|
Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com |
From: Acee Lindem (acee_at_REDBACK.COM)
Date: Mon Sep 09 2002 - 15:21:24 CDT
Christian,
It is closer to #2 than #1. Assuming a single area topology,
it would be the neighbor inactivity event that triggers
re-origination of router LSAs for A and B. Refer
to section 12.4 in RFC 2328.
Christian Glomb wrote:
> Dear all,
>
> let's assume we are monitoring the flow of OSPF messages within one OSPF
> area. All routers have successfully come up and are running stable, i.e.
> only HELLO messages are exchanged between adjacent routers. Now the link
> between Router A and Router B fails such that Router A does not receive
> HELLOs from Router B and vice versa. After the timer RouterDeadInterval
> has expired at each Router, Router A as well as Router B assumes that the
> other router (or the link to the other router) has crashed. All OSPF
> information concerning the other router is removed from the database of
> Router A and Router B respectively and both Routers are recalculating their
> routing tables.
>
> However Router A as well as Router B have working adjacencies to other
> Routers C...E and F...H respectively and are still sending HELLOs to these
> neighbors.
>
> What happens then?
>
> 1.
> Due to the fact that the ID of Router B is missing in the neighbor list
> of the HELLOs send from Router A to Routers C...E and the ID of Router A
> is missing in the HELLO neighbor list sent from Router B to Router F...H,
> Routers C...E as well as Routers F...H are sending a LINK_STATE_REQUEST to
> Router A and Router B respectively, since they notice that the topology has
> changed (i.e. one link is missing).
> Routers A and B resp. are responding with a LINK_STATE_UPDATE. Then the
> Flooding Procedure starts due to the receipt of such a message at Routers
> C...E and F...H.
>
> OR
>
> 2.
> Routers A and B resp. send a LINK_STATE_UPDATE directly to Routers C...E
> and Router F...H after having updated their routing tables to trigger the
> Flooding Process.
>
> OR
>
> 3. I'm completely wrong, it's working completely different.
>
> Thank you for your comments,
>
> Christian.
>
>
-- Acee
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]