Clarifications sought on Kerberos SA: TGS_REQ and Server Auth??

Surendra Babu A surendra.a at samsung.com
Mon Dec 12 00:54:50 EST 2005


Hi,

Could you please explain how can user enjoy the Server Authentication
feature with IP address of the KDC? If I give host name, it will form the
correct principal in TGS_REQ and send that to KDC.

If I give IP Address of the Server, forming of TGS_REQ is failed. How can
Client, KDC and Server agrees up on single principal name? That means, Host
name of server to be configured some where? if so, where?

My aim is: without using the Host name (FQDN) of Server, I should be able to
form TGS-REQ and get TGS-REP.

Is it possible? Please let me know your thoughts.

Thank you,
-Surendra


----- Original Message ----- 
From: "Douglas E. Engert" <deengert at anl.gov>
To: "Surendra Babu A" <surendra.a at samsung.com>
Cc: <kerberos at mit.edu>
Sent: Wednesday, December 07, 2005 7:34 PM
Subject: Re: Clarifications sought on Kerberos SA: TGS_REQ and Server Auth??


>
>
> Surendra Babu A wrote:
>
> > Dear Team,
> >
> > Thanks a lot for the response.
> >
> > Issue#1. Regarding Hostname in Server Pricipal:
> > So it is better to use the hostname in the prinicpal.
>
> Yes. As the FQDN is unique, and both the client and server can agree that
that is
> the name of the server.
>
> > So for Server Authentication, HOst name should be known. Right?
>
> Hopefully. If is in not, then the client has to know the principal name
that
> the server is using, so it can ask the KDC for a ticket for that service.
> The user/client, KDC and service must all agree on how services are named,
> so as to avoid attacks where the user is directed to the wrong service
> withoutknowing it.
>
> Host names fit this pattern. A KDC admin would/should not add a principal
> for host.x.y and give the key to anyone but the admin of the host.x.y
>
> A user may use the command "ssh host.x.y" to establish a session to the
host,
> The kerberos libs use the host.x.y in the principal. The conection is made
and
> authenticated in both directions because the names have ben agreed upon in
advance.
>
>
> >
> > Without knowing the Hostname, it is diffifcult to resolve the Server
Authentication. Right?
> >
>
> Yes, but not impossible. If the user client, KDC and server can agree on
what principals
> are to be used, then you could get Kerberos to work without host names.
>
> > Additionally, Application Server side, should we configure the KDC
server IP address?
>
> That is an  implementaiton not a protocol convention. I believe most
Kerberos libs expect
> hostnames in the krb5.conf file. Never tried it with an IP address. Using
names can also
> allow for multiple IP addresses on a single KDC.  But the name of the KDC
or its IP address
> are not used in tickets.
>
> > I think It is not required. Because we are already sending the Server
Pricipal in TGS-REQ. Right?
>
> If it works, then you can use it.
>
> >
> > Conclusion: To enjoy the benefit of Server Authentication of Kerberos,
User should enter the Hostname of Applciation Server. Right?
> >
>
> Yes.
>
> >
> > Issue#2: Regarding GSSAPI.
> > My assumption is, if ESMTP supports Kerberos, then it will give AUTH
GSSAPI response for the EHLO command.
>
> If is supports GSSAPI, then forget about using Kerberos and getint tickets
... and use GSSAPI!
> The Kerberos veriosn of GSSPAI does all of that for you.
>
>
> >So I am sending the Servcie Ticket from our esmtp client. Is it right?
>
> Don't know anything about ESMTP or how it uses Kerberos. Sorry.
>
> >
> > Let me expalin clearly:
> > - AS_REQ and AS_REP happend and User Authentication is done
successfully.
> > - TGS_REQ and TGD_REP is happend with Kerberos client and KDC. So
Service tkt is at kdc client side.
> >
> > - Now we initiated the email operation from  Multi Funtion Printer. It
contacted the ESMTP server from ESMTP client.
> > - Sent the EHLO command. ESMTP Server Gave me the AUTH GSSAPI response.
