[c-nsp] BGP KEEPALIVE maximum frequency

[M. V.]

thank you Keegan.

> is it possible that these were just retransmissions cause by some sort 
> of error in your lab setup? 

i don't think that's the case here, 'cause each packet has its own sequence 
number, plus it always happens.

>  if it is a bug have you tried the test on multiple platforms and 
> multiple versions of IOS?  

actually i have. i have tested and seen this behaviour on 3 different systems:
1) virtual cisco3600 (IOS version: 12.3(22)) on Dynamips
2) a real-box "Cisco2821 (IOS version: 12.4(15)T13)
3) Quagga (cisco-like open-source routing software)
i don't think it's a bug. i think they've implemented it this way for a reason. 
i just can't figure it out.

> If I had to guess I would say that the BGP process may start before 
> the actual layer2/3/4 path is ready to carry the control traffic so 
> sending multiple keepalives is a way to overcome delays in things 
> like arp or even LSP establishment or something I've missed.

i'm not sure, but this could be the reason.

thanks again.