startGsProcessor script

6.2	startGsProcessor script

All the processing launch scripts with the ‘run’ prefix call the general purpose ‘startGsProcessor’ script which is responsible for calling the gsRunProcessor executable. Specifically, startGsProcessor is a wrapper script which first creates an output data directory with a regions subdirectory, and either reads an existing dataRunParams.xml input XML file or creates one from the imageLog.parse and image files, and finally calls the appropriate launching commands.

startGsProcessor --pipe=XMLFILE SOURCE_DIRECTORY

Command	Description
startGsProcessor	A wrapper script for the gsRunProcessor. It creates the output directories, prepares the Run for processing and calls the appropriate launching commands.

Argument	Description
--pipe=XMLFILE	Specifies a pipeline processing script file. (Specifying the .xml extension is optional.)
SOURCE_DIRECTORY	Data directory containing files to be processed.

An important environmental variable that affects the operation of the startGsProcessor script is the ‘GS_LAUNCH_MODE’ variable. This value can be set in each user’s environment by adding an environmental variable definition line to the .bash_profile file in the user’s home directory. Site administrators may have set these values site-wide in /etc/profile or by adding a file in /etc/profile.d.

GS_LAUNCH_MODE

Value	Description
SINGLE	Starts a single copy of the gsRunProcessor. (Default)
MULTI	Starts multiple copies of the gsRunProcessor, equal to the number of processors in the current workstation. (Non-cluster)
MPI	Uses ‘mpiexec’ for launching jobs on a compute cluster.
GSRPM	Starts the job using the gsRunProcessorManager. Will submit jobs to the same job queue as users who use gsRunBrowser to submit processing jobs. Recommended as a multi-user job queuing system.