/clustertools

Scripts to install an overlay of useful software for data analysis

Primary LanguageShell

Using an installed toolset

If software is installed at /path/to/software (i.e. you are reading this README at /path/to/software/README), then run the following:

eval `/path/to/software/setup.sh`

This will set up your PATH, PYTHONPATH, LD_LIBRARY_PATH, and a few other things appropriately. To have this occur on login, add the above line to your .profile, .bash_profile, or .csh_profile, depending on which shell you use.

The setup.sh script will point your environment at the standard toolset, which is currently py3-v4.

Current defined toolsets:

  • py2-v1: Python 2.7.11-based tools (deprecated)
  • py3-v1: Python 3.5.2-based tools (deprecated)
  • py3-v2: Python 3.6.1-based tools (deprecated)
  • py3-v3: Python 3.7.0-based tools (default)
  • py3-v4: Python 3.10.1-based tools (latest)
  • py3-v5: Python 3.12.3-based tools (dev)

Installing software for the first time

If you are just downloading this, make sure the repository is checked out to the location at which you want to install the software. Then run the buildall.sh script as follows:

/path/to/software/buildall.sh /path/to/some/scratch/directory

The scratch directory is used for temporary files only and will be deleted when the script completes. It should not be a network filesystem. It usually takes 2-3 hours to finish compiling and installing each toolset.

If you have a heterogeneous cluster with with multiple CPU architectures and/or operating systems, then run buildall.sh on each of them. The software will be automaically installed in subdirectories for each system architecture.