US ATLAS hosted two shared Tier 3 at BNL and SLAC, also known as Analysis Facilities (AF). These
two faclities are available to all US ATLAS physicists and computer scientists.
…They are
orgniazed and managed to support US ATLAS users’ need on computing resources, including login,
run interactive and batch jobs, access ATLAS data, store private data, etc.
These two facilities also support tools specific for users analysis, including ATLAS/CERN
software in CVMFS, Grid middleware, Rucio clients, Machine Learning packages, MPI, Jupyter
Lab with PyROOT, Xcache with auto data discovery, GPUs, etc.
The two facilites are backed by staffs to support software environment, unix systems and
storages.
The processes can take days as both BNL and SLAC are DOE national labs. Do NOT wait until the last minute.
Applying BNL computing accounts is a multiple-step process.
The steps are summarized at here
Applying SLAC computing accounts
is a two-step process: becoming a SLAC laboratory user, and then obtain computing account(s).
SLAC is building a new computing facility - SLAC Shared Scientific Data Facility (SDF). On the technical side, it is an HPC environment built upon SLURM, Lustre and InfiniBand. Future US ATLAS resource at SLAC will be invested at SDF. At this stage, we would like to ask the JupyterLab users at the SLAC AF to prepare yourselves to login and switch to use JupyterLab at SDF ASAP by following the instruction below.
SDF will use a new identity management system (aka “SLAC ID” - it will be a computer account to login to everything at SLAC). If you already have a SLAC Windows account, you are all set (SLAC ID = SLAC Windows account) and go to the next step. If you don’t have a SLAC Windows account, please go to SLAC SDF page and click “Accounts Portal”. After this, give it a hour for the changes to be proprogated through SLAC computing.
ssh sdf-login01.slac.stanford.edu (or sdf-login02) using your “SLAC ID”. The first time you login, a new home directory of 25GB will be created automatically. You can then logout and follow the JupyterLab link below.
What will happen to your GPFS or AFS spaces? The ATLAS GPFS spaces will be accessible at SDF. It just won’t be your home directory. For AFS spaces, you will need to manually copy your files in AFS to SDF since SDF does not support AFS.
Note: Except the JupyterLab, the existing computing accounts and environment for ATLAS will continue until the hardware retires, which is a year or two from now (September 2020).
A tutorial of analysis example at Tier3s of BNL/SLAC is available at here. It was initially given at the Aug. 2019 US ATLAS Week at University of Massachusetts Amherst.
Documents, examples and entry points of the JupyterLab at BNL and SLAC.