The python-nest conda environment

The python-nest environment is a conda python environment with the minimum requirements for ansible to partition and install an OS.
It is build for the tuxifier ansible collection. An example playbook repository can be found here: tuxifier-playbook.
A prebuild python-nest.squashfs is available for download from here: python-nest.squashfs
It contains the packages listed in python-nest-environment.yml, among them python, blivet, parted, e2fsprogs, xfsprogs, mdadm, cryptsetup and rpm-tools. The exact versions of the last build are in python-nest-environment_versioned.yml.

For the people that would like to improve or extend the functionality of this python environment, below the steps to build it.

Installing conda

On the following website is described how to install conda: https://conda-forge.org/download/

Conda prerequisite - geertsky channel

The directories in this repository are the recipes of the conda packages that are created for python-nest, or that are changed with respect to the conda-forge versions. Not all these changes have been merged yet to conda-forge. The geertsky channel contains the packages built from these recipes. For building the python-nest environment the geertsky anaconda channel needs as addition to the conda-forge channel. This can be done with the following command:

conda config --add channels geertsky

To see the configured channels we can issue a conda config --show-sources which should return:

channel_priority: strict
channels:
  - geertsky
  - conda-forge
report_errors: False

As an additional test we can issue a conda search parted which should list a build from the geertsky channel.

Conda prerequisite - conda-pack

Additionally, the conda-pack conda package needs to be installed in the base environment to pack the python-nest conda environment in a squashfs. This can be installed using the following command:

conda install -n base conda-pack

Building the python-nest environment

In this repository there is a conda environment file, python-nest-environment.yml, which can be used to build the python-nest environment. Using the following command, from the root of this repository, we can build the python-nest environment:

conda env create -f python-nest-environment.yml

To build exactly the package versions of the last release use python-nest-environment_versioned.yml instead.

Per default, the python-nest conda environment will be placed in ~/miniforge3/envs/python-nest/.

This environment can be activated for inspection using:

conda activate python-nest

Packing the python-nest environment

To pack the python-nest conda environment we need to use the following command:

conda-pack --compress-level 9 -j 8  --dest-prefix /local/conda/envs/python-nest