From: Gregg Malcolm (greggm@xxxxxxxxxxxxx)
Date: Thu Apr 11 2002 - 01:52:14 GMT-3
This is definitely not a real world question, only to help my understanding.
Here's the scenario :
Physical diagram :
R7---serial---R6---serial---R9
| |
60 <---------- 90 <------ Loopback/Token-ring Addresses
BGP Neighbor Connectivity and AS #'s
R6---peers---R7---peers---R9
65256 79 79
Test is to ping/trace between the 60 and 90 int's. Must use BGP routing for
reachabilty between these 3 routers. There are many issues as I see it with
this config.
Here are the IP route tables from the 3 routers:
R6
C 150.20.60.0/24 is directly connected, Loopback60
B 150.20.90.0/24 [20/0] via 150.20.50.7, 01:31:21 <----50.7 is R7
R7
B 150.20.60.0 255.255.255.0 [20/0] via 150.20.50.6, 01:32:52 <---50.6 is R6
B 150.20.90.0 255.255.255.0 [200/0] via 150.20.69.9, 01:33:10 <---69.9 is R9
R9
B 150.20.60.0/24 [200/0] via 150.20.50.7, 00:00:09
C 150.20.90.0/24 is directly connected, TokenRing0
I know the logical solution to this problem is to peer across the directly
connected interface (I.E. R6 should peer with both R7 and R9), but I don't
want to do that. One of the problems I'm seeing is when I try to ping from R6
to R9. R7 sends and ICMP redirect back to R6 saying that R6 know how to get
to R9, even tho the route table clearly indicates that the .90 subnet has the
correct entries in both R6 and R7. R6 is in fact the path that the packet must
take since it has the interface connected to R9. Then of course, R6 sends the
packet back to R7 again..classic routing loop. BTW - R7 cannot ping R9 either.
Has anyone else experimented with a config like this (or already knows the
solution) ? I ran across it accidentally when I was configuring local pref
and med.
Thanks, Gregg
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:58:04 GMT-3