question about pam_tally and the faillog

Bill Tangren bjt at usno.navy.mil
Tue Mar 18 13:53:15 UTC 2008


I am running a number of RHEL ES 4.5 systems, fully updated.

The problem I am having is baffling me. I am using pam_tally so that three
consecutive unsuccessful logins will lock out the user, until an hourly
cron script unlocks the account. It has worked fine for a number of years.

The problem is this:

If I log in as user bjt, and I SUCCESSFULLY su - to user bdna_user, the
faillog records this as a failed login attempt, even though nothing
untoward appears in the logs, that I can find. In essence, I can log in as
bjt, su - to bdna_user three times, and cause bdna_user's account to be
locked out, so that if bdna_user attempts to log in, their access will be
denied. Below is the logs where I do just that.

Mar 18 09:40:35 doggett sshd(pam_unix)[14176]: session opened for user bjt
by (uid=0)
Mar 18 09:40:49 doggett su(pam_unix)[14201]: session opened for user
bdna_user by bjt(uid=500)
Mar 18 09:41:27 doggett su(pam_unix)[14201]: session closed for user
bdna_user
Mar 18 09:41:36 doggett su(pam_unix)[14226]: session opened for user
bdna_user by bjt(uid=500)
Mar 18 09:41:38 doggett su(pam_unix)[14226]: session closed for user
bdna_user
Mar 18 09:41:44 doggett su(pam_unix)[14250]: session opened for user
bdna_user by bjt(uid=500)
Mar 18 09:41:48 doggett su(pam_unix)[14250]: session closed for user
bdna_user
Mar 18 09:42:03 doggett sshd(pam_unix)[14176]: session closed for user bjt
Mar 18 09:42:07 doggett sshd(pam_unix)[14150]: session closed for user
bdna_user
Mar 18 09:42:23 doggett pam_tally[14278]: user bdna_user (1029) tally 4,
deny 3

[root at doggett ~]# pam_tally
User bdna_user  (1029)  has 4
[root at doggett ~]#

This is what is in /etc/pam.d/system-auth:

auth        required      /lib/security/$ISA/pam_tally.so onerr=succeed
no_magic_root
account     required      /lib/security/$ISA/pam_tally.so deny=3
no_magic_root reset

NOTE: I have systems where "onerr=fail" is set, and it makes no difference.

My question is, why is it doing this?

-- 
Bill Tangren
U.S. Naval Observatory

Si hoc legere scis nimium eruditionis habes




More information about the redhat-list mailing list