Installing (Third Party) applications

NeSI provides a long list of various applications on its systems. Nevertheless, if you need additional applications or libraries (below called package), we distinguish:

  • you need a newer version of an already installed package: ask NeSI support for an update
  • you need an older version of an installed package: please use the Easybuild installation procedure (below) to install it into your working space
  • you want to test a new (not installed) package: below we collected some hints, how you can install it in your user space.

In any case, if you have issues, do not hesitate to open a ticket and ask NeSI support for help.

Additional Packages to installed software

If you are looking for additional packages, e.g. for Python or R, please have a look to the application documentation

Using Easybuild

Most of the applications and libraries on the NeSI systems are installed using Easybuild. This helps to provide a reproducible installation process. The installation process is controlled by easyconfig files.

In the following an older version of GST is installed into your project directory (/nesi/project/<projectID>):

  1. setup the EasyBuild environment
    export NESI_EASYBUILD_PROJECT_ID=<projectID>
    module load project
    This sets the Easybuild environment with related directories for source files, module files and software into your project directory of project <projectID>.

    Note for advanced users: You can change the directories to a different location, using the following environment variables:

    export EASYBUILD_INSTALLPATH_SOFTWARE=/nesi/nobackup/<projectID>/<dirToInstall>

    The path to the modules could be also set to a more general or more easy accessible location as /nesi/nobackup/<projectID>/modules using the EASYBUILD_INSTALLPATH_MODULES environment variable.
  2. create a temporary directory and get the EasyBuild easyconfig file
    cd  <pathToEasybuildDir>   # where to handle the Easybuild configuration files

    cp /opt/nesi/mahuika/easybuild-resources/ebfiles_repo/GSL/GSL-2.3-gimkl-2017a.eb ./GSL-1.6-gimkl-2017a.eb
  3. modify the easyconfig, e.g. change the version of the package (pay attention that with version changes checksum changes, you could update or remove a checksum line).
    vi GSL-1.6-gimkl-2017a.eb
  4. start Easybuild (we already set directories with the environment variables above)
    eb GSL-1.6-gimkl-2017a.eb

This should create the application in the specified directory.

To use these modules you simple load

module load project
module load GSL/1.6-gimkl-2017a


Third party applications

Installation instruction vary from application to application. In any case we suggest to read the provided installing instructions. Nevertheless, the following should give you an impression which steps you usually need to consider:

  • Change into a desired source code directory. We suggest to use /nesi/nobackup/<projectID> or /nesi/project/<projectID>
  • download the source code. This could be done via a repository checkout (git clone <URL to the application source repository>) or via downloading a tarball (wget <URL to the tarball>). Unpack the tarball using tar xf <tar file name>. Change into source directory.
  • load compiler module and modules for additional libraries (module load gimkl FFTW)
  • run the configure with appropriate options ./configure --prefix=<desired install directory> --use-fftw=$EBROOTFFTW  (options can be listed using ./configure --help)
  • In other applications you need to adjust the provided Makefile to reflect compiler, and library options (see below)
  • compile code (make)
  • install the binaries and libraries into the specified directory (make install)

Create your own modules

You can create personalised module environments, which can load modules and set up environment variables. For example, you could define a modules in a project directory /nesi/project/<projectID>/modulefiles/ProdXY as the following:


conflict ProdABC   # conflicts with other versions of itself
module load CMake  # load other additional modules
module load netCDF/4.4.1-gimkl-2017a
module load Python/3.6.3-gimkl-2017a
# provide a description whatis "The ProdXY for doing clever things." proc ModulesHelp { } { puts stderr "This module loads the ProdXY tool. It requires CMake and netCDF." puts stderr "\t the executable prodXY is provided." }
# set the path to the software product (can be used later in the module) set PKG_PREFIX /path/to/software/package/ProdXY
# add the location of binaries to PATH, such they are immediately accessible prepend-path PATH $PKG_PREFIX/bin # add to library path for dynamically linked applications prepend-path LD_LIBRARY_PATH $PKG_PREFIX/lib
# add a location for Python packages
prepend-path PYTHONPATH $PKG_PREFIX/lib/python3.6/site-packages/
# for example, you can set environment variables for compiling setenv CFLAGS "-DNDEBUG"

In the first lines, we can set conflicts with other modules (here named ProdABC). Then we load some dependency modules and provide some description. The additional lines depend on your requirements for the module. With set you can define internal variables (within this module file). The command setenv defines a environment variable. And prepend-path and append-path extend an environment variable at the front or end.

There are common environment variables like:

  • PATH for providing executabl,
  • LD_LIBRARY_PATH for self created libraries,
  • PYTHONPATH for providing Python modules,
  • CONDA_ENVS_PATH for providing Conda environments,
  • etc.

And others which are very application specific.

To use the module (or all in that directory and sub-directories) we need to register that directory to the module environment. This can be done by setting the following environment variable:

module use /nesi/project/<projectID>/modulefiles/ 

by adding that line to your $HOME/.bashrc you will have the modules always available.

The module then can be loaded by:

module load ProdXY

These modules can easily be shared with collaborators. They just need to specify the last two steps.


Conda environmenta and python packages:

Please see here.



Was this article helpful?
0 out of 0 found this helpful