Prerequisites

  • Get a UNIX S3DF account and have it activated via coact
  • Can connect using Putty, MobaXterm, terminus (recommended as you can pre configure hosts to ssh with one click), warp (highly recommend - rust based super fast ai terminal for mac) etc
  • Note that this can be accomplished using Windows 10+ using SSH in CMD / Mac terminal.
  • Security settings need to be specified
  1. Open Windows command line (type cmd in start menu) / mac terminal
  2. Type ssh  <UNIX_USERNAME>@s3dflogin.slac.stanford.edu
  3. Enter UNIX password
  4. (You’re in! But to do anything useful, ssh away from login node.)
  5. ssh into iana: ssh iana          (Now you can do something) Iana is one of the interactive compute nodes for general tasks. Please don't run any intensive compute tasks on iana.


Accessing AD and FACET Resources

For now there are two types of shared resources: coact manages SLURM on S3DF which is what is used to submit batch jobs and allows access to CPUs and GPUs. Shared storage/data is managed via posix on RHEL. When you submit a request to be added to the S3DF FACET facility someone from FACET will manually add you to the posix that allows write privileges to FACET shared storage. Please confirm with your supervisor or FACET contact which resources (AD or FACET or both or other) you should request access to. Resources on S3DF are divided up into "facilities" and "repos".

How to request to be added to the AD and FACET facilities

Login to coact using your SLAC unix account. On the top of the bar is a "repos", select that.

If you need access to FACET specific resources:

  • Select "Request Access to Facility" and select "FACET" from the drop down and submit.
  • At the moment there is only the "default" FACET repo, so you do not need to request access as it will automatically add you to the default repo when you join the FACET facility.
  • Someone will manually add you to posix so you can access data. This can take several hours.

If you need access to AD specific resources:

  • Select "Request Access to Facility" and select "AD" from the drop down and submit.
  • When you're part of the AD facility, you can request access to "beamphysics" by selecting "Request Repo Membership" and entering "beamphysics" then on the drop down selecting "AD" and submitting.
  • To be added to the posix group for beamphysics see direction here, under prerequisites: Getting started on s3df

First time only - setup Conda

Here are the instructions how to setup conda on s3df:

https://s3df.slac.stanford.edu/#/reference?id=conda

From the instructions: It is not recommended to store your conda environments in your $HOME due to 1) quota limits, and 2) an inability to share conda environments across groups. We generally recommend that you install software into your $GROUP space (eg /sdf/group/facet/sw - please see $GROUP storage).

  1. ssh to an interactive cluster (after this step ssh  <UNIX_USERNAME>@s3dflogin.slac.stanford.edu)
  2. ssh iana
  3. run the miniconda setup script
  4. (optional) You can also use mamba instead of conda for a faster time creating environments. Mamba is used by first installing miniforge and then using the mamba command instead of the conda command. curl -L -O "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh"
    bash Miniforge3-$(uname)-$(uname -m).sh

  5. These four commands quickly and quietly install the latest 64-bit version of the installer and then clean up after themselves. To install a different version or architecture of Miniconda for Linux, change the name of the .sh installer in the wget command.

    mkdir -p /sdf/group/facet/$USER/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O /sdf/group/facet/$USER/miniconda3/miniconda.sh
bash /sdf/group/facet/$USER/miniconda3/miniconda.sh -b -u -p /sdf/group/facet/$USER/miniconda3
rm -rf /sdf/group/facet/$USER/miniconda3/miniconda.sh


     initialize miniconda for bash 

    /sdf/group/facet/$USER/miniconda3/bin/conda init bash
  6. create your a conda env
  7. conda create -n <NAME> python=3.11
conda activate <NAME>
#install some packages
conda install numpy scipy matplotlib h5py jupyterlab

More information on conda:

ARD's S3DF + conda + beamphysics tutorial and Getting Started on s3df with Python/Jupyter


  • No labels