Re: soft state (was Re: shim6 and bit errors in data packet headers

[Iljitsch van Beijnum <iljitsch@muada.com>]

On 27-mei-2005, at 16:17, marcelo bagnulo braun wrote:

> > Also note that the feedback doesn't always have to be from the  
> > actual transport protocol: just observing packets flowing in both  
> > directions will tell you that all is well.

> well, this wouldn't really tell you that the path is working for  
> sure, right?

Node A could be sending and receiving while there is a failure, but  
in that case, B wouldn't be receiving data. So if both sides do this,  
it works well.

(Ok, this has the potential to be a big problem: what happens if  
there is significant packet loss, but enough packets make it so the  
shim doesn't see the problem?)