Spring Semester HPC Downtime (March 27 – April 3)

From 8:00 a.m. on Monday, March 27, until 9:00 a.m. on Monday, April 3, the entire HPC cluster—including all head nodes, file systems, and compute nodes—will be unavailable for use during our annual spring upgrade.

During this downtime, we will:

  • Upgrade the operating system on all compute nodes and head nodes. We will be grading the operating system, applying security and operating system patches on file servers, running file system checks on the majority of the HPC file systems, and installing firmware patches on various storage devices.
  • Rebuild /staging. We will be rebuilding the /staging file system as part of our standard downtime procedure. The rebuild will remove all data from the file system in accordance with our policy on the /staging and /scratch file systems, posted at hpcc.usc.edu/support/infrastructure/temporary-disk-space. Please remember that this file system is temporary working space, so no data on it is backed up. Please migrate any data residing on /staging that you would like to keep at least one week prior to the downtime to ensure the migration completes before the downtime.
  • Make default symbolic link (symlink) changes for software installed under /usr/usc to newer versions. We anticipate making default symlink changes for software installed under /usr/usc to newer versions. Older software versions will still be available on HPC should you prefer to use those. We will provide a list of all updated default symlinks as we get closer to the downtime.
  • Add Mellanox Infiniband infrastructure to address Infiniband cluster growth.

As always, we will release components before the scheduled uptime, if we are able to do so. However, due to the work on the file systems, we will be unable to determine ahead of time if we will be able to release components early.

If you have any questions or concerns about this downtime, please contact us at hpc@usc.edu.