From: Dennis Dumont (dfdumont@xxxxxxxxx)
Date: Wed Feb 20 2002 - 15:03:10 GMT-3
I've got an interesting situation while playing around
with obscure BGP commands. I've got a six router pod
with three conections to an external AS-254, and
another connection to an external AS-253. The two
externals are themselves connected behind the pod.
The two central routers have connections to all of the
edge routers. All routers in the pod are the same
AS-100.
There are no direct connections from the edges to any
routers other than R3 and R4.
Here's the funny part. With normal behavior, the
selected paths on R3 (the test box for this) BGP
chooses the shortest AS-path, like its supposed to. I
have the three paths to AS-254 each with specific MEDs
for later fun, but the MED's are ignored unless trying
to break a tie between equal-length AS-paths.
When I turn on the above command on R3, I don't get
the best path as determined now by the MED, but rather
I get the preferrence of the LONGEST AS-path!
Here's R3 BGP block:
router bgp 100
no synchronization
bgp log-neighbor-changes
bgp bestpath med missing-as-worst
bgp bestpath as-path ignore
neighbor local peer-group
neighbor local remote-as 100
neighbor local password cisco
neighbor local update-source Loopback0
neighbor local soft-reconfiguration inbound
neighbor 176.14.1.1 peer-group local
neighbor 176.14.2.2 peer-group local
neighbor 176.14.4.4 peer-group local
neighbor 176.14.5.5 peer-group local
neighbor 176.14.6.6 peer-group local
Here's the BGP table - note the selected paths:
Network Next Hop Metric LocPrf
Weight Path
* i192.14.1.0 176.14.2.2 0 100
0 253 i
* i 176.14.5.5 200 100
0 254 253 i
* i 176.14.6.6 300 100
0 254 253 i
*>i 176.14.1.1 4294967294 100
0 254 252 253 i
* i192.14.2.0 176.14.2.2 0 100
0 253 i
* i 176.14.5.5 200 100
0 254 253 i
* i 176.14.6.6 300 100
0 254 253 i
*>i 176.14.1.1 4294967294 100
0 254 252 253 i
* i192.14.3.0 176.14.2.2 0 100
0 253 i
* i 176.14.5.5 200 100
0 254 253 i
* i 176.14.6.6 300 100
0 254 253 i
*>i 176.14.1.1 4294967294 100
0 254 252 253 i
* i192.14.4.0 176.14.2.2 0 100
0 253 i
* i 176.14.5.5 200 100
0 254 253 i
* i 176.14.6.6 300 100
0 254 253 i
*>i 176.14.1.1 4294967294 100
0 254 252 253 i
* i192.14.5.0 176.14.2.2 0 100
0 253 i
* i 176.14.5.5 200 100
0 254 253 i
Network Next Hop Metric LocPrf
Weight Path
* i 176.14.6.6 300 100
0 254 253 i
*>i 176.14.1.1 4294967294 100
0 254 252 253 i
* i192.14.6.0 176.14.2.2 0 100
0 253 i
* i 176.14.5.5 200 100
0 254 253 i
* i 176.14.6.6 300 100
0 254 253 i
*>i 176.14.1.1 4294967294 100
0 254 252 253 i
* i200.200.220.0 176.14.2.2 4294967294 100
0 253 254 i
* i 176.14.5.5 200 100
0 254 i
* i 176.14.6.6 300 100
0 254 i
*>i 176.14.1.1 0 100
0 254 252 i
* i200.200.221.0 176.14.2.2 4294967294 100
0 253 254 i
* i 176.14.5.5 200 100
0 254 i
* i 176.14.6.6 300 100
0 254 i
*>i 176.14.1.1 0 100
0 254 252 i
Has anyone else seen this? OR is my code bad [12.1(11)]
This archive was generated by hypermail 2.1.4 : Thu Jun 20 2002 - 13:46:28 GMT-3