GRASS can be very resource intensive so it should not be run directly on SciClone's login servers. Instead, GRASS jobs should be submitted to an appropriate compute node on the cluster via the job ...