Dears.
There are the necessary for  help and ideas.
Very strange situation exists in work gLiteCE server.
System of Server - Linux 3.06 CERN with openssh-server-4.3p2-2.cern, 
glite-yaim-3.0.0-34,
glite-CE-2.4.23-1, torque-server-1.0.1p6-13.SL30X.st applications.
Installation by Yaim tools passed successfully with one error -
...
gLite Security Utilities configuration successfully completed

running /opt/globus/setup/globus/setup-globus-gram-reporter-fork..[ Changing 
to /opt/globus/setup/globus ]
Setting up fork gram reporter in MDS
-----------------------------------------
Error  fork GRAM reporter entry jobmanager-fork. Aborting!
loading cache ./config.cache
...

In addition, the commands were executed

/opt/glite/yaim/scripts/run_function site-info.def config_bdii

and

/opt/glite/yaim/scripts/run_function site-info.def config_mkgridmap

Last command run because the file /etc/grid-security/grid-mapfile
was very small.

Jobs run from UI is executed orderly.
Now begin the miracle.
Job continues to run on CE, but on UI comes message -

*************************************************************
BOOKKEEPING INFORMATION:

Status info for the Job : https://wmslb.itep.ru:9000/O6BDOJSZi4EXkrE1VLTeHg
Current Status:     Done (Failed)
Exit code:          0
Status Reason:      Got a job held event, reason: Unspecified gridmanager 
error
Destination:        testbed01.itep.ru:2119/blah-pbs-alice
Submitted:          Sat Dec 23 17:45:08 2006 MSK
*************************************************************

On CE is started one more copy of the JOB as a result -

[lublev@uiitep JOBS]$ glite-job-logging-info 
https://wmslb.itep.ru:9000/O6BDOJSZi4EXkrE1VLTeHg | grep Run
Event: Running
Event: ReallyRunning
Event: Running
Event: ReallyRunning

Repetition of the new job start occurs several times whereupon on UI we get 
message

*************************************************************
BOOKKEEPING INFORMATION:

Status info for the Job : https://wmslb.itep.ru:9000/lKhDzZLsPZE9SXN2e01vBA
Current Status:     Aborted
Status Reason:      hit job shallow retry count (0)
Destination:        testbed01.itep.ru:2119/blah-pbs-alice
Submitted:          Sat Dec 23 12:05:35 2006 MSK
*************************************************************

But several copies of the job on CE continue be considered and end orderly.
Best regards. Yevgeniy.