Cluster jobs which do not use files in /scratch/cluster should be
unaffected. Condor jobs attempting to use files in the cluster scratch
space may be put on hold automatically by Condor. Once the downtime is
over, the condor_release command should get them running again.