GT Home : : Campus Maps : : GT Directory

Archive for January, 2017

PACE clusters ready for research

Posted by on Friday, 20 January, 2017

Our January 2017 maintenance period is now complete, far ahead of schedule.  We have brought compute nodes online and released previously submitted jobs.  Login nodes are accessible and data available.  Our next maintenance period is scheduled for Thursday May 11 through Saturday May 13, 2017.

Removal of obsolete /usr/local/packages

  • This set of old software has been made inaccessible. Loading the ‘oldrepo’ module will now result in an error and instructions for contacting PACE support for assistance in utilizing the current software repository.

Infiniband switch swap

  • Replacement complete, no user action needed.

Readdressing network management

  • Work complete, no user action needed.

Upgrade of scheduler server for the NovaZohar cluster

  • Upgrade complete, no user action needed.  Further detail has been provided to the users of this cluster.

 

PACE quarterly maintenance – January 2017

Posted by on Thursday, 12 January, 2017

Dear PACE users,

It is again time for our quarterly maintenance. Starting at 6:00am Thursday, January 19, all resources managed by PACE will be taken offline. Maintenance is scheduled to continue through Saturday evening. Our next maintenance period is scheduled for Thursday May 11 through Saturday May 13, 2017.  We have a reduced scope this time around, as compared to our previous maintenance periods, with only one item visible to users.

Removal of obsolete /usr/local/packages
We will be removing (nearly) all content from /usr/local/packages. This set of software represents a repository two versions old, much of which is incompatible with the currently deployed operating system. We believe that this software is not currently in use – with one exception. We will continue to work with that user to accommodate their needs. Newer and/or compatible versions of all software being removed are available in the current repository.

Old modules, including the module that has been used to access to this old repository (oldrepo) will be removed. If attempt to load this module(s) in your environment or in PBS scripts, you will get an error. Please contact pace-support@oit.gatech.edu if you need assistance with finding replacement modules in the current repository.

Infiniband switch swap
We will replace a small infiniband switch used by infrastructure servers with one that has redundant power supplies. This was identified during the recent electrical maintenance by OIT. No user action is required.

Readdressing network management
With the assistance of the OIT Network Engineering team, we will move the management IP addresses for a number of network devices. This will make room for additional user-facing services. As these devices are not accessible to the user community, no user action is required.

Upgrade of scheduler server for the NovaZohar cluster
The scheduler server responsible for the NovaZohar cluster will be upgraded during the maintenance period. This will provide for additional performance for scheduler related tasks (submitting jobs, querying status, etc.) Previously submitted jobs will be retained, and resumed at the conclusion of maintenance. No user action is expected.