HPC Upgrades

Our yearly hardware and software updates are coming up soon. We have a lot of exciting new software packages and hardware upgrades.

The major upgrades include:

  • We have purchased and are installing the TotalView Debugger package for all users
  • We are upgrading to RedHat Enterprise Linux v6.5
  • We are implementing the new,faster FDR standard for parts of our InfiniBand Network
  • We are replacing older, slower nodes with new hardware
  • We recompiling and upgrade many software packages to newer versions

These upgrades will require several maintenance periods where our services will be offline. We will try to minimize the impact on research as much as possible during these periods.

Our tentative schedule for these maintenance periods is:

  • August 15
    We will take the 2008 and 2009 nodes offline. Running jobs will continue, but new jobs will not be scheduled to use these nodes.
  • August 18
    We will cancel all jobs running on the old 2008 and 2009 nodes at 9am. This will not impact jobs on newer nodes.
    At this point, we are going to shutdown and unmount all old nodes from the HPC racks at Sliger. These nodes are going to be refurbished made part of the Condor cluster. More details about this
  • September 1 - September 8
    We will turn off all HPC compute and login nodes. This will cause all running jobs and login sessions to terminate. HPC will be unavailable.
    At this point, we are going to install and activate our new 2014 nodes, move our older 2012/2013 in the data-center and make performance/stability upgrades to the Panasas filesystem.
  • September 4
    Login nodes will be back online, but most compute nodes will be offline.
  • September 8
    All upgrades complete. All nodes back online, including new ones.

Let us know if you have any questions or concerns.