From: Vytautas Valancius (valas@mail.lt)
Date: Mon Nov 29 2004 - 23:55:32 GMT-3
Hi Group!
I have a general problem with route redistribution feedback.
There are two redistribution points between ISIS and OSPF.
I play with distance to make ISIS process prefer its native routes -
redistribution is muttual. Now lets say OSPF connects several other routing
domains (as ussual) - RIP, EIGRP etc. EIGRP anounces some route X. Route
X gets installed in all ISIS and OSPF domain routers (because of mutual
redistribution).
The problem is that when I withdraw route X from EIGRP domain (lets say I
was asked to configure summary, or ppp interface went down and /32 route
dissapeared) it starts 'circulating' in ISIS/OSPF!!!
Here is how it happens:
1.First ISIS-OSPF redistribution point gets OSPF LSA without route X.
2.OSPF withdraws route X from routing table
3.OSPF redistributes route X from ISIS (as route X is still there!)
4.OSPF sends new external LSA
5.Second ISIS-OSPF redistribution point gets two LSAs - one is saying that X
no longer is reachable from EIGRP side, the other is from first
redistribution point saying that route X is reachable! This is the loop for
nonexistant route!
This is very general problem. I mean I see this behaviour in almoust any set
of protocols and in any lab scenario. Routers I posses have quite different
processing power so maybe it adds to such bad synchronization of things. The
cure for that is to clear routing processes.
Is there any way to overcome this situation? Should we try to overcome it?
:-)
Comments are welcomed,
Vytautas Valancius
This archive was generated by hypermail 2.1.4 : Thu Dec 02 2004 - 06:57:51 GMT-3