KAUST Supercomputing Laboratory Newsletter 8th February 2017

In this newsletter:

  • Purge policies on Lustre project filesystem
  • Shaheen Queuing Policy
  • HPC Saudi Arabia 2017: 13th - 15th March 2017
  • Shaheen Operating on Reduced Capacity
  • RCAC Meeting
  • Follow us on Twitter
  • Previous Announcements

Purge policies on Lustre project filesystem

Following a prolonged period in which, for technical reasons, we maintained a somewhat relaxed implementation of Shaheen policy for purging of older files from the /project filesystem, the policy is now in full operation. The policy is described in detail in our User Guide at https://www.hpc.kaust.edu.sa/content/lustre. In terms of the user experience, files which are listed in a directory may no longer exist on disk and exist only on tape, so it may take several seconds or even minutes, depending on the size of the file, to list the contents of a file. From the User Guide:

To query the HSM status of a particular file in /project, run the following command:

$ lfs hsm_state /project/<projectname>/<filename>

A response of this type:

/project/<projectname>/<filename>: exists archived, archive_id:1

indicates that a copy of the file has been made to tape, whereas:

/project/<projectname>/<filename>: released exists archived, archive_id:1

indicates that the file has been purged from disk and only the stub file remains. A simple read action will cause the file to be recalled from tape back to disk.

The fastest and most efficient to restore multiple files from tape is to use:

$ lfs hsm_restore FILE1 [FILE2 ...]

Shaheen Queuing Policy

In order to provide more functionality on Shaheen and at the request of representations made to the RCAC, some changes have been made to the queuing policy:

The are now 16 nodes dedicated to a ‘debug’ queue, jobs are limited to a maximum of 2 nodes with a wall clock limit of 30 minutes.
The 72 hours queue now has up to 512 nodes available and the 1 node job limit will be removed from 1st Feb 2017.

Use of the 72 hours queue will be authorised by the RCAC, please note that any current project requiring ongoing access after the 31st January 2017 will be required to submit an application. Please could we request that the PI for a project submits a brief justification along with details of the code they are running to help@hpc.kaust.edu.sa

HPC Saudi Arabia 2017: 13th - 15th March 2017

KAUST is proud to be hosting High Performance Computing Saudi Arabia (HPC Saudi), this is the premier regional event in the field, where participants can meet each other, share ideas and experiences, and discuss cooperation and collaboration. This is the seventh HPC Saudi event, and this year it will focus on coordinated efforts for the advancement of the HPC ecosystem in the Kingdom. Please see www.hpcsaudi.org for more information and registration. Please note that registration closes on the 1st March 2017.

Shaheen Operating on Reduced Capacity

Due to maintenance on the UPS system to replace the batteries, Shaheen will be operating on sligthly reduced capacity from 13th Feb 2017 until 12th March 2017. Up to 515 nodes will be offline to reduce power load, this will mean that the maximum size job that can be run during this period is 5643 nodes.

RCAC Meeting

The project submission deadline for the next RCAC meeting is 28th February 2017. Please note that the RCAC meetings are held once per month. Projects received on or before the submission deadline will be included in the agenda for the next RCAC meeting. The next meeting is scheduled for March 2017. The detailed procedure and the updated forms are available here:


Any new project allocation will be considered only using the new project proposal template. This should include up-to-date PI and CoPI C.V., list of publications  and optional suggestions and/or exclusions of reviewers.

Follow us on Twitter

Follow all the latest news on HPC within the Supercomputing Lab and at KAUST, on Twitter @KAUST_HPC.

Previous Announcements