HPC News/Updates

[Print-friendly version]

General Information

The first Wednesday of each month is identified as the time in which system maintenance is performed when required. Depending on the work to be carried out, some or all of the HPC services may be unavailable.

Update ~ 03/06/08

MAJOR IMPACT:

vega: The OS is being upgraded. All ssh/sftp sessions on Vega and running jobs in the batch system will be stopped.

MINOR IMPACT:

hpc-fs: OS upgrades and configuration changes will be made. Changes will be made to the inactive server, a forced failover will be triggered. Changes will then be made to the other server. A second failover will then be triggered to test the changes on both systems. During the failover period (approx 5-10minutes) access to HPC-FS (eg mapped network drives) will be lost - please unsure that all PC's accessing HPC-FS do not have open files.

hadar: Cluster management software is being upgraded which will require a reboot of the system. While the reboot is in progress access to all license managers will be disrupted (including: MATLAB, Abaqus, PATRAN, Ansys). The disruption should be for a minimum period, approx 5 - 10 minutes. During this time no new batch jobs on Lyra will be accepted, and 'qstat' commands will fail but running jobs will not be affected.

If you have any queries about this maintenance and/or its impact please contact HPC (qut.itshpc@qut.edu.au) as soon as possible.   

[Mark Barry]

Update ~ 29/02/08

There is maintenance required for the HPC-FS system (the HPC file servers, providing storage to Vega, Lyra and network mapped drives on PC's/Mac's). System upgrades/modifications are required to correct known issues. The disk space of the /work partition will also be increased.

IMPACT:

All ssh/sftp sessions on Vega and Lyra will be terminated, running jobs in the batch system will be stopped. Vega, Lyra and mapped network drive connections to HPC-FS will be unavailable for the duration of the maintenance.

If you have any queries about this maintenance and/or its impact please contact HPC (qut.itshpc@qut.edu.au) as soon as possible.   

[Mark Barry]

Update ~ 21/02/08

"It is with much sadness that I need to inform you that after more than 6.5 years of very faithful service to users at QUT and beyond, Sirius (in the SGI Origin 3000 incarnation) is about to come to an end.

Generations of PhD, Masters and Undergraduate students as well as research staff have made use of Sirius, from the initial 60 processor version back in June 2001 though to the final 128 processor configuration of the past couple of years. Over 650 users have had an account at one time or another on Sirius.

Sirius' last day will be Wednesday 27th February 2008. The system decommissioning process will then be undertaken including the final system backups. Sirius will be gone but long remembered.

Ah, I remember the day that the process load hit 230, the load graphs were off the scale, but on she went ... What about the job set that ran for over 3 years, now that was what I call a JOB! ...memories..."   

[Mark Barry]

Update ~ 02/05/07

QUT has recently purchased new high-performance computing infrastructure from SGI which is due to be operational by the second semester, 2007. The new installation includes two compute systems and a large data store. The first machine is an SGI Altix 4700 SMP supercomputer (96 64bit Itanium 2 cores and 198GB RAM), and the second machine is an SGI Altix XE Cluster (containing 14 compute nodes, with 112 64bit Intel Xeon cores).

Both systems will share a heirarchical data storage system which will provide over 100 TeraBytes of storage to QUT researchers. The compute capacity of the new system is approximately ten times that of the previous HPC machine.

For more information about these new systems, please refer to Hardware Resources. For additional information, contact the HPC and Research Support Group.