Module system

To use the software that is installed on the cluster (programs, compiler and librarys) it has to be loaded using the "module-system". The most used moduls are load per default and they put on screen at login on the cluster. To display this list again you can use module list:

To display all available modules you can use the command module avail (only module names) or module whatis (module names with description):

With module avail <directory> you can list all modules, which starts with the given directoryname. This is case-sensitive. In the first example we search about moduls, whos directory starts with "mpi". In the second example we search case-insensitive for all modules with "mpi" in the name:

A module which is displays using module avail can be load with module load <modulename>:

You can load serval modules with one command using module load <modulename1> <modulename2>.

Some modules exclude each other e.g. different versions of the same program. To switch a module, the old one has to be unload using module unload <modulename>:

To switch a module version with only one command you can use module switch <old-module> <new-module>:

To test a new program you may wish to unload all modules with module purge at the same time:

The software you can load using the module system is installed on only one server of the Linux-Cluster. If you unload a module there is maybe a (usually older) version of the same program installed directly on the node. For example after module purge there will be a GCC compiler available on every node.

For further information on the module system type the command module help.