Damon Getsman schrieb:
Well first of all I'd very much like to thank you for that detailed explanation of the pam modules. It's helping me to understand much better what is going on in those pam scripts, especially /etc/pam.d/gdm, which is where I believe the issue is most probably taking place.

I did run across a module that you didn't mention, however. <sunlib>/pam_kiosk.so, and I've been unable to find any more detailed information than the manpage for it (in /opt/SUNWKio...). The manpage is still leaving me a little confused, as it appears that this module has many different possible behaviors.

The core sentence for your case is:

     1. [all modules] Return PAM_IGNORE and  perform  no  further
        action,  if  the  session is known not to be a Kiosk ses-
        sion.

That means that unless you have configured some/all X sessions to be kiosk sessions (using the kioskconfig(8) tool or the SRSS functionality built on top of it) this module, will do nothing and return PAM_IGNORE. What is unclear about this description?

For most intents and purposes this means that, if you know that you'll never need kiosk mode, you could remove these entries from the pam.d/gdm file without doing any harm. OTOH this should not change anything for your PAM processing and SRSS may restore the entires it has made (including the pam_kiosk ones) under certain circumstances.

Honestly I'm not even sure what kiosk mode would be used for in our setting, as every user has to authenticate at the gdm screen and we have no unauthenticated sessions. This is probably due to the fact that I don't have much understanding of the module, however.


Typical uses include running a full screen Sun Ray Windows Connector session or running another full-screen frontend to some network-based service (e.g. in a stripped down web browser) without requiring a Unix login.

This is the context in which it appears:

-=-=-
#%PAM-1.0
password include        common-password
# BEGIN: added to gdm by SunRay Server Software -- gdm
auth requisite /opt/SUNWut/lib/sunray_get_user.so.1 property=username
auth required /opt/SUNWut/lib/pam_sunray_amgh.so.1
auth sufficient /opt/SUNWkio/lib/pam_kiosk.so log=user ignoreuser
auth requisite /opt/SUNWkio/lib/pam_kiosk.so log=user
auth required /opt/SUNWut/lib/sunray_get_user.so.1 prompt
auth required /opt/SUNWut/lib/pam_sunray_amgh.so.1 clearuser
auth     include        common-auth
session required /opt/SUNWkio/lib/pam_kiosk.so log=user
session  required       pam_loginuid.so
session  include        common-session
session  required       pam_resmgr.so
account sufficient /opt/SUNWkio/lib/pam_kiosk.so log=user
account  include        common-account
-=-=-


This looks all sane. Unless you actually use any of the relevant SRSS functionality (AMGH or Kiosk Mode) you could remove all of the SRSS-generated PAM entries (all the entries that point to /opt/SUNW*) without doing harm. But in that case these modules should all be passive, so I would be surprise, if that improved your situation.

Is there any chance that you could explain to me the behavior of that particular module a bit?


See above. The behavior if a session is configured for Kiosk Mode is described in the man page in quite some detail. What is lacking there?

Also, thanks to the greater pouring over documentation that I did today compared to previously, I figured out where I think the PAM modules are dumping their logging information on these systems. The only authentication error I saw from my testing the other day was the following:

-=-=-
/var/log/warn:May 26 13:22:15 srss-bismarck-gamma gdm[4175]: WARNING: Couldn't authenticate user /var/log/warn:May 26 13:22:28 srss-bismarck-gamma gdm[4175]: WARNING: Couldn't authenticate user /var/log/warn:May 26 13:29:16 srss-bismarck-gamma gdm[13323]: pam_krb5[13323]: error resolving user name 'dgetsman' to uid/gid pair /var/log/warn:May 26 13:29:19 srss-bismarck-gamma gdm[13323]: WARNING: Couldn't authenticate user /var/log/warn:May 26 13:36:45 srss-bismarck-gamma gdm[6638]: WARNING: Couldn't authenticate user /var/log/warn:May 26 13:37:17 srss-bismarck-gamma gdm[6846]: WARNING: Couldn't authenticate user /var/log/warn:May 26 13:41:12 srss-bismarck-gamma gdm[6846]: WARNING: Couldn't authenticate user /var/log/warn:May 26 13:42:33 srss-bismarck-gamma gdm[6846]: WARNING: Couldn't authenticate user

These warnings are rather unspecific. It is hard to tell where things go wrong. It might be useful, if you have a syslog target which captures more messages (at notice, info or even debug level). It may also be helpful if you add 'debug' options to some of modules in /etc/pam.d/common-auth, which is where the real authentiaction behavior is defined.

/var/log/warn:May 26 13:52:21 srss-bismarck-gamma sshd[8323]: fatal: login_init_entry: Cannot find user "dgetsman" /var/log/warn:May 26 13:52:21 srss-bismarck-gamma sshd[8323]: fatal: login_init_entry: Cannot find user "dgetsman" /var/log/warn:May 26 13:52:21 srss-bismarck-gamma sshd[8328]: pam_krb5[8328]: error resolving user name 'dgetsman' to uid/gid pair /var/log/warn:May 26 13:52:21 srss-bismarck-gamma sshd[8328]: pam_krb5[8328]: no user info for dgetsman (shouldn't happen) /var/log/warn:May 26 13:52:21 srss-bismarck-gamma sshd[8328]: pam_krb5[8328]: error resolving user name 'dgetsman' to uid/gid pair /var/log/warn:May 26 13:52:21 srss-bismarck-gamma sshd[8328]: pam_krb5[8328]: no user info for dgetsman (shouldn't happen)
-=-=-


If 'dgetsman' is supposed to be a valid login name, then your name service configuration (/etc/nsswitch.conf) is probably wrong.

You said before that LDAP login works for ssh or console login. Is that really using the same login name and password which fails under gdm?


Now I can't be sure as I'm still learning about the SRSS PAM modules and admittedly still have a LONG way to go, but I'd imagine that if the problem encompassed more than just kerberos, I'd see other modules complaining in the log.


Yes. If you wish you can try removing all SRSS modukes from the gdm file. If the failures still occur, you know for sure that the problem is a in one of the modules listed in the common-{auth,account,session} files.

HTH

- Jörg
_______________________________________________
SunRay-Users mailing list
[email protected]
http://www.filibeto.org/mailman/listinfo/sunray-users

Reply via email to