[j-nsp] Problems with OSPF stub area (Juniper and Cisco)
Wed, 6 Nov 2002 10:34:34 -0500
I believe that makes it pretend that the interface is a broadcast,
causing a different representation in the LSDP, DR election, and type2
LSA generation, etc. We can let you set the interface to NBMA or P2MP,
but not broadcast. I'm a little surprised the adjacency became full,
with the two routers disagreeing on the link type. I'd love to see the
router LSA's for those two routers, while that command is active.
I suspect the fact that it causes the cisco to coalesce an unused range
is an unintended feature. What you are suggesting is essentially
abusing an unrelated aspect of the protocol to trigger a side effect of
coalescing routes that you don't have.
I suggest that you are better off having the routes. Can you add a /32
from the range to a loopback somewhere? More importantly, why do you
need this route in OSPF if there are no destinations from it in the
area? Perhaps there is a more natural way to affect your desired
Juniper Pro Services
On Wednesday, November 6, 2002, at 01:31 AM, Matti Saarinen wrote:
> Avram Dorfman <email@example.com> writes:
>> I tried this config between two junipers, and at first the /25 did not
>> show up either, but it shouldn't have - there were no more specific
>> active routes from the range to coalesce.
>> Then I added a 22.214.171.124/30 hanging off of the "cisco" in your
>> scenario, and both it and the /25 discard route showed up on both
>> Do you have such a subordinate route actually in your area?
> No, I don't have such a route. But, now I looked at Cisco's
> configuration again and noticed the the following command
> under the atm interface's configuration.
> ip ospf network broadcast
> When I remove the command the /25 route disappears from the
> routing table. So, I think there's the problem. I couldn't find
> the similar configuration statement under Juniper's OSPF
> configuration. Is there any way to configure the similar
> functionality in JUNOS? I mean that the way OSPF sees the ATM
> interface should be broadcast instead of point-to-point.
>> (As a side note, I don't see how you got your ATM interface to come up
>> without having specified the atm-options vpi 0 max-vcs statement, or
>> did you just omit it from your email?)
> Yes, I did omit it from my email.
> Matti Saarinen, Tampere University of Technology / Network
> Phone: +358 3 3115 3337 # Address: P.O. Box 692
> Fax: +358 3 3115 2172 # FIN-33101 Tampere
> Email: firstname.lastname@example.org # Finland