I thought of replacing the default shell with a shell that
keeps its pid file in a central place. If such a file already
exist, it'll kill the other running shell before logging in.
Something like
export
LOCKFILE=/tmp/singleshell.pid
while [ "$(cat
$LOCKFILE)" -eq "$LOCKFILE" ]; do
kill $(cat
"$LOCKFILE")
sleep 1
kill -9 $(cat
"$LOCKFILE")
echo $PID
>$LOCKFILE
done
exec /bin/sh
The benefits of
that are, making sure it is always one man fixing the server
(what can happen is two people get a call "help, server is
not working", and forgetting to check who else is logged in
in the rush). And, the option of streamline logging. I'll
use sudosh2 or some similar recording shell/recording
terminal, and I'll have the exact log of what happened, no
need to correlate events in two shells.
I'd like to hear
your opinion and/or experience with such a scheme.
Thanks
