BGP Dampening???

[Mavrick25]

Hello Everyone,

What happens if you don't have BGP Dampening on??

Does it grab default settings?? If so what are they?

We have some internal confederations within our network and we notice that the ebgp peering we have with them isn't stable.

I receive the following message all the time:

BGP-3-NOTIFICATION: received from neighbor x.x.x.x (cease) 0 bytes

BGP-5-ADJCHANGE: neighbor x.x.x.x Down BGP Notification received

BGP-5-ADJCHANGE: neighbor x.x.x.x Down NSF peer closed the session

All the time.. Let me know

[milan.kulik]

Hi,

no, it's disabled by default.

See http://www.cisco.com/en/US/docs/ios/12_3/iproute/command/reference/ip2_a1g.html#wp1039828

for details and default values.

IMHO, the messages you see are just saying the BGP peering went down (neigbor keepalive timeout possibly?).

BR,

Milan

[marikakis]

I cannot see how dampening is related to those messages. BGP Route Flap Dampening has to do with suppressing unstable routes, not suppressing/closing BGP neighbors. Can you please check if you have a physical layer issue between the neighbors (link flap)? Depending on the interconnection method used between the routers, such issues might be observed even as layer 2 issues or you might just see the BGP flapping (you have to check all individual physical segments between the routers). Also check for heavy load conditions or other malfunctions in both router endpoints of a single BGP session.

[Giuseppe Larosa]

Hello Paolo,

other side is sending a notification message asking to close the eBGP session.

BGP-3-NOTIFICATION: received from neighbor x.x.x.x (cease) 0 bytes

BGP-5-ADJCHANGE: neighbor x.x.x.x Down BGP Notification received

BGP-5-ADJCHANGE: neighbor x.x.x.x Down NSF peer closed the session

have you added the neigh x.x.x.x ebgp-multihop on the other side? I think yes because the session was established.

eBGP confederation sessions are eBGP sessions in some aspects.

if it was a routing problem you would see timeouts unless they happen on the other side that doesn't see all BGP keepalives coming from local node.

as noted by Maria it can be a question of heavy load on the other device if it is a route reflector server within the mini AS

Hope to help

Giuseppe

[marikakis]

Hi Giuseppe,

I have seen constant physical link flaps causing high load (using real-time cpu usage of critical routers on my screen) in routers running BGP with many routes (both when link goes up and when it goes down). Do you consider this scenario impossible?

Kind Regards,

Maria

[Giuseppe Larosa]

Hello Maria,

I don't consider impossible the scenario you are describing.

Hope  to help

Giuseppe