From owner-freebsd-hackers Wed Nov 12 17:50:34 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id RAA01730 for hackers-outgoing; Wed, 12 Nov 1997 17:50:34 -0800 (PST) (envelope-from owner-freebsd-hackers) Received: from ns1.yes.no (ns1.yes.no [195.119.24.10]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id RAA01718 for ; Wed, 12 Nov 1997 17:50:30 -0800 (PST) (envelope-from eivind@bitbox.follo.net) Received: from bitbox.follo.net (bitbox.follo.net [194.198.43.36]) by ns1.yes.no (8.8.7/8.8.7) with ESMTP id BAA19129; Thu, 13 Nov 1997 01:50:06 GMT Received: (from eivind@localhost) by bitbox.follo.net (8.8.6/8.8.6) id CAA29329; Thu, 13 Nov 1997 02:50:00 +0100 (MET) Date: Thu, 13 Nov 1997 02:50:00 +0100 (MET) Message-Id: <199711130150.CAA29329@bitbox.follo.net> From: Eivind Eklund To: Brian Somers CC: doconnor@ist.flinders.edu.au, freebsd-hackers@FreeBSD.ORG In-reply-to: Brian Somers's message of Wed, 12 Nov 1997 12:31:20 +0000 Subject: Re: Dial on demand with dynamic IP References: <199711121115.VAA04516@holly.rd.net> <199711121231.MAA29463@awfulhak.demon.co.uk> Sender: owner-freebsd-hackers@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk [Dial on demand with dynamic IP] > > Is it worth a try? Is anyone alreay doing it? :) > > This is only half the problem, but it would probably make things > better for most sites. > > The ``other'' problem is when a process bind()s to an IP number that > gets dynamically re-assigned. > > Perhaps the *real* answer is to never change the IP number of the > interface, and to change all IP numbers that are the same as the one > netogitated through IPCP to that of the interface. > > Hmmm, that would work wouldn't it ? Yeah, it'd likely work, but would deny one IP address from being accessed from your network. Not too big a loss, though. I'm doing it by using an interface route to tun and adding the dynamic addresses to the lo0 interface; I could probably just as well add them to tun, as long as the addresses are removed when the call is brought down - demand TCP streams must not bind to the addresses, and for an interface with addresses at demand time they will. Eivind.