[15718] in Athena Bugs

home help back first fref pref prev next nref lref last post

Sendmail/moira interaction bug

daemon@ATHENA.MIT.EDU (Patrick J. McCormick)
Fri Dec 5 00:40:57 1997

Date: Fri, 5 Dec 1997 00:40:48 -0500 (EST)
From: "Patrick J. McCormick" <pmccormi@MIT.EDU>
To: bug-moira@MIT.EDU
Cc: postmaster@MIT.EDU, bugs@MIT.EDU, 50k-online@MIT.EDU


I think this is a known bug, but having been recently bitten again, I
would really appreciate someone fixing it.  I'm cc'ing this to postmaster
since this may be wholly in the sendmail rules and not really a moira
problem.

If a poorly-formed email address is added to a moira list, and this list
is propogated to the mailhubs, then mail sent to the list is bounced and
NOBODY on the list receives the mail.

An expn of such lists reveals something like:

550 "sib@worldnet.att.net"... User unknown: Bad file number

In my case, this entry in moira looked like:

STRING:"sib@worldnet.att.net"

where the quotation marks were part of the value, which is an illegally
formed email address and renders the entire list and any lists which
include it unusable.  Other kinds of malformed string values will cause
the same behavior.

When I discussed this earlier on -i help, someone mentioned that the
client should throw up a warning.  Maybe it did; I didn't enter this
address, so I'm not sure.  Even so, sendmail or moira should properly deal
with bad values and bounce the message while sending mail to every valid
address on the list.

This is a serious bug since people assume that when they receive a bounce
from a list with this problem that all other recipients recieved the
message successfully.  Serious confusion has ensued when in fact nobody
receives the message. 

The fact that MIT still has a list propagation system that doesn't allow
for instant updates is bad enough, but this limitation becomes
unacceptably frustrating when a mailing list can be taken out of
commission for a whole day due to a single bad entry.

Please fix these problems.

--Pat McCormick



home help back first fref pref prev next nref lref last post