From: Moffat, Ed (EMoffat@fsci.com)
Date: Mon Apr 05 2004 - 02:03:35 GMT-3
The reason for this is that BB3 is not going to announce a route to BB1 if
it has already determined that the best path is through BB1.
Actually, it initially will announce the route to BB1, then it gets a route
announcement from BB1 with a better path, so BB3 will withdraw its
announcement of the route.
The same is true for the opposite direction.
As you noted, if the best path (through BB1) goes away, it will announce its
new best path to BB1.
You can check this out through deb ip bgp updates.
-Ed-
-----Original Message-----
From: Ahmed Mustafa [mailto:ahmed.mustafa@sbcglobal.net]
Sent: Sunday, April 04, 2004 1:04 PM
To: ccielab@groupstudy.com
Subject: BGP Redundant Paths
BGP 100 multihoped with BGP 54
There are two router in AS 54 BB1 and BB3 both are IBGP peer.
BB1 is connected to R6 via ATM Link
BB3 is connected to R5 via Ethernet Link
Both R6 and R5 are in AS 100.
As 100 is advertising network 174.1.3.0/24, 174.1.4.0/24, 174.1.7.0/24, and
174.1.44.0/24
As you can see on below:
i174.1.3.0/24 54.1.8.6 100 0 100 i
*> 204.12.1.5 0 100 i
* i174.1.4.0/24 54.1.8.6 100 0 100 i
*> 204.12.1.5 0 100 i
*> 174.1.7.0/24 204.12.1.5 0 100 i
* i 54.1.8.6 100 0 100 i
* i174.1.44.0/24 54.1.8.6 100 0 100 i
*> 204.12.1.5 0 100 i
I set a MED policy for networks 174.1.4.0 and 174.1.44.0 learned through ATM
link, and networks 174.1.3.0 and 174.1.7.0 through Ethernet Link.
I defined my configuration on Border router R5 and R6
Router 5
!
ip prefix-list VLAN4&44 seq 5 permit 174.1.4.0/24 ip prefix-list VLAN4&44
seq 10 permit 174.1.44.0/24
ip prefix-list VLAN3&7 seq 5 permit 174.1.3.0/24 ip prefix-list VLAN3&7 seq
10 permit 174.1.7.0/24
route-map MED permit 10
match ip address prefix-list VLAN4&44
set metric 20
!
route-map MED permit 20
match ip address prefix-list VLAN3&7
set metric 10
!
route-map MED permit 30
R6
ip prefix-list VLAN4&44 seq 5 permit 174.1.4.0/24 ip prefix-list VLAN4&44
seq 10 permit 174.1.44.0/24
ip prefix-list VLAN3&7 seq 5 permit 174.1.3.0/24 ip prefix-list VLAN3&7 seq
10 permit 174.1.7.0/24
route-map MED permit 10
match ip address prefix-list VLAN4&44
set metric 10
!
route-map MED permit 20
match ip address prefix-list VLAN3&7
set metric 20
!
route-map MED permit 30
Now on my BB3 and BB1, I achieved exactly I wanted except
BB3
*> 174.1.3.0/24 204.12.1.5 10 0 100 i
*>i174.1.4.0/24 54.1.8.6 10 100 0 100 i
* 204.12.1.5 20 0 100 i
*> 174.1.7.0/24 204.12.1.5 10 0 100 i
*>i174.1.44.0/24 54.1.8.6 10 100 0 100 i
* 204.12.1.5 20 0 100 i
BB1
* 174.1.3.0/24 54.1.8.6 20 0 100 i
*>i 204.12.1.5 10 100 0 100 i
*> 174.1.4.0/24 54.1.8.6 10 0 100 i
*>i174.1.7.0/24 204.12.1.5 10 100 0 100 i
* 54.1.8.6 20 0 100 i
*> 174.1.44.0/24 54.1.8.6 10 0 100 i
on BB3 it only shows the best bath for network 174.1.3.0 and 174.1.7.0 and
wouldn't show the redundant path through 54.1.8.6 and opposite on BB1. If I
were to bring the Ethernet link down for 174.1.3.0 and 174.1.7.0 then
ofcourse I will see those prefixes are reachable through 54.1.8.6.
The question why it would not show those routes also reachable through
54.1.8.6 as it showing for networks 174.1.4.0 and 174.1.44.0 via two paths,
and same for BB1, but in opposite directions.
Sorry for the long email.
Regards,
Ahmed
This archive was generated by hypermail 2.1.4 : Mon May 03 2004 - 19:48:42 GMT-3