Setting Up Your Environment =========================== This page explains how to set up your **bash or Python environment** in a Slurm job runtime. We adopt a modular approach to software and package distribution. The information reported here applies to all jobs submitted both from the Magellano shell and the OnDemand interface. See also `User Interfaces `__. Environment Modules (``module``) -------------------------------- The ``environment-modules`` utility allows you to load packaged and libraries from a shared location. The job runtime environment is a minimal AlmaLinux9 instance. Only a small set of packages are installed in that environment. With the ``module`` you can browse an load software into your session. You can get a list of available modules with the following command:: module avail This will print a list of softwares grouped by platform and/or type. You can load one or more modules into your environment. For example, to load the latest installed version of Python:: module load python This will append the correct binary paths to your ``PATH`` environment variable, so that Python will be available in your shell:: $ which python /cvmfs/software.pi.infn.it/spack/prod/linux-x86_64_v3/python-3.13.5-jvkjfssfgkupkpfczeupvfat7w6nekjv/bin/python You can specify the exact version of the module you want to load:: $ module load python/3.13.5-gcc-11.5.0 $ python --version Python 3.13.5 Finally, you can unload a module as follows:: module unload python Modules can be loaded in the **job runtime** by adding the proper ``module`` commands in your batch script. .. Note:: Modules are loaded into your **current session**. When starting a new session, you will have to reload them. You can request the installation of new modules by issuing a ticket to our `Service Desk `__. For more info about Environment Modules, see the `official documentation `__. Python Virtual Environments (``venv``) -------------------------------------- A virtual environment is an isolated directory tree where you can install the dependencies of your Python code. In Magellano, Python versions are distributed as modules, as explained above. You can install **Python packages** by loading the desired Python version and creating a virtual environment. First, create a virtual environment in a chosen path (``VENV_ROOT``). A typical choice is to put it in a subdirectory of your home directory:: module load python python -m venv $HOME/my_environment In this case ``VENV_ROOT`` is ``$HOME/my_environment``. After creating the environment, you can **activate** it by sourcing the activation script in the ``VENV_ROOT/bin`` directory:: source $VENV_ROOT/bin/activate When your virtual environment is active, the name of the environment will appear in your command prompt between brackets (for example, ``(my_environment)``). .. Hint:: After creation, you won't need to load the Python module again. The path to the correct Python version is coded in the ``activate`` script. With your environment activated, you can install Python packages using **PyPI**. For example:: pip install pandas You can easily check that the packages have been installed under ``VENV_ROOT``:: (my_environment) $ python -c "import pandas; print(pandas.__file__)" /gpfs/ddn/users/neo/my_environment/lib/python3.13/site-packages/pandas/__init__.py You can deactivate your environment with the ``deactivate`` command. .. Hint:: It's a good practice to have separated Virtual Environments for different use cases. You can request the installation of shared Python Virtual Environments by issuing a ticket to our `Service Desk `__. Learn more about Python Virtual Environments in the `official documentation `__.