Pleiades OSG User Policies

Cluster Access

  1. Authenticated Access

    Pleiades users must already have an Grid certificate and a Globus Client installation to submit jobs on the cluster.

    To access Pleiades resources please contact your local organization on how to become a member its Virtual Organization.

    All user accounts share the same group id grid3.

Resource Allocation Policy

Compute Resources

  1. Fairshare

    For the purpose of fairshare balancing of compute resources over a one week averaging period, cycles are allocated by user affiliation in the following proportion:

      PSURG

    LSC

    iVDGL
    (OSG)

    IDLE

    Fairshare Allocation

    50

    40

    10

    0

    OSG users are directed to the 'ivdgl' queue by default. OSG users do not have authority to use the other higher priority queues. Users could optionally choose to submit to the 'idle' queue if they so desire.

  2. Job Time Limit

    Each queue has an allocated walltime for jobs. The walltime is active from the moment a job begins to run on a node, not while queued. Jobs that need to run longer than walltime need to be checkpointed.

      PSURG LSC iVDGL
    (OSG)
    IDLE
    Job WallTime (hours) 24 24 24 12
  3. Processors Per Job

    Users are permitted to submit multiple jobs but each job has a limit on how many processors it can use (see table below). Users are requested to run one process thread per processor at maximum and limit memory usage to 1GB per processor.

      PSURG

    LSC

    iVDGL
    (OSG)

    IDLE

    Max. CPUs/Job

    64

    16

    16

    16

Storage Resources

  1. Disk Quota

    The disk space for all OSG resources is 300GB. This includes $APP, $DATA, and home directories of all OSG users.

    $APP, $DATA, and home directories are mounted across the cluster.

  2. Scratch Space

    Local scratch space for jobs is /tmp and follows standard system cleanup policy.

    Note: $APP, $DATA, and home directories are NOT backedup or archived. They are, however, on RAID5 mounts.