RE: Man In The Middle Attacks and STARTTLS

Paul Hoffman / IMC <[email protected]> Thu, 15 Apr 1999 16:20:11 -0700
Newsgroups gmane.ietf.apps-tls
Message-ID <[email protected]>
At 03:54 PM 4/15/99 -0700, Scott Roberts (Exchange) wrote:
>Bodo, your points are good. I would think that it is better to state that a
>reasonable amount of time SHOULD pass before the message is returned to the
>sender rather than stating a default time because of issues that might arise
>in the future.

You are only seeing two options: use TLS or return to sender. The third 
option, which I think will get used much of the time, is send anyway. That 
is, various client-server pairs with lots of cycles to spare might be 
willing to use STARTTLS to encrypt because it only causes a second or two 
delay and, hey, why not. If TLS isn't available, they'd send the mail anyway.

>We could add to section 5 stating:
>         A client that receives a 454 reply code from a server that has
>advertised the STARTTLS verb SHOULD attempt to retry the mail in a
>reasonable amount of time or reroute it before returning it to the sender.
>The amount of time before retrying the message and/or NDRing the message
>should be left up to the implementation and is outside the scope of this
>document.

I think the wording needs to cover much more than that. I'll work on some 
and stick it in the first draft, and we can discuss from there.


--Paul Hoffman, Director
--Internet Mail Consortium