Errors-to: owner-drums@CS.UTK.EDU
To: drums@CS.UTK.EDU
Subject: Re: RFC 821 problem list 
In-Reply-To: Your message of "Wed, 31 May 1995 15:50:52 -0400."
             <cjnAWQq00WBwM9xk93@andrew.cmu.edu> 
Date: Thu, 01 Jun 1995 06:56:02 +1000
Message-Id: <840.801953762@munnari.OZ.AU>
From: Robert Elz <kre@munnari.oz.au>

    Date:        Wed, 31 May 1995 15:50:52 -0400 (EDT)
    From:        John Gardiner Myers <jgm+@CMU.EDU>
    Message-ID:  <cjnAWQq00WBwM9xk93@andrew.cmu.edu>

    I suppose you'd like to turn all 4XX codes into 5XX codes then?

No, my basic rule of thumb would be that if the error is of
the kind that will probably fix itself, without human intervention,
then 4xx is the right one to use (queue filesystem full fits here,
as the queue should eventually drain, leaving space for more mail,
mailbox full however needs the user to read mail, or the admin to
move the mail elsewhere).  If some human action is needed, then
5xx is better.

Clearly there's no absolute way to mandate rules that apply
in all cases, however it seems to me that at the minute there
are enough 4xx and 5xx codes already defined to cover all of
the really necessary cases, and that this is something that
doesn't really need to be fiddled with.  All that is required
is that mailers actually pick the right code to use.

    Our users frequently go over quota for
    only minutes or hours at a time.

Either your quotas are way too low, or your users are
cheating you.   If I couldn't tell the users here that if
they let their mailbox get anywhere near the limits (they
normally never come within megabytes) they will lose mail,
(that is, it will bounce), then they'd simply make use of
mail spool space for all kinds of files to defeat other
limits.

Those users who seem to like to use those brain deat UA's that
like to leave all the user's mail in the system mailbox soon
learn that this is not a real good idea.

kre