On Mon, 8 Jul 2002 16:27:34 +0100 (BST) Philip Hazel <ph10@???> wrote:
> On Mon, 8 Jul 2002, Phil Chambers wrote:
>
> > I am concerned that just disconnecting may result in a sending site re-connecting
> > straight away and causing a loop. An alternative would be to just dump any
> > further input until the sending site finishes and disconnects. If the sending site
> > thinks it has finished then it is more likely to go away.
>
> Do you mean give no more responses, but just swallow input? In theory,
> the sender shouldn't send any more input unless it gets a response. So
> it will time out - which is equally likely to make it send again.
>
> > I presume that in the case of the HTML based attack there would be no loop because
> > the web browser would not keep re-trying, but other situations might be different.
> >
> > Am I being too pessimistic?
>
> I doubt whether one can be too pessimistic in the matter of
> junk/spam/attacks. :-(
>
> However, I don't see that there is anything to be gained, and you tie up
> your process longer.
>
> Or do you mean send 5xx responses to all subsequent commands?
I think it would have to be a case of sending 5xx to everything to prevent the
likelyhood of the sender timing out waiting for a response.
Phil.
---------------------------------------
Phil Chambers (postmaster@???)
University of Exeter
