jarmo@tut.fi (Jarmo Sorvari) (08/22/90)
We have a sendmail configuration problem that resulted in an infinite
loop.
Our system: HP 9000/840 with HP-UX 7.0.
I am sorry to admit that we don't use the default sendmail.cf that HP
promises to support, because something else did not work (can't
remember now what it was). The person who configured our sendmail
copied sendmail.cf from another machine, and made some small changes.
But if I describe briefly the symptom, I hope some sendmail wizard out
there will have some idea where we could start looking for the
problem.
Mail works fine, EXCEPT in the following situation:
when a process *STARTED BY CRON* produces output, it is
automatically sent to the owner of the cron process by mail. This
mail cannot be delivered, and it is appended to
/usr/tmp/dead.letter.
Here is an excerpt of /usr/spool/mqueue/syslog:
Aug 12 02:04:45 AA15501: message-id=<9008120004.AA15501@karhu.tut.fi>
Aug 12 02:04:45 AA15501: from=root, size=547, class=0
Aug 12 02:04:57 AA15501: to=root, delay=00:00:12, stat=unknown mailer error 1
:
:
Aug 12 02:05:04 AA15501: to=/usr/tmp/dead.letter, delay=00:00:19, stat=Can't return mail to anyone
When the local delivery agent was /bin/mail, this was all that
happened. But when I changed the delivery agent to a PD "rmail-biff"
(which sends a user@mailbox-offset to comsat), and then execs
/bin/rmail, then the infinite loop started: a notification of the
failure was sent to the originator, which again failed. Which
generated a notification, etc. etc.
Root has no problems of sending mail from interactive sessions.
I would be very grateful for any ideas! Which types of things could
be wrong in sendmail.cf? (Note: stdout-mail did get through with the
default HP sendmail.cf)
--
-----------------------------------------------------------------------------
! Jarmo Sorvari Control Engineering Laboratory !
! jarmo@tut.fi Tampere University of Technology !
--------------------------------------- BOX 527, 33101 Tampere, Finland -----