Re: [DNA] retaining current IP configuration as DNA kicks in

[Greg Daley <gregd@research.panasonic.com>]

Hi Thomas,

Thomas Narten wrote:
[cut]
>> I'd like to make a clarification. IMO, upon "link up" event,
>> though a host can keep the existing configuration information,
>> it should not use the information until DNA procedure confirms
>> that the host remains at the same link.
> 
>> For example, a host can retain its current IP address but should not send
>> a packet with the address as a source address till DNA comletes.
> 
> Why?
> 
> I've thought about this some, and the question I ask is "what breaks
> if a host does send such packets?" If that answer is "nothing", then
> no harm done.
> 
> It may be that sending such packets is useless, but that's not a
> problem. If it actually causes things to break or bad things to
> happen, then yes, we should think about disallowing such behavior. But
> lets evaluate the cost/benefit explicitly.

I agree.

This was actually one of the first topics on the mailing list a _long_
time ago.  There's no need to specify that packets can't be sent until
DNA completes, as globally addressed packets (except ND) can't interfere
with packet streams when sent from another link.

There may be problems for TCP if these packets are wasted, but I
don't think that's necessarily DNA's direct responsibility at
this stage (we don't even have the base protocol standardized).

Greg