On Thu, 2 Mar 1995, Eric Thomas wrote:
> On Thu, 2 Mar 1995 09:38:07 -0500 Laura Toms <[log in to unmask]> said:
>....
> >there a way to configure LISTSERV to recognize auto-replies and not send
> >them out to the entire list of subscribers,  or do we have to get on the
> >subscribers themselves  about it? It  seems like an incredible  waste of
> >resources.
>
> Unfortunately, auto-replies are very good at avoiding anything that could
> help software  identify them as  auto-replies. Personally, I  just delete
> the  users until  they stop  their auto-reply  program. Until  auto-reply
> programs clearly identify themselves (in some form of computerese), there
> isn't much else you can do.
 
  If auto-reply programs are like those vacation programs (I think they are)
the only thing they do not reply to are "XXXX-Request, UUCP, Postmaster,
MAILER and MAILER-DAEMON' in the Sender field.  If they find a header
of 'Precedence: bulk' or 'Precedence: junk' I think they will not reply
either, but that would mean...well, we know what that would mean, and
I think Eric has more priorities other than including another header.
What makes these annoying for lists is that if they find a 'Reply-to'
field, they send there, and often that is the list.  I agree that deleting
the user is the only alternative because the program is directly under
the individual's control, and if set up incorrectly with an interval of
zero instead of infinite, will send a reply out *every* time mail is
received.  Such a waste.
 
  I have a question on another topic that we are trying to solve.  I
wonder if anyone could give me a quick tutorial on how listserv handles
memeory.  That is to say, does listserv require a specific amount of
'real memory' or can listserv (1.8a running on Unix IRIX 5.2) make use
of swapping on disk. The reason I ask is because Listserv seemed to
freeze about 1:30 pm today and spooled incoming mail for a few hours.
At first we thought it may have crashed, but it did not.  We had another
program running on the same machine, and when we checked the stats,
there was between 2.0 - 2.5 M of real memory available with about 13M
of disk swapping space available.  We increased the disk space to 64M,
but listserv didn't seem to come alive until real memory jumped up to
about 3M or more.  These statistics I obtained from osview because I
thought it might be more reliable than from top.  The unix box is
an sgi INDY.
 
  Another related question.  It seems to me that when listserv processes
a request locally, once listserv hands over the job to sendmail for local
distribution, listserv has nothing more to do with those specific job
requests, is that correct?  One system administrator here told me that
listserv was slow or froze because our main node was down so mail was
queued and re-processed every 15 minutes.  We changed sendmail to 'keep
trying' every hour instead, but I can't see how the two are related and
I argue that changing sendmail only conserves disk space eaten up by
the syslog and has no impact on listserv at all.  Granted, I'm guessing
because I can't find a one-to-one correspondence between what sendmail
passes to Listserv's mail interface and what listserv then passes to
sendmail for distribution...and around we go.  Thus, there is no evidence
for or against it in the syslog.  So, can someone tell me I'm correct :-)
or even incorrect would help.  I'm betting Listserv is more efficient and
that the real problem is that a program was run on the machine that should
have been run on a different test machine first so as not to chew up
real memory for production software.  Many thanks for your time.
 
  Sincerely, Trish
--
Trish Forrest
Computing Services, University of Windsor, Windsor, Ontario, Canada