Setting up staging cluster with:

• Two sets of login node and compute nodes
  • login001 and c000[1-2] on gpfs4
  • login002 and c000[3-4] on gpfs5
• Have node feature, gpfs4, gpfs5 setup
• Have job_submit plugin setup so when user submit a job, it selects correct set of compute nodes for the user, based on their group
• Have apache proxy setup to route users to correct OOD instance, based on their group
• Have sshpiper setup to route users to correct login node, based on their group
  • key based
  • password based
• Have account app setup to add the user to default group, e.g. gpfs4
• Have account migration script to move users between two groups, gpfs4 and gpfs5