Re: [dhcwg] RFC4388: Revision or new option needed?

[Niall O'Reilly <[e-mail]>]

	Kim,

	Thanks for your thoughtful reply.

	In particular, I liked the following.

On 03/06/10 16:10, Kim Kinnear wrote (much more than this fragment):

> One thought that I've had for years but never acted upon would be
> to write a general draft for a new option which would be a
> "return to sender" option.  This option would override the normal
> addressing for a return DHCP packet(and would work on any packet,
> not just leasequery).  It would cause the return packet to be
> sent to the IP and port of the sender regardless of the "normal"
> RFC 2131/4388/etc.  mandated behavior.
>
> This would be useful for a number of scenarios (including
> testing).
>
> I could also imagine such an option have a couple of flags, one
> which said "send reply to incoming IP address", and another which
> said "send reply to incoming port".  That way you could get just
> the port behavior you want, but others could use it in general.

	Neat!

	Some reference to protection against indirect DoS attacks
	may need to be mentioned, along the lines of the "... SHOULD
	protect ..." matter in 4388.

> In any case, I would support a draft to fix up RFC4388 in some
> way that doesn't break existing clients that depend on the
> current behavior.

	Great!

	Thanks again, and best regards.


	Niall O'Reilly