CCL: Calculations on a DRBL cluster

[Alexander Martins Silva <alex.msilva..uol.com.br>]

 Sent to CCL by: Alexander Martins Silva [alex.msilva#%#uol.com.br]
         Hi CCLers,
              I have installed a small linux cluster with the DRBL
 program (drbl.sourceforge.net).  I can execute small parallel jobs with
 mpi and even short Gamess calculations. However, the parallel jobs
 stops
 without error message when these jobs require some hours of
 calculation.  Even when I start serial jobs on each node the error is
 the same.  The same job can stop at differents points.  What's
 happening? How can I solve this?  I'm using a Mandrake 10/Athlon XP
 2600
 server machine with Raid1, 10 P4 clients and a PLanet
 24port/gigabit/switch.
             Any suggestion or advise is welcome.
                        Thanks,
                     Alexander.