US HPC Facilities Meeting
Meeting notes:
Attendees: Doug, Rob, Paolo, Xin, Wei
Need a champaign to drive the HPC allocation @ NSF - From Rob: need ALTAS physicists to come out with a convincing use case for (NSF) HPC such as the Bluewater. MWT2 already spent manpower on it and all it need is allocations.
DOE HPCs: discussing with DOE about long term commitment - 75% of simulation on HPCs.
NERSC: Want more space. considering moving NERSC_DATADISK to global scratch instead of project GPFS space - this means ATLAS data there can't be for permanent. and need to coordinate NERSC purging and RUCIO update; Need to run deletion sooner, and get storage dump (cron job) back and running.
Using containers may help reduce load to shared file systems; Xin is helping setup a container building machine at BNL - expect daily builds and shipping out Shifter and Singularity container images (for HPCs) at a O(10^2) GB / day.