Subject: Re: Strange routing problem: NetBSD goes outside for a local IP address
To: Stephane Bortzmeyer <bortzmeyer@nic.fr>
From: Christian Hopps <chopps@allegronetworks.com>
List: netbsd-users
Date: 02/12/2003 07:48:08
This is how BSD (not just NetBSD) works.

The routing daemon should also have a loopback route internally for the 
local
IP of the p2p interface. The user should have a choice as to the 
preference
for this loopback route, and thus be able to choose which route gets 
installed
into the kernel. This is how GateD works.  If you run zebra but not 
OSPF, does
it not install a loopback route for the local IP of the p2p interface?

Chris.

On Wednesday, February 12, 2003, at 02:23 AM, Stephane Bortzmeyer wrote:

> On Tue, Feb 11, 2003 at 11:07:35PM +0100,
>  Manuel Bouyer <bouyer@antioche.eu.org> wrote
>  a message of 77 lines which said:
>
>> Actually I can't see how this is wrong. If you have an entry in the 
>> routing
>> table, it's expected to be used :)
>
> No, it is expected to be used *after* the local interfaces are
> consulted. Remember that Linux does not do it and OSPF gurus think
> that it should not be done.
>
> Anyone runs OSPF PtP links on NetBSD? How do you address that issue?
>
>
>
>