BGP advertisement - why is it doing this?

John Blakley · ‎09-08-2009

All,

On a router (in gns) I did the following:

Dallas#sh ip bgp neigh 172.30.1.2 advertised-routes

BGP table version is 45, local router ID is 10.125.1.1

Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,

r RIB-failure, S Stale

Origin codes: i - IGP, e - EGP, ? - incomplete

Network Next Hop Metric LocPrf Weight Path

*> 10.125.1.0/24 0.0.0.0 0 32768 i

*> 10.125.100.0/24 172.30.1.2 0 13979 65100 i

r> 10.127.1.0/24 172.30.1.2 0 13979 65100 ?

*> 172.16.1.0/30 172.30.1.2 0 13979 65100 i

*> 172.30.1.0/30 0.0.0.0 0 32768 i

*> 172.30.3.0/30 172.30.1.2 0 0 13979 i

*> 172.30.100.0/30 172.30.1.2 0 0 13979 i

Dallas#sh ip bgp neigh 172.30.1.2 advertised-routes

BGP table version is 45, local router ID is 10.125.1.1

Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,

r RIB-failure, S Stale

Origin codes: i - IGP, e - EGP, ? - incomplete

Network Next Hop Metric LocPrf Weight Path

*> 10.125.1.0/24 0.0.0.0 0 32768 i

*> 172.30.1.0/30 0.0.0.0 0 32768 i

As you can see above, I did the "sh ... advertised" immediately after another and the routes were chopped down. There are no policies on this router that would affect bgp neighbors:

router bgp 65001

no synchronization

bgp log-neighbor-changes

network 10.125.1.0 mask 255.255.255.0

network 172.30.1.0 mask 255.255.255.252

neighbor 10.127.1.1 remote-as 65500

neighbor 172.30.1.2 remote-as 65000

no auto-summary

Its neighbor doesn't either:

router bgp 65000

no synchronization

bgp log-neighbor-changes

network 172.30.1.0 mask 255.255.255.252

network 172.30.3.0 mask 255.255.255.252

network 172.30.100.0 mask 255.255.255.252

neighbor 172.30.1.1 remote-as 65001

neighbor 172.30.100.1 remote-as 65100

no auto-summary

I'm at a loss. Debugging shows "duplicate dropped", but the whole point of redundant paths is to be able to have both in the table. =)

Thanks,

John

HTH, John *** Please rate all useful posts ***

Giuseppe Larosa · ‎09-08-2009

Hello John,

check what are the best routes on Dallas for the "disappeared" prefixes.

There are chances that Dallas has installed routes coming from neighbor 172.30.1.2 in a second time.

If this happens for a logical split horizon rule Dallas router stops to advertise the prefixes to the router it prefers.

So you see a transition that is the propagation time of BGP routes in your topology (I guess it is the same topology described in the other thread unless you have a server farm in your office running GNS ...:) )

Hope to help

Giuseppe

John Blakley · ‎09-08-2009

Giuseppe,

Is there a way to get around this? You're right; I only have routes from the 172.x.x.x address and not the other router that I need it from. It makes sense, but is there a way for me to be able to advertise the routes that I'm learning from the other router? If they're getting overridden in the bgp table, then there's nothing to advertise....

Thanks,

John

HTH, John *** Please rate all useful posts ***

Giuseppe Larosa · ‎09-08-2009

Hello John,

until you don't play with BGP attributes the shortest AS path wins.

to have the Dallas router to keep advertise the routes it has to prefer the advertisements coming from the other side/eBGP neighbor of your topology regardless of AS path attribute length:

you can use weight or local preference to achieve this.

BGP has a hierarchy of criteria for best path selection it uses it until it finds a best path.

May you explain what you would like to achieve?

Hope to help

Giuseppe

John Blakley · ‎09-08-2009

Giuseppe,

I'm trying to use a middle router as a transit router for failover:

1

/ \

2-3-4

I need 2 to advertise a route to 1, but I also need 3 to advertise it as well as a backup path. I've been able to modify the as path attribute, but it stops at the router that I just posted.

I've been able to get this to work with redistributing eigrp into bgp.

Thanks,

John

HTH, John *** Please rate all useful posts ***

Giuseppe Larosa · ‎09-08-2009

Hello John,

>> I've been able to get this to work with redistributing eigrp into bgp.

locally originated BGP routes have weight 32,768 including those coming from redistribution.

Hope to help

Giuseppe