I have only a few hundred lists and no performance problem with average 
cpu less than 1%.  My shutdown statistics show many more jobs in the 
"Other" category than all others combined; is this where x-spam jobs are 
counted?

Does Notre Dame get more x-spam jobs than I do?  If not, then maybe poor 
web performance w/ x-spam  is just a symptom and not the cause of the ND 
problem?

Here are my stats from yesterday, fwiw:

* The average  CPU time  used by  LISTSERV for this  session is  0.1%. Since
* LISTSERV was last rebooted (on 13 Nov 2005 at 23:58, ie 1 day and 0 minute
* ago), the following requests have been processed:
*
* - Interactive messages from users    1000
* - Postings to distribution lists      289 (55268 recipients)
* - DISTRIBUTE jobs (internal)          559
* - DISTRIBUTE jobs (from network)     3544
* - Other incoming files               8331
*
* - Database searches (interactive)      26
* - Database searches (batch mode)     None

cheers, wayne

Paul Russell wrote, in part,  on 11/15/2005 10:11 AM:
> On 11/14/2005 22:49, Andrew Bosch wrote:
>> We haven't noticed, but then our mail volume is not as great as Notre
>> Dame's. Perhaps it is possible to split the WA from the Listserv
>> installation and have it run on another host?
>>
>
> I believe this suggestion has been offered in the past, and I believe 
> that
> the response has always been that it is not possible to do this. The 
> problem
> is not that the server is overloaded; the problem appears to be two-fold:
> (1) everything must go through lsv and lsv is single-threaded, and (2) 
> inbound
> messages in the listserv/spool directory are given precedence over 
> pending web
> requests.
>
> The rationale for X-SPAM jobs is no longer valid. Spammers used to send
> hundreds, even thousands of messages with the same sender address, so
> blocking or quarantining all messages with a sender address seen on spam
> was effective. Spammers' techniques have changed, but LISTSERV is still
> using the same old model that used to work "back in the old days". I 
> think
> it is time to review that model to determine whether it is still valid in
> the current spam environment. It appears to me that the cost of this
> feature significantly outweighs the benefit. Does anyone share this view?
>
> -- 
> Paul Russell
> Senior Systems Administrator
> OIT Messaging Services Team
> University of Notre Dame