But IIRC the Solaris "su" can use PAM, and thus could use pam_krb5 or
some other PAM routine to acomplish the same thing as the MIT su?


Heilke, Rainer wrote:

Thanks for the response. Please see inline...


In Solaris 10, all of the Kerberos services are already bundled,
there is no longer any external packages that need to be added.


Right.


Whoever told you 'ksu' was part of the encryption kit was mistaken,
ksu has never been part of SEAM.


OK, thanks for that clarification. It was a bit of a surprise to me when
I was told it was there. So, does the Solaris 10 SEAM have any
functionality similar to ksu, or just the standard su command?


The encryption kit for Solaris 10 enhances the overall crypto
capabilities of the system, the only benefit Kerberos gets is
that it can support AES-256 with the S10 encryption kit.
Without the S10 encryption kit, the strongest AES crypto
available for Kerberos in S10 is AES-128.


And this fits more with what I understood, before my co-worker's
comments.


On the S10 system, you must make sure to enable the "eklogin" service.
Run this command (as root):

# svcadm enable eklogin


Hmm. That may be a good part of my problem. I added the inetd.conf entry
for the old (MIT) eklogin, and ran inetconv. So, this is probably really
confusing the system. I'll try to revert that, and do the svcadm.


For Solaris 8 with the SEAM rlogin daemon, make sure your inetd.conf entries
are correct.


We don't actually run SEAM on any Sol8 systems; it's all MIT.


Don't bother with inetd.conf in S10, S10 uses the new SMF
system for managing services (its very nice once you get used to it).


I'm still getting used to it, but I like it. I just wish there was more
info on how to convert from old-style to new-style (beyond running
inetconv -i inetd.conf).


You indicated below that you are using and MIT kerberos KDC on the
Solaris 8 systems.  So, the key to making things work with the S8
SEAM kerberos clients is to make sure that the host principals
for those Solaris 8 systems are only issued DES keys.   The rlogin
servers in SEAM only support DES since that is all that was
available when the S8 SEAM packages were created.

'kadmin -q 'addprinc -e des-cbc-md5:normal host/foo.bar.com"'
'kadmin -q 'ktadd -e des-cbc-md5:normal host/foo.bar.com"'

(Im not sure if the syntax for those commands is exactly correct,
but you get the idea).

Solaris 10 systems can be issued AES keys (AES-128 if the encryption
package is not installed, AES-256 otherwise) or RC4, 3DES, or DES.


Can we force the Sol10 box to only use DES, to be compatible with the
Sol8/MIT systems (which is everything but the one Sol10 box)?


SSH in Solaris 10 does support GSSAPI authentication and ticket
forwarding, you may need to enable debugging to get more information
about why the GSS auth is failing (ssh -vvv  hostname).   If the host
keys on S10 are AES-256, but the system doesnt have the enhanced
crypto package, then that may be causing the failure, but I would
check the debug output from ssh first.


OK, thanks. I'll poke harder at this once I have the previous issues
resolved. I'll make SSH a separate problem. (I think I can do that...)


Solaris 10, SEAM going one way, Solaris 8/MIT going the other.

We have tested this and it does work, but you have to make sure that the S8 system has only DES keys.


All Solaris 8 systems are MIT, so if I understood your earlier comments,
they already are DES; is that correct?


Solaris 8, MIT for rlogind in one direction, and Solaris

10/SEAM in the

other. Typically, I am SSH'ing in to the Solaris 10 system initially
(from our production lab) and then trying to rlogin to the

Solaris 8/MIT

systems in the test lab (where the Solaris 10 system sits).

I have also

SSH'd to a test lab Solaris 8/MIT system, and tried to rlogin to the
Solaris 10/SEAM system.

All of this should work, start by getting detailed log information from the
SSH session to see why that part is failing.


Thank you for all of the help. I'll look at correcting the eklogin
service first, and move down the list from there. :-) Answers to the
above questions/clarifications will be greatly appreciated along the
way. :-)

Thanks again.
Rainer

________________________________________________
Kerberos mailing list           [email protected]
https://mailman.mit.edu/mailman/listinfo/kerberos




--

 Douglas E. Engert  <[EMAIL PROTECTED]>
 Argonne National Laboratory
 9700 South Cass Avenue
 Argonne, Illinois  60439
 (630) 252-5444
________________________________________________
Kerberos mailing list           [email protected]
https://mailman.mit.edu/mailman/listinfo/kerberos

Reply via email to