19 Sep 2024
Dear HKUST SuperPOD Users,
I am writing to provide an update regarding the HKUST SuperPOD, following my email dated 13-Aug-2024.
As the new High Performance Computing Infrastructural Center (HPCIC) approaches completion, we plan to physically relocate the HKUST SuperPOD DGX machines (currently being hosted in an external data center) back to the HPCIC on campus. Details of the relocation exercise are as follows:
- Access to HKUST SuperPOD Scratch Area: HKUST SuperPOD users are reminded to back up their important data in the scratch area before 30-Sep-2024, as the corresponding storage server will be physically relocated.
- Access to HKUST SuperPOD Home Directory: During the relocation period, access to user Home directory will remain normal as the corresponding storage system is hosted on campus.
- Job Submission: New job submissions will be disabled on 1-Oct-2024, at 00:00:00. All executing or queued jobs will be terminated or cleared at that time.
- Relocation Timeline: The physical relocation exercise is expected to take approximately six weeks, from 1-Oct-2024, to 11-Nov-2024. During this period, the availability of resources in the HKUST SuperPOD cluster will be as follows:
- Between 1-Oct-2024 to 28-Oct-2024: No service available.
- By 29-Oct-2024: Access to the storage system will resume, with approximately 30% of GPU resources becoming available.
- By 5-Nov-2024: Up to 60% of GPU resources will resume service.
- By 11-Nov-2024: HKUST SuperPOD services will resume at 100%.
We understand that the downtime of the HKUST SuperPOD service may impact researchers' planned schedules. The ITSC is committed to continue collaborating with all relevant parties involved in the relocation exercise and will endeavor to minimize and shorten the service interruption to the greatest extent possible.
Thank you for your understanding and cooperation.
Regards,
Alan Wong
Head of Teaching Technologies
Information Technology Services Center
Hong Kong University of Science & Technology
Direct Line: (852) 2358-6246