Under Construction

The S3DF (SLAC Shared Science Data Facility) is a SLAC-wide computing facility. For LCLS the S3DF will replace the PCDS current computer and storage resources used for data processing.  

Accessing the S3SDF

To access to the S3DF systems  connect to the s3dflogin

To get into the LCLS photon computing system you can connect to the pslogin pool from anywhere:

ssh pslogin.slac.stanford.edu

The SDF using the SLAC Active Directory authentication.  This means you need to use your SLAC Windows account rather than your UNIX account (even if the SDF is entirely Linux based).  If you don't have a SLAC Windows account, users with a SLAC UNIX account (this is the same as your LCLS account) can generate an Active Directory password by following the instructions here: Accounts Portal.  It's supposed to be straightforward for U.S. residents/citizens.  foreign nationals will get a  may have to fill out an "HTTP Request Failed" error and will receive email about filling out a DOE "FACTS" form.

To login, from a node inside slac (e.g. psnxopr, pslogin,...) type "ssh sdf-login".

SDF Documentation

Documentation on the SDF can be found here (scroll down a little to see the documentation).

The SDF uses SLURM.  LCLS users should use the partition (queue) names "lcls".  Documentation on submitting jobs can be found at the SDF SLURM installation page.

To see your disk quota usage use this command: /sdf/sw/sdf-quota.sh

Using psana in SDF

The psana software conda installation can be activated with "source /sdf/group/psdm/sw1/conda/manage/bin/psconda.sh" (python 2).  Add a "-py3" flag at the end to get python3.

Experiment Folder

The experiment data are stored below  /sdf/group/lcls/ds/data/<instr>/<expt>.

Access to the data is granted to all members of the unix group that has the same name as the experiment name. To check ones group membership either check the groups of a user: % id [user-name] or get all members of a group: % getent group <group-name>. There is currently an issue with the group access and the user should make the experiment group it's primary one using % newgrp <group-name> (newgrp will start a new shell with a new primary group).