Environment Module

From Leo's Notes
Last edited on 18 April 2023, at 16:20.

Environment modules provides a reversible way to alter a user's environment to make software packages available.

Cheat Sheet[edit | edit source]

Action Command
Load a module module load gcc/6.1.1
Unload a module module unload gcc
Show what's available module avail
List what's loaded module list

Creating a module[edit | edit source]

module avail will list all modules that it can find in the search path. Any modulefile that is found will be listed as a loadable module. To get started, create a directory containing all your modulefiles. For example, /share/modulefiles.

Add this path to module's search path with:

module use --append /share/modulefiles

Each new software package should have its own directory. Each version of the software should be its own file. For example, to create a new rstudio module for version 1.4:

# vi /share/modulefiles/rstudio/1.4

The modulefile[edit | edit source]

A modulefile is basically just metadata outlining what the module is for and what paths module should update for the software to work. Continuing from the rstudio example, we can have something similar to the following:

#%Module
set vers 1.4
global version

proc ModulesHelp { } {
    puts stderr "\tLoads rstudio.\n"
}

module-whatis "Loads rstudio 1.4 into the environment"

module load R
module load libpq/12.4

set rootdir "/software/modules/software/rstudio-1"

prepend-path PATH "${rootdir}/usr/lib/rstudio/bin"
prepend-path LD_LIBRARY_PATH "${rootdir}/usr/lib/rstudio/lib"
# prepend-path MANPATH /usr/local/modules/mpi/gnu/man

A few things to keep in mind:

  • Modulefiles start with the magic #%Module string
  • ModulesHelp is what gets run when running module help module-name
  • Dependencies are injected as 'module load' commands.
  • Anti-dependencies or package conflicts are defined with conflicts module-name.

See also[edit | edit source]

Alternatives[edit | edit source]