• Re: is there a way to detect if user is using same incorrect password in authentication

    From Ken Hornstein@kenh@cmf.nrl.navy.mil to Jim Shi on Fri Aug 9 21:03:01 2024
    From Newsgroup: comp.protocols.kerberos

    Hi, we have a required to detect -aif a client is using same incorrect >password in in authentication against KDC. Is it possible the KDC
    server can determine if client is using same incorrect password? Thanks

    Ouch, is this some dang compliance requirement? I thought I had dealt with
    SO MANY weird compliance issues, but that's a new one to me. I'm interested
    in where this is coming from. If I understand you, it seems like you mean
    that a single client is repeating the same incorrect pasword over and over.
    If you mean that different clients are trying to use the the same incorrect password, I don't believe that's possible (nor do I understand why that
    would be a requirement). Upon further thought, this seems like a completely ridiculous requirement and I cannot imagine why anyone would ask for it.

    I _think_, in theory ... my first guess as to what you mean is possible.
    But it won't be trivial. I believe you could accomplish this by using
    encryped timestamp preauth, detecting when a wrong password is seen, remembering that on the KDC, and then sending the same encrypted timestamp
    back to the client upon further password requests and detecting if the
    response was the same. That would be a lot of code and have issues if
    the requests went to different KDCs. It's very possible I could be wrong
    about that. And again, that only works with requests from the SAME client
    due to password salting.

    --Ken
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Ken Hornstein@kenh@cmf.nrl.navy.mil to Brent Kimberley on Sat Aug 10 12:28:07 2024
    From Newsgroup: comp.protocols.kerberos

    The definition of an argon salt is predicated on a nonce - number used
    once. Reusing the salt, changes the definition.

    I ... don't see how that's relevant to this discussion? The document
    you linked is not used by any extant Kerberos implementation that I am
    aware of.

    --Ken
    --- Synchronet 3.21d-Linux NewsLink 1.2