Hawk Phase 2 (Updated)

TL;DR

  • Hawk service at-risk for the weeks of the 9th and 16th December
  • Hawk compute unavailable on one day to be re-arranged.
  • Users should backup important scratch data in advance.

Hawk will shortly be undergoing a Phase 2 Expansion which will provide Hawk with new compute capabilities via AMD Rome CPUs, enhanced GPU capabilities with new nodes containing Nvidia V100 GPUs and will also grow the scratch filesystem.

Primary installation will take place during the week of 9th December. The entire hawk service should be considered at-risk as hardware installation will be taking place in the data centre.

Additionally, the lustre expansion will require a compute outage in order to quiesce the file system. This is currently planned for the 12th December. Login nodes and job submission will remain available, but compute will not.

IMPORTANT:
The expansion of the lustre /scratch file system presents a small risk of data loss. As such, please BACKUP important or difficult-to-reproduce data from the scratch file system before the installation begins.

UPDATE @ 11/12/2019
During installation and configuration, due diligence identified two factors requiring additional resolution. As such, we are postponing the compute outage due Thursday 12th and will advise the new date in due course.

Job dispatch on Hawk has been fully resumed.

Please don’t hesitate to Contact Support if you need further information or any assistance with backing up any data.

Leave a Reply

Your email address will not be published. Required fields are marked *