[891] in Release_7.7_team

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

why sendmail doesn't send mail (?)

daemon@ATHENA.MIT.EDU (Dan Winship)
Mon Mar 10 14:13:09 1997

Date: Mon, 10 Mar 1997 14:12:56 -0500 (EST)
From: danw@MIT.EDU (Dan Winship)
To: ops@MIT.EDU, release-team@MIT.EDU

I got some mail this morning that had been sent Friday morning from an
8.0K sun. This prompted me to go investigate the machine.

It turns out it was just the same old sendmail lossage, but then
someone else crashed the machine before the cron job had a chance to
free it, but then it got sent on reboot.

BUT... while I was there, I noticed something interesting:

hostname# last foobar
foobar   ttyp0        :0.0             Fri Mar  7 11:00 - 11:19  (00:19)

hostname# more syslog
Mar  7 11:05:54 hostname sendmail[28645]: AA28645: message-id=<9703071605.AA28645@hostname.MIT.EDU>
Mar  7 11:05:58 hostname sendmail[28645]: AA28645: from=foobar, size=334, class=0, received from local
Mar  7 11:05:59 hostname sendmail[28647]: AA28645: to=someone, delay=00:00:05, stat=Sent
Mar  7 11:19:38 hostname sendmail[28656]: AA28656: message-id=<9703071619.AA28656@hostname.MIT.EDU>
Mar  7 11:19:39 hostname sendmail[28656]: AA28656: from=foobar, size=1113, class=0, received from local
Mar  7 14:22:43 hostname sendmail[372]: AA00372: message-id=<9703071922.AA00372@hostname.MIT.EDU>
Mar  7 14:22:43 hostname sendmail[372]: AA00372: from=bazquux, size=114, class=0, received from local
Mar  7 14:22:44 hostname sendmail[374]: AA00372: to=someoneelse, delay=00:00:01, stat=Sent

`foobar' is the person who sent me mail. The message he sent me was the
one at 11:19, which lists a `from' but no `to' (presumably sendmail
crashed before writing that line).

The neat bit is that he also logged out at 11:19... My guess is that
sendmail is losing when the user's passwd entry disappears from
/etc/passwd while it's running, or something else related to
athena-ish logout that wouldn't happen on non-athenized machines or
private workstations, neither of which seem to have been bitten by
this bug.

-- Dan

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