Phase one of the Lewis cluster upgrade will become available soon. The new system (Lewis 3) consists of 20 24-core machines with 128 GB of RAM. It will run CentOS 7 and the SLURM job scheduler and form the basis of future upgrades. As a result, most of the software will need to be recompiled and job files will need to be updated to run in the new environment.
The new and old systems will run simultaneously for a short time to allow users to migrate over to the new system. To facilitate this, the data and home directories will be the same on both systems.
We will also update our documentation and policies. The first change will be to reduce the maximum job time to ensure we can transition in a timely manner without disruption.
Watch this space for future updates and information about the upgrade.
— March 27, 2015