Resource based kerberos constrained delegation

Martin Burkhart martin.burkhart at ergon.ch
Mon Jul 4 10:08:14 EDT 2016


Hi Greg

I am the product manager of the Single Sign-on solution Airlock. We are interested in adding support for resource based Kerberos constrained delegation (RBKCD) to our solution but currently miss the corresponding feature in krb5-libs. You have been discussing this before with Stefan Dietiker (see below). Therefore, I’d like to ask a couple of questions:

- According to your experience, what’s the estimated effort for adding RBKCD to krb5-libs? 
- Is RBKCD somewhere on the roadmap?
- Is there a way of sponsoring a feature? 

Thanks in advance for your time
Best regards
Martin

-- 
Dr. Martin Burkhart
Head of Product Management
Application Security
https://www.airlock.com  

martin.burkhart at ergon.ch +41 44 268 83 27
Ergon Informatik AG, Merkurstrasse 43, CH-8032 Zürich
http://www.ergon.ch    
______________________________________________________________
e r g o n    smart people - smart software

> -----Ursprüngliche Nachricht-----
> Von: Greg Hudson [mailto:ghudson at mit.edu]
> Gesendet: Dienstag, 28. Juni 2016 16:59
> An: Stefan Dietiker <stefan.dietiker at ergon.ch>; kerberos at mit.edu
> Betreff: Re: AW: Resource based kerberos constrained delegation
> 
> On 06/28/2016 06:03 AM, Stefan Dietiker wrote:
>> A few months ago I have asked you whether it is possible with
>> krb5-libs to do Resource Based Kerberos Constrained Delegation or not.
>> You mentioned that the Kerberos libs does not include the
>> PA-PAC-OPTIONS which are required for this purpose. Recently I was
>> tracking the changes in the git repo and realized that a new option
> "--request-pac" is available.
> 
> I don't believe this change bears any relation to resource based
> constrained delegation.  PA-PAC-REQUEST is different from PA-PAC-OPTIONS.
> 
> (I would also assume there is substantially more to implementing resource
> based constrained delegation on the client than just sending the
> PA-PAC-OPTIONS bit, or there would be no reason to have the bit in the
> protocol.)





More information about the Kerberos mailing list