[prev in list] [next in list] [prev in thread] [next in thread] 

List:       zebra
Subject:    [zebra 5927] FW: FW: FW: Zebra/ospfd over unnumbered interfaces?
From:       "Bulger, Tim" <TBulger () ea ! com>
Date:       2000-11-29 17:50:01
[Download RAW message or body]

After playing with this for a few days, I found some interesting information
that might be helpful towards having a fix for this.  If there are three
interfaces on the system sharing the same IP address (one ethernet, two
PtP), ospfd will send a HELLO packet for each interface to the ethernet,
meaning three HELLO packets.  The reason for this, I believe, is a section
of ospf_packet.c that says:   /* Select outgoing interface by destination
address. */  ... It seems as though it exits this loop as soon as it finds a
match for the IP, without looking for more matches (I'm not a C programmer
:( ).  I'm just speculating here, but I suspect that the next step beyond
getting the adjacency stuff worked out is finding a way to tell ospfd/zebra
that a route learned over an unnumbered interface uses the interface as the
next hop instead of an IP address.

I appreciate you guys looking into this... Zebra is a huge part of my
production network and I'm trying to make it bigger. :)

Thanks all,
Tim


-----Original Message-----
From: Kunihiro Ishiguro [mailto:kunihiro@zebra.org]
Sent: Tuesday, November 28, 2000 6:02 PM
To: TBulger@ea.com
Cc: zebra@zebra.org
Subject: Re: [zebra 5908] FW: FW: Zebra/ospfd over unnumbered
interfaces?


>Does anyone have any ideas regarding this at all?  I am looking at
deploying
>zebra in a network that already exists with "unnumbered" tunnels and
slicing
>up IP space for each end of the tunnel is not appealing. :)

In zebra, there is alredy unnumberd nexthop support.  Toshiaki, would
you mind to check support of unnumbered address configuration in ospfd?
-- 
Kunihiro Ishiguro

[prev in list] [next in list] [prev in thread] [next in thread]