There are up to three quotas you have to take into account when running jobs on the cluster.

  1. Your Linux account, from which you submit your jobs, has a quota. The default quota for this area is 2GB, although some people have requested increases in this quota so your individual quota may be different. You can read more about this quota here.
  2. Any output that your job prints to either standard output (STDOUT) or standard error (STDERR) gets stored on the individual cluster node where your job runs. Each person gets 4GB of quota per cluster node. If your job will print a lot of information or you have a lot of jobs running simultaneously, you may need to redirect the output to a file. You can do this with something like:
    ./program > ${PBS_JOBID}-stdout.txt 2> ${PBS_JOBID}-stderr.txt

    This redirects the standard output to *-stdout.tx and standard error to *-stderr.txt (where * is the job ID that’s automatically generated). By including the job ID, this ensures that the file names are unique for every job, but using the job ID in the name of the files is not required.

  3. Some cluster nodes have scratch space available. Please refer to this page for information about scratch space quota.