Re: removing <exiting> processes -- SUMMARY

From: Jay Bourland (jayb@gauss.Stanford.EDU)
Date: Thu May 21 1992 - 13:25:16 CDT


> Occasionally one of my users will exit ungracefully and will leave a
> process hanging around along the lines of:
> 12242 ? I 0:09 <exiting>
> This invariably comes from one of the terminals we have hooked up to
> an mti board. I know this because a "w" command will show the user as
> logged in and idle. I have tried everything I know to remove these
> processes, but the only thing that ever works is rebooting the
> machine. I would like to know how to get rid of these processes,
> because they also block the terminal line - i.e. no one can login on
> that terminal. Also, is there any way to find out what the original
> process was, so that I can make sure that this stops happening. For
> what it's worth, the offending machine is a Sun 3/160 (temporarily
> diskless) running SunOS 4.1.1 (no patches). My guess is that, because
> of the slow speed of the machine, some kind of communication error is
> occurring between the login shell and some large process.
>

This question generated a lot of responses, so I thought I'd give a
short summary of them.

(1) Ignore the process, it doesn't do anything.
        This doesn't work because the process shows the terminal
        in use and thus blocks the getty so no new users can log in.
        In fact the terminal is totally hung. Turning it off, kill -1 1,
        etc. cannot restart it.
(2) Find the parent process and kill it.
        The parent process is given as 1. Thus, killing it reboots the
        machine, which is what I'm trying to avoid.
(3) Attach a trace to the process, ala "trace -p 12242".
        When I tried this, all I got was "No such process".
(4) from Jim Lick <jim@piggy.ucsb.edu>: Apply patch 100358-01.
        Technically, this patch is for dial-ups not serial line terms,
        but it looks like it addressed the same problem, so I applied
         it and rebuilt the kernel on the offending machine.
        Unfortunately, (or maybe fortunately), these events are
        relatively rare, so it may be a while before I find whether or
        not this fixed the problem.

Thanks again to all who responded,

--
Jay Bourland
jayb@math.stanford.edu



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:42 CDT