ltsp-cluster-accountmanager not functioning properly

[datube]

Package: Ubuntu Karmic - ltsp-cluster-accountmanager - 2.0.3-0ubuntu1

We have 3 application servers which have the same issue described below. I will give the output of several commands to describe the issue.

1) Logged on users:
~# w
 17:49:40 up 24 days, 1:41, 9 users, load average: 0,11, 0,14, 0,32
USER TTY FROM LOGIN@ IDLE JCPU PCPU WHAT
userA pts/1 xxx.yy.z.67:7.0 17:48 0.00s 0.31s 0.09s w
userB pts/2 xxx.yy.z.69 16:11 1:38m 0.03s 0.00s /bin/sh -
userC :1002 xxx.yy.z.37 12Apr10 ?xdm? 17:48m 0.01s /bin/bash /usr/
userD pts/6 xxx.yy.z.141 09:59 7:49m 0.01s 0.00s /bin/sh -
userE pts/8 xxx.yy.z.34 16:15 1:34m 0.03s 0.01s /bin/sh -
userA pts/18 xxx.yy.z.67 11:46 6:02m 0.03s 0.00s /bin/sh -
userF pts/11 xxx.yy.z.162 11:39 6:09m 0.03s 0.01s /bin/sh -
userG pts/23 xxx.yy.z.157 13:53 3:55m 0.05s 0.02s /bin/sh -

2) Unique users derived from running processes:
~# ps auxwww | awk '{ print $1 }' | sort | uniq | wc -l
120

3) Syslog messages:
~# tail /var/log/syslog
Apr 22 17:52:48 app-server ltsp-cluster-accountmanager: Killing user "user?"
Apr 22 17:52:49 app-server ltsp-cluster-accountmanager: Killing user "user?"
Apr 22 17:52:49 app-server ltsp-cluster-accountmanager: Killing user "user?"
Apr 22 17:52:49 app-server ltsp-cluster-accountmanager: Killing user "user?"
Apr 22 17:52:49 app-server ltsp-cluster-accountmanager: Killing user "user?"
Apr 22 17:52:49 app-server ltsp-cluster-accountmanager: Killing user "user?"
Apr 22 17:52:49 app-server ltsp-cluster-accountmanager: Killing user "user?"
Apr 22 17:52:49 app-server ltsp-cluster-accountmanager: Killing user "user?"
Apr 22 17:52:49 app-server ltsp-cluster-accountmanager: Killing user "user?"
Apr 22 17:52:49 app-server ltsp-cluster-accountmanager: Killing user "user?"

Note:
1) "user?" are not users seen in the list which is produced with the 'w' command and this does not account for one uid
2) these messages are filling up the syslog and logged throughout the day

4) There are also a lot of "<defunct>" processes:
~# ps auxww | grep defunct | wc -l
97

I do not know why there is a big of difference is the number of users actually logged on and the number of running processes at the same time. I also cannot grasp why there are so many "<defunct>" processes, they do not seem to resolve after a period of time.

If I need to supply any additional information I will provide them.
Thanks in advance.