Em Thu, Jan 31, 2002 at 10:13:53AM -0600, Douglas E. Engert escreveu:
> > Also, I can also just query the kerberos server just like NIS if
> > preauthentication is not in place.
> 
> 
> You just answered your own queation. PREAUTHENTICATION
> 
> The paper if I remember correctly was addressing Kerberos V4. V5 
> addresses these problems with preauthentication. Password change protocols

Preauth with a timestamp is not enough either, it doesn't prevent an
eavesdropper. Quoting the paper:

"Although this [preauth] prevents an attacker from requesting TGTs, it
does not protect against an eavesdropper who captures either Ek(t) [sent
by the client] or Ek(tgt) [sent by the server, i.e., the reply]. Either
of those quantities constitutes verifiable plaintext that can be used to
mount a dictionary attack."

In fact, I have another paper where a student at a university here made
such attacks against W2K, which uses timestamp-preauth. I don't have
an URL now, it's at home, sorry.

> can force passowrd rules too. 

Yes, weak passwords will always be a problem. This is a moving target,
what is weak today wasn't weak 10 years ago.

Reply via email to