Hi

I am having problems running "real" jobs on my perceus/xcpu cluster.
This is true both without moab/torque and within it.  Especially
within moab/torque.

The xcpu installation is sxcpu, as that is what perceus uses.  When I
try to run a piece of code on one of the slave nodes, it kind of works
if I invoke xrx with the -c option, so that it changes directory to
wherever I want it to run.  For example, if I am in the directory
/home/danny, AND the /home filesystem is available to the nodes (e.g.
via NFS), then if I issue the command

xrx -c /home/danny myjob

it runs properly, performing all the required I/O.

However, when I try to run the code from within moab/torque, I can't
get it to do this.  I realize that it may simply be due to my
ignorance of how to create a proper job script, so if anybody could
provide some pointers it would be truly appreciated.

Apart from this, the issue of moab/torque not running jobs properly
distributed on all the cluster nodes was not resolved (the thread with
Hugh just stopped...).  Any hopes for getting this resolved?

Thanks,
Daniel

Reply via email to