From: Gene (ccie_iraq@myway.com)
Date: Mon Oct 02 2006 - 12:56:34 ART
Cagri-Try looking at the OSPF database in R3 before and after you set up the virtual link. I think of it like this--if R1 is going to have an interface in area 0 (and the virtual link endpoint is an interface in area 0) then it has to know about everything area 0 knows. In particular, R1 needs to know about those unsummarized /24s.So how can R1 find out about those? One way would be to encapsulate the routing updates from R5 to R1, so that R3 doesn't see them. This is the GRE tunnel solution.Another way is to populate R3's database with area 0 information, then pass that info on to R1. This is the way the developers of OSPF chose to implement virtual link. So in the virtual link solution R3 sees the /24s.Check the database in R3 with and without the virtual-link and with and without the tunnel.-gt--- On Mon 10/02, Cagri Yucel < cyucel@gmail.com > wrote:From: Cagri Yucel [mailto: cyucel@gmail.com]To: ccielab@groupstudy.comDate: Mon, 2 Oct 2006 12:27:44 +0100Subject:
OSPF questions (IE VOL1 LAB7)I am really confused with this one, below is a quick diagramArea 0--------- R5 ---Area1-----R3 ------Area1------R1------Area2Two /24 ranges in area0 are summarised using R5 so they received as a /23 onR3, no problems.Later we add Area2, then naturally a Virtual Link between R5 and R1 fine,suddenly R5 starts to leak /24s to R3.Why is this ? I am not talking about /24s coming from R1 which naturallyneed to be summarised further.IE's solution for the problem is using a tunnel and it works perfectly. Butwhat is the underlying logic ?Why VLINK leaks routes at the first place and why tunnel avoids thatbehaviour.-cagri_______________________________________________________________________Subscription information may be found at: http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Wed Nov 01 2006 - 07:29:04 ART