RE: Login Failure and Sessions

[Edward Lewis <edlewis@arin.net>]

For what it matters, speaking as a state machinist and not as the chair:

        |
        |  /-------------------\
        |  |                   |
        v  v                   |
    +-------+                  |
    | login |--fail < N tries--/
    +-------+
     |     |
  success  fail
     |     >= N tries
     |     \
     |      ------------------>
     v

This is roughly how I would draw option 2 - not that I am suggesting 
that option 2 is the preferred solution.  This is a shorthand, 
strictly speaking you shouldn't have state (the value of N) on the 
arcs.  But I think this would fly in a protocol specification.

At 3:41 PM -0400 8/5/02, Hollenbeck, Scott wrote:
>>  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.
>
>You didn't address the problem I presented: the need to craft a
>deterministic state diagram.  If what you're saying, though, is that you
>prefer option 2 (blast the connection after N failures) with the value of N
>(0 <= N <= inf) to be determined by the server operator, I can work with
>that.  That's at least easier to describe formally than "it's a policy
>issue". ;-)
>
>-Scott-

-- 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis                                          +1-703-227-9854
ARIN Research Engineer