[c-nsp] limited bgp traps

Saku Ytti saku at ytti.fi
Tue Sep 6 04:31:09 EDT 2011


On (2011-09-06 12:03 +0400), Nikolay Shopik wrote:

> I understand this because it actually changes state from down to
> active(trying to establish tcp session) and after timeout goes down
> again thus sending trap again.
> 
> Is this behavior by design? I don't actually expect receive another
> trap until it change state from what I define(down or up).

Unfortunately this is by design, people who designed BGP traps were dreadfully
confused what people need. Now you must save previous BGP state in your NMS, so
you are able to determine what happened.

However some platforms, like IOS, extend standard BGP traps and include
'previous state' field, this will allow you to react only on establihes->down
and X->established events, without saving states in NNS. Some other platforms,
like JunOS unfortunately do not provide you with this luxory.

-- 
  ++ytti


More information about the cisco-nsp mailing list