A user typo lead to a segfault on pbs_server: #PBS -l walltime=24:00:00,nodes=1,ppn=4,mem=30gb should have been: #PBS -l walltime=24:00:00,nodes=1:ppn=4,mem=30gb pbs_server crashes consistently when jobs are submitted with this resource request.