The CSF2 has been replaced by the CSF3 - please use that system! This documentation may be out of date. Please read the CSF3 documentation instead. To display this old CSF2 page click here. |
Getting Started on the CSF (Info for New Users)
Please look through the getting started pages which cover the initial tasks new users will want to perform. A brief introduction to the CSF is given below. The sections in the menu on the left provide more details.
Requesting an Account
Please see the user accounts page for how to request an account, vistor/collaborator access, extensions to accounts and policies applicable to all users.
The CSF is used Remotely from your Desktop or Laptop Machine
The CSF is accessed from the comfort of your office (or home, via the correct method). Full details are provided in the Connecting to CSF documentation, for access from Linux, MacOS and MS Windows.
The CSF is a Cluster
The CSF is a cluster of machines: users of the system start a session on the login node (by means of an SSH client). From here you can submit computational jobs (see below) and do common tasks such as editing files, transferring files on/off the system.
Running Computational Work
All computational work must be submitted to the batch system, SGE, which farms out submitted jobs to the 100s of compute nodes. The compute nodes have more cores and much more memory than the login nodes.
We strongly recommend that all new users on the system try the 10 minute batch tutorial to become familiar with how to submit jobs to the batch system.
Please note: Any computational work found running outside of the batch system, for example applications started directly on the login node, will be killed immediately, without warning. Users who repeatedly run work outside of the batch system will have their account suspended.
Full documentation on the batch system and how to run jobs is provided on other pages on this site.
N.B. The maximum time for which a job may be run is one week. After one week jobs are automatically killed.
Use of Scratch Space
Users have a home directory on the central RDS file system and a scratch directory on the local, fast, scratch file system.
Users are encouraged to run jobs from their scratch area so that job outputs will be written to files in the scratch filesystem, as this filesystem is bigger and faster than others available on the CSF. However, scratch is suitable for only short-term storage of files — any file over three months old may be deleted without warning. Run your jobs from scratch then copy the important result files back to home.
More information about CSF filesystems and how to use the please see here.
MS Windows Users
The CSF uses the linux operating system. If you are a windows users there are some common pitfalls you may encounter when starting to use the CSF. Please read our guide on Using the CSF from Windows to understand how best to use the CSF from your MS Windows desktop/laptop.
Training Courses
The Intro to CSF course runs a couple of times each semester – please see the details and registration form (uni login required).
Users who are completely new to this kind of computational environment, and research computing in general, may also wish to consider attending other IT Services’ training courses (some of the materials are available online, see individual course pages).
Getting Help
The information given here is specific to the CSF and assumes some familiarity with Linux and/or working on HPC systems. The documentation is comprehensive but please do contact us at its-ri-team@manchester.ac.uk if you have any questions about using the system.