So Kerberos is supported at server sid. Right??
> > - After seeing this GSSAPI string, sending the Service ticket to ESMTP
server.
>
> If the server is expecting GSSAPI, then the cilent should be use the
Client side of GSSAPI, which is
> a generic layer on top Kerberos.
>
> >  Is it correct? We are assuming that, GSSPAI is supported means,
Kerberos is supported. Is it right?
>
> Maybe. GSSAPI is a generic API for security systems. Kerberos is one of
these. 99% of the time
> this refers to Kerberos.
>
> >
> > Could you please explain about the Application Server transactions???
How to communicate with GSSAPI?
>
> No. Google for gssapi.
>
>
> >
> > Thank you,
> > -Surendra
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > ----- Original Message ----- 
> > From: "Douglas E. Engert" <deengert at anl.gov>
> > To: "Surendra Babu A" <surendra.a at samsung.com>
> > Cc: <kerberos at mit.edu>
> > Sent: Wednesday, December 07, 2005 5:11 PM
> > Subject: Re: Clarifications sought on Kerberos SA: TGS_REQ and Server
Auth??
> >
> >
> >
> >>
> >>Surendra Babu A wrote:
> >>
> >>>Hi Kerberos Team,
> >>>
> >>>Could you please let me know your thoughts on the following questions?
Thank a lot in advance,
> >>>
> >>>1. While forming the TGS-REQ pkt, I need to send the Server name with
that TGS_REQ packet. For
> >>
> >>this reason, I need to use krb5_parse_name(). second Parameter for this
API is a Server Principal.
> >>
> >>Should I need to send a qualified Hostname with that?
> >>
> >>As long as the server and the client agree on what is in the name, and
the principal
> >>is regstered in the KDC, and the server has a copy of the key, it can
work.
> >>
> >>Servers usually have a two component name and the realm: <service>/<fqdn
of host>@<realm>
> >>and many of the Kerberos routines assist in making sure the host is
converted to a fqdn.
> >>
> >>You could have more components, DCE had some three component names.
> >>
> >>You could use IP addresses, but IP addresses don't real identify a host,
they
> >>identify an interface. Hosts with multiple interfaces, VPNs, and NAT can
> >>make this dificult. IP numbers change, so the client, server and KDC all
need
> >>to be updated. And what will you do about IPV6 addresses in a principal?
> >>
> >>Use names if you can, DNS or even names are in pre distributed
/etc/hosts
> >>files.
> >>
> >>
> >>  That means, we should know the Host NAme of
> >>
> >>the server? Without knowing the HOst Name of the Application Server
(i.e. If we know only IP Address),
> >>
> >>can't we form the TGS_REQ packet and get the successful response
TGS-REP?? I tyried with IP Address in
> >>
> >>Principal. But it was not successding. COuld you please let me know your
thoughts?
> >>
> >>
> >>
> >>
> >>>2. For Server Authentication feature: if the Application Server is a
Kerberised ESMTP server, how it
> >>
> >>should proceed? After sending the Service ticket to ESMTP server, what
should happen? Could you please
> >>
> >>let me know the Client and Applciation Server handshake and transfer
machanism till Server Authentication
> >>
> >>feature happens?
> >>
> >>Rather then using raw Kerberos, can you use gssapi? Gss addresses many
of these issues.
> >>
> >>
> >>
> >>>Please let me know your thoughts.
> >>>
> >>>Thank you,
> >>>-Surendra
> >>>________________________________________________
> >>>Kerberos mailing list           Kerberos at mit.edu
> >>>https://mailman.mit.edu/mailman/listinfo/kerberos
> >>>
> >>>
> >>
> >>-- 
> >>
> >>  Douglas E. Engert  <DEEngert at anl.gov>
> >>  Argonne National Laboratory
> >>  9700 South Cass Avenue
> >>  Argonne, Illinois  60439
> >>  (630) 252-5444
> >>
> >>
> >
> >
>
> -- 
>
>   Douglas E. Engert  <DEEngert at anl.gov>
>   Argonne National Laboratory
>   9700 South Cass Avenue
>   Argonne, Illinois  60439
>   (630) 252-5444
>
>



More information about the Kerberos mailing list