[6019] in Athena Bugs
Leftover processes
daemon@ATHENA.MIT.EDU (Richard Basch)
Thu Sep 13 10:11:06 1990
Date: Thu, 13 Sep 90 10:10:32 -0400
To: bugs@ATHENA.MIT.EDU
In-Reply-To: bugs[6018]
From: Richard Basch <probe@MIT.EDU>
I have seen this on almost every machine I have logged into out in the
public clusters... We now have had reports from all architectures (if
not, I will at least claim VS3100s and RTs have both had this problem,
since that is what I have been using).
During my login sessions, the leftover processes are almost invariably
emacs processes, possibly with a "comp" parent that is still around, and
possibly still a parent shell. I have yet to see any that do not
include an emacs process somewhere.
Of course, all this means is that emacs may not be listening to the X
error that the connection is broken. Perhaps it is filtering out
SIGPIPE?
I also agree it is a big problem, since it is eating a lot of CPU
time... I have rarely seen a case where it has not already eaten a few
thousand minutes of CPU time, and the number increases visibly in
real-time.
-Richard
[6018] daemon@ATHENA.MIT.EDU Athena Bugs 09/13/90 10:02 (33 lines)
Subject: nasty leftover processes
From: rsargent@ATHENA.MIT.EDU
Date: Thu, 13 Sep 90 10:02:10 -0400
To: bugs@ATHENA.MIT.EDU
I've noticed, for the new release, that ALMOST EVERY TIME I LOG IN, there
are nasty leftover processes eating up CPU time. This happened
occasionally during the last release, but seems to happen much more often
in this one. An example:
/mit/rsargent% ps -aux | more
USER PID %CPU %MEM SZ RSS TT STAT TIME COMMAND
24128 79.6 3.4 970 396 v0 R N 19184:40 /usr/athena/emacs
rsargent 18336 9.8 2.2 161 137 p2 R 0:00 ps -aux
...
Often there are two or three. Usually they are an emacs process, although
I've seen others as well. They really slow down the machine, and I feel
sorry for the people who don't know how to find these processes and kill
them.
I only use VAXstation 3100s, so I don't know if the problem occurs on other
machines. The problem occurs on both normal and test machines running
their respective versions of release 7.1.
Thanks, and good luck.
Randy Sargent
