From owner-freebsd-isp Sat Mar 16 08:51:27 1996 Return-Path: owner-isp Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id IAA18306 for isp-outgoing; Sat, 16 Mar 1996 08:51:27 -0800 (PST) Received: from brasil.moneng.mei.com (brasil.moneng.mei.com [151.186.109.160]) by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id IAA18289 for ; Sat, 16 Mar 1996 08:51:23 -0800 (PST) Received: (from jgreco@localhost) by brasil.moneng.mei.com (8.7.Beta.1/8.7.Beta.1) id KAA24399; Sat, 16 Mar 1996 10:47:50 -0600 From: Joe Greco Message-Id: <199603161647.KAA24399@brasil.moneng.mei.com> Subject: Re: csh hanging around after disconnect To: map@iphil.net (Miguel A.L. Paraz) Date: Sat, 16 Mar 1996 10:47:50 -0600 (CST) Cc: freebsd-isp@freebsd.org In-Reply-To: <199603161611.AAA07898@marikit.iphil.net> from "Miguel A.L. Paraz" at Mar 17, 96 00:11:14 am X-Mailer: ELM [version 2.4 PL24] Content-Type: text Sender: owner-isp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > I wrote: > > > > How is this done? Is this a separate daemon, or an option to compile > > > in these daemons? > > Joe Greco wrote: > > > It's something that doesn't exist but would be implemented by adding a > > little code to all the networking daemons. Particularly the daemons without > > timeouts.. > > Looks like a bigger task than it should be. Perhaps this should > be more configurable... or compilable with a #define, ISP - > "define -DISP if you're running the daemons on an ISP machine." Actually, people may frown on this, but I'd sorta like a way to define the use of keepalives as a systemwide default.. > The effect is the machine runs out of ptys. I would want to > install that, but this is a client machine that runs remotely, > and I don't want to take chances unless I'm on-site. :-) > > > Another way - can we looked through the logged in users at regular > > > intervals and find out who's disconnected? You could compare this > > > against the list of logged in users ('show sessions'), but > > > we'd have to differentiate from users who are telnetted in from > > > somewhere else... > > > No, that gets to be very difficult very quickly. > > Yes... is there any other way? Not really. > On another client box, a Linux machine that did the PPP by itself, > I made a script that kills all user processes when the user logs out. > Drastic... but stray user processes were screwing up the > accounting. That only worked coincidentally... you had just the right circumstances. A more general problem is, for example, a busy news server... you don't have folks logging in and out. ... Joe ------------------------------------------------------------------------------- Joe Greco - Systems Administrator jgreco@ns.sol.net Solaria Public Access UNIX - Milwaukee, WI 414/546-7968