Skip to content

Advanced topics#

This section covers aspects of Sherlock that go beyond the basics covered in the Getting started and User guide sections, and require more configuration or understanding of the underlying system.

  • Connecting

    The default login.sherlock.stanford.edu alias is suitable for most use cases. This page covers scenarios where more control is needed: connecting to a specific login node, available authentication methods, and SSH client configuration options.

    Connection options

  • Jobs

    Goes deeper into Slurm job configuration: the full set of #SBATCH directives, queue and wall-time limits, Slurm constraints for targeting specific hardware, and the service partition for lightweight recurring tasks such as data transfers.

    Job submission options

  • Scheduling

    How Slurm decides when and where to run your jobs: the backfill algorithm, job distribution across nodes, and practical tips for reducing wait times.

    Scheduling