Skip to content

Our policies

  • Abide by the Policy on Responsible Use of NYU Computers and Data
  • Moderate risk research data is permitted on the cluster
    • Prohibited data: the cluster is not equipped to handle High-Risk data, such as Personally Identifiable Information (PII), Protected Health Information (PHI), or financial data. Researchers needing high-risk data storage and processing should get in touch with the Secure Research Data Environment team
  • Don't set incorrect resource requests: the system will assume that you need the requested resources and will not put them towards another user's workload. Try to set a request as close as possible to your average need.
  • Don't rely on containers running forever: though we will try to keep the cluster stable, individual nodes will go down for maintenance periodically. At that time your pods will be "evicted". Using Deployments or Jobs instead of Pods is recommended, in that case new pods will be automatically started to replace the lost ones.

Backups

Persistent volumes have daily backups which are kept for 30 days. Contact us if you need to recover data.

Communication and maintenance plan

Types of maintenance

  • Planned maintenance without user impact:
    • Some maintenance is necessary for the health and performance of the network but does not impact users. It will be performed during normal business hours and will not be subject to a prior notification.
  • Planned maintenance with potential user impact:
    • Some maintenance activities can temporarily affect services to our users. Planned maintenance will typically be carried out on the first worked Monday of the month, following the HPC maintenance policy.
  • Emergency maintenance or failure:
    • Urgent maintenance or equipment failure will sometimes happen outside of planned maintenance windows.

Notification channels

  • For Kubernetes cluster users: emails to hsrn-kubernetes-users@nyu.edu
    • Every user that is given API access to Kubernetes is added to this list.
  • For Ceph cluster users: emails to hsrn-ceph-users@nyu.edu
    • Users will be added to this list when they are given access to Ceph, either directly or via the HPC data transfer nodes (DTN).

Notification timing

  • Planned maintenance without user impact:
    • We will not notify users if there is no impact in availability and no change in the service.
  • Planned maintenance with potential user impact:
    • Notification will be sent by email at least seven days before the start of the maintenance.
  • Emergency maintenance or failure:
    • We will notify users as soon as possible via email.

Support channel

Users can reach us by email at hsrn-support@nyu.edu.