RE: Login Failure and Sessions

["Liu, Hong" <Hong.Liu@neustar.biz>]

Scott,

I understand your concerns, but the retrial number N is really a
configuration parameter. It is not unusual to leave this type of parameters
for run-time configuration, take the windowing protocol as an example. The
size of the window is not fixed in the spec.

It may help, though, to give guidelines to selecting an appropriate value of
N in the spec.

--Hong

-----Original Message-----
From: Hollenbeck, Scott [mailto:shollenbeck@verisign.com]
Sent: Monday, August 05, 2002 12:40 PM
To: 'Liu, Hong'; ietf-provreg@cafax.se
Subject: RE: Login Failure and Sessions


> I agree with Patrick that this is a server policy issue. The 
> protocol should
> not specify the exact value of N.

If so, then we have a protocol with a non-deterministic state diagram.  The
state of the server WRT to login failures ends up being
implementation-dependent, and I think this is going to get us in trouble
with the IESG *.

-Scott-

* I say this because I've been told by our AD that we need a state diagram
in the specs, and I can't draw such a diagram if I can't document how the
server behaves when it has to deal with a client authentication failure.
I'm open to suggestions as to how this kind of implementation choice might
be described in a state diagram...