From owner-freebsd-questions Mon Jun 17 14:59:01 1996 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id OAA03482 for questions-outgoing; Mon, 17 Jun 1996 14:59:01 -0700 (PDT) Received: from Rigel.orionsys.com (root@rigel.orionsys.com [205.148.224.9]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id OAA03476 for ; Mon, 17 Jun 1996 14:58:57 -0700 (PDT) Received: (from dbabler@localhost) by Rigel.orionsys.com (8.7.5/8.6.9) id OAA02593; Mon, 17 Jun 1996 14:57:35 -0700 (PDT) Date: Mon, 17 Jun 1996 14:57:35 -0700 (PDT) From: Dave Babler To: Terry Lambert cc: questions@FreeBSD.ORG Subject: Re: Rlogin delay In-Reply-To: <199606172129.OAA08654@phaeton.artisoft.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-questions@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Mon, 17 Jun 1996, Terry Lambert wrote: > > I've observed something concerning rlogin that is a bit odd to me, and > > increasingly annoying for my users. When a user logs off and then tries > > to re-log within a reasonable delay from another machine on the local > > network, the login will be ingored for a very long time (10-60 seconds > > seems typical). I checked the /etc/inetd.conf to make sure it was okay: > > Did you turn on the global SO_KEEPALIVE that everyone was talking about? > People who wanted this were MUD people and similar sites, without regard > for the impact on things that didn't expect it to be on. > I basically installed/built whetever was the standard configuration (2.1-stable)... I also tried the -n option for rlogind in /etc/inetd.conf to no effect. The first couple of relog attempts usually work in a second or two - subsequently, the delays required get pretty long. > > This would be a natural side effect of the thing having data to drain > that your client didn't wait for because of the disconnect on the OOB. > > This is really a problem with setting SO_KEEPALIVE, not with rlogin. > > If you want it to operate correctly, you would need to delay the OOB > disconnect for the flush interval for the non-OOB data (or add a > "disconnect ack wait" to the server and a "disconnect ack" to the > client. > The client here is not modifiable at all, unfortunately, and may implement things less than perfectly (I've already found a problem with what it propogates as window size to FreeBSD's rlogind). -Dave Babler