk/

WWWOFFLE users page:
        http://www.gedanken.demon.co.uk/wwwoffle/version-2.8/user.html

Reply via email to