From: ccie candidate (ccie1@xxxxxxxxx)
Date: Thu Aug 15 2002 - 16:39:26 GMT-3
claudine ;
i believe this normal behaviour since the default route propagate into the Eigr
p as external to the domain so that it is admin distance is 170 which is greate
r than the rip (120).
since you have two points of redistribution , the situation of Blackhole should
definitely apprear .
on R2 for example , the default route comes twice , one from R1 with admin dist
ance of 170 and from R3 with admin distance of 120 ( becausse of redistribution
on R3) ..now R2 will believe R3 !
the same will happen on R3 , he will not beleive R1 ..but R2 .
u end up that R2 pointing to R3 and R3 points to R2 !
the solution for this is explained clearly in doyle vol1 .
there are two ways to get rid of this problem
1- in the two redistribution points , you filter the routes learnt from certain
domain to be learnt on the same router from the other domain ...simply on R2 f
or example , prevent Eigrp routes in the incoming ethernet [R2 will not learn t
he default route from R3 by distribute list in its ethernet .
2- by changing the admin distance so that the routes coming from one domain sho
uld remain learnt from the same domain .
for example , decrease the admin distance in R2 for the default route and same
for R3
HTH
--On Thu, 15 Aug 2002 18:40:39 Claudine DEMAR wrote: >Scenario & problem: Propagating a default route from eigrp to rip caused a >routing loop ?? > > |lo0 192.168.1.0/24 > R1 > / \ > / \ >R2 R3 > | | >-------- 192.168.234.0/24 > > >Serial hsrp connections between R1-R2 (192.168.12.0/24) and R1-R3 >(192.168.13.0/24) are running EIGRP >R2 and R3 are on an Ethernet segment which is running RIPv2. > >R1: >router eigrp 1 >net 192.168.1.0 >net 192.168.13.0 >net 192.168.12.0 > > >R2: >router eigrp 1 >redistribute rip metric 10000 1000 254 1 1500 >net 192.168.12.0 > >router rip >ver 2 >redistribute eigrp 1 metric 14 >net 192.168.234.0 > > >R3: >router eigrp 1 >redistribute rip metric 10000 1000 254 1 1500 >net 192.168.13.0 > >router rip >ver 2 >redistribute eigrp 1 metric 2 >net 192.168.234.0 > > >The problem appeared when the I wanted to propagate a default route. >Lo0 of R1 is supposed the default route. > >On R1: >ip default-net 192.168.1.0 > >This causes the 192.168.1.0 tagged as the default route in the EIGRP cloud. >This is redistributed in RIP but I did not expect R2 and R3 to have both >default routes pointing to each other: >R2: >route 0.0.0.0/0 metric 2, next-hop is R3 > >R3: >route 0.0.0.0/0 metric 14, next-hop is R2 > >This default route never times out !!!! I expected to have R2 pointing to R3 >but not R3 pointing to R2. >This RIP and this default route is definitively behaving wierdly. >I know split horizon might be involved but I don't understand why there's no >timeout of the route whereas each router does not receive any updates from >the other .... > >Anybody have an explanation for this ?? > > >The solution I found was to advertise 0.0.0.0 to R2 and R3 as a EIGRP >internal route (distance 90). >In order to do that I just defined net 0.0.0.0 under router eigrp in R1. >The problem with that is that all connected interfaces are advertised with >that network statement. > >Any other suggestions ? > > > > > > > > > > > >
This archive was generated by hypermail 2.1.4 : Sat Sep 07 2002 - 19:48:26 GMT-3