May HiPerGator Maintenance
Network Switch Upgrades
UFIT will apply software updates to HiPerGator’s main network switches on May 5 starting at 8:00 am. These updates are expected to take one hour and be mostly transparent to users, though there may be some momentary slowness and unavailability.
Storage Maintenance
HiPerGator Storage Maintenance and Open OnDemand outage, May 6,7 and 8
UFIT Research Computing staff will apply software updates to the storage systems starting on May 6 (schedule below). These updates are expected to take four hours each and be mostly transparent to running jobs, though there may be some momentary slowness. For interactive use, moments of slowness and unavailability are expected.
As Open OnDemand is more sensitive to these updates, it will be down during the maintenance on each storage system. Users will not be able to connect to Open OnDemand during the maintenance. Other HiPerGator services, such as JHub and SSH access, will remain in production.
Storage update schedule:
- Red storage will be updated on Tuesday, May 6, starting at 7:00 am.
- Blue storage will be updated on Wednesday, May 7, starting at 7:00 am.
- Orange storage will be updated on Thursday, May 8, starting at 7:00 am.
Operating System Maintenance
HiPerGator OS Upgrade: RHEL 8 to RHEL 9
HiPerGator is beginning a rolling upgrade from Red Hat Enterprise Linux version 8 (RHEL-8) to version 9 (RHEL-9). The upgraded software environment will include updates to important components, including NVIDIA GPU drivers, CUDA, communication libraries, and compilers.
Many programs/workflows will run in the updated software environment without any modification, but some will require recompilation or adaptation.
Please check our website for additional information and updates.
Timeline
- Monday, May 19, 2025:
- A subset of HiPerGator compute resources will go online with the updated software environment.
- A new login node pool will be introduced as hpg-el9.rc.ufl.edu that runs the updated software environment.
- If you need to rebuild your application, you can do it here.
- May 19 - Aug. 1: Additional compute resources will be upgraded at a steady rate until they are all upgraded.
- Once more than half of the compute resources are upgraded, the new login nodes will become the default as hpg.rc.ufl.edu. Until all RHEL8 nodes are converted, some RHEL8 login hosts will be at hpg-el8.rc.ufl.edu.
We recommend moving to RHEL-9 at your earliest convenience.
Choosing RHEL-8 or RHEL-9
Using SLURM directives, it is possible to submit jobs targeting: only RHEL-8, only RHEL-9, or by default, whichever is available first (i.e., operating system doesn’t matter, take the first available)
To target a specific OS in your job scripts, use the SBATCH options: --constraint=el8 or --constraint=el9. If neither is specified, the scheduler will not consider the operating system in placing your jobs. These constraints can be entered in the “Additional SLURM Options” field of Open OnDemand forms.
Questions or concerns?
If you have concerns about access to RHEL-8 systems on HiPerGator after Aug. 1, 2025, please contact us as soon as possible to discuss.