Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!uwm.edu!rpi!gem.mps.ohio-state.edu!apple!voder!pyramid!csg
From: csg@pyramid.pyramid.com (Carl S. Gutekunst)
Newsgroups: comp.sys.encore
Subject: Re: Problems with biff/comsat notifying wrong people
Message-ID: <86512@pyramid.pyramid.com>
Date: 4 Oct 89 22:24:56 GMT
References: <2069@bu-tyng.bu.edu>
Reply-To: csg@pyramid.pyramid.com (Carl S. Gutekunst)
Organization: Pyramid Technology Corp., Mountain View, CA
Lines: 13

In article <2069@bu-tyng.bu.edu> oneill@bu-tyng.bu.edu (Brian O'Neill) writes:
>Lately I've heard complaints that on occasion people receive notification of
>new mail, when if fact the mail is for another person.... It seems that after
>the Multimax has been running for a long time (> 3-4 weeks), comsat isn't
>getting the right information from /etc/utmp.

The problem is it isn't rereading /etc/utmp. This is from a bug in the 4.2BSD
version of sleep(3). (It goes to sleep an never wakes up.) The bug is fixed in
4.3BSD; a quick diff of the two makes it painfully obvious. There is also a
bug in the 4.2BSD interval timers that will only show up on a multiprocessor
system and that exhibits the same behavior.

<csg>