February 21, 2025

Omnibus Maintenance, March 26, 2025

By Rebecca Hartman-Baker

Posted in Outages
Perlmutter Mural Phase II lighted

On March 26, 2025, coinciding with the Perlmutter monthly maintenance, NERSC will perform a number of upgrades that will impact user access to NERSC.

First, NERSC will upgrade the network that connects the NERSC global file systems to other resources. Upgrading to the new network, which will result in improved bandwidth and network speed capabilities, requires a full downtime of the global file systems, including the Community File System (CFS), Global Common, Global Homes, and DNA from 6:00 am to 3:00 pm. Users should plan for these file systems to be unavailable for the majority of the day.

In addition, the broader NERSC network, connecting systems within NERSC to the router connecting with ESnet, will be upgraded between 8:00 a.m. and 4:00 p.m. This will result in periodic, 30-minute connectivity losses throughout the maintenance window. Systems that may otherwise be available will not be reachable during these equipment changeovers. If, therefore, you encounter a system other than Perlmutter that is not reachable during this period, we recommend trying again after 30 minutes.

Finally, we will take advantage of this downtime to introduce new equipment and disk storage to CFS. This will provide a temporary boost to CFS capacity, until old equipment is retired in the upcoming months.

A summary of availability and expected impacts throughout the day for most NERSC systems and services follows (all times Pacific time zone):

  • Perlmutter: Unavailable 6:00 a.m. - 6:00 p.m. Includes Perlmutter scratch file system.
  • Global File Systems (CFS, Global Common, Global Homes, DNA): Unavailable 6:00 a.m. - 3:00 p.m. Any systems depending on these file systems will be unavailable, including Data Transfer Nodes, and any Spin workloads depending on CFS.
  • HPSS: Available via direct connections only, until other systems depending on global file systems become available; periodic outages due to network maintenance.
  • DTNs: Unavailable 6:00 a.m. - 3:00 p.m., coinciding with global file systems unavailability. Return to service degraded, with no access to Perlmutter scratch (until Perlmutter returns to service).
  • Jupyter: Unavailable 6:00 a.m. - 6:00 p.m. Dependent upon Perlmutter for operation; logins may be possible but with error messages displayed due to unavailability of Perlmutter.
  • Globus: Unavailable 6:00 a.m. - 3:00 p.m., coinciding with global file systems unavailability.
  • Spin: Workloads depending on global file systems unavailable 6:00 a.m. - 3:00 p.m.; periodic outages due to network maintenance.
  • NoMachine: periodic outages due to network maintenance.
  • Science Databases: periodic outages due to network maintenance.
  • Registry: periodic outages due to network maintenance.
  • Iris: Available all day; not impacted by network upgrade.
  • NERSC websites (help desk, docs, www): Available all day; externally hosted and not impacted.
  • NERSC Network: Intermittent 30-minute outages from 8:00 a.m. - 4:00 p.m., impact across all systems at different times throughout the time window.