Subject: Re: Slow turnaround of unit
From: raj@rijhwani.org (Raj Rijhwani)
Date: 10/01/2004, 18:17
Newsgroups: alt.sci.seti

On Friday, in article <OBiiysRsdp$$EA8O@unseen.demon.co.uk>
     tpratchett@unseen.demon.co.uk "Terry Pratchett" wrote:

Since the beginning of January the (theoretically) fasted machine in my 
network,  a 2.2 Ghz P4, has been turning around units in 15-20 hours. 
It's running the graphical version under XP

This is as opposed  to a  4-5hr turnaround on another machine, the1.8 
Ghz  P4  running the CLI).  It's slower even than the old 450Mhz P2!  I 
didn't worry at first because all the machines get the occasional big 
one, but *every* unit take the better part of a day now. 4-6 hours per 
unit was the norm until the end of the December.

No new software has been put on the machine, no specs have been altered, 
and in every other respect it appears to be working fast and 
efficiently.  Can anyone give me a clue about what might be happening?

1) Any exposure to the outside world?  (Possible infection or attack.)

2) You may not have changed anything, but the "beginning of January" 
is a red flag.  Are you running something that schedules an annual task 
on or around New Year?  (Process stuck/deadlocked/in a permanent run state.)

3) Don't know XP, but if there is any means of examining resource usage, 
it would be wise to see what's
a) eating the most memory
b) eating the most processor time
Is the (hard) disk access very high?  If something has overflowed available 
memory and is causing a lot of paging, this would slow you down immensely 
because access times are now dependent not on RAM access times but disk 
access times.  Usual culprits are growing databases, indexes, mail spools 
or print spools.

Just a few suggestions.
-- Raj Rijhwani | This is the voice of the Mysterons... raj@rijhwani.org | ... We know that you can hear us Earthmen http://www.rijhwani.org/raj/ | "Lieutenant Green: Launch all Angels!"