Running Jobs

Running L2 FP Through Populator Produced Scripts

The Populator supports creating scripts customized for several cluster management systems. The target cluster is selected using -t argument and one of the following strings:

pbs_pro PBS Professional
torque Torque
pleiades Pleiades Supercomputer

The default cluster when the -t argument is not supplied is PBS Professional.

When the Populator is run with the -b option specifying a path to a L2 FP binary the following scripts may be created: Runs an individual sounding, created for all target clusters Launches jobs on PBS Profession and Torque clusters Launches jobs on the Pleiades Supercomputer machines

You normally would not run directly, although it can be useful when debugging a single sounding. This script will take an argument specifying the index of which sounding to process. To run multiple instances of simultaneously on one computer you can use GNU Parallels. For instance to run the first four soundings you would run it as follows:

$ parallel ./ ::: 0 1 2 3

The script can only on be run from the head machine of a PBS Professional or Torque cluster, the machine where qsub exists. Any arguments specified to will be passed directly to qsub. On the SCF system the one required argument missing from the qsub call done in is the argument specifying which queue to use. This argument is specified by the -q option and hence the normal use case for running is as follows:

$ -q long

The above command will launch all soundings into the long queue. Should you wish to use a different queue just swap out “long” for the desired queue name. Any additional options for qsub can also be specified in the call to the script.

Additional Cluster Documentation

For full documentation on TORQUE job submission, please consult this document:

For additional documentation on the TORQUE job submission command qsub, please consult this document:

PBS Professional documentation can be found here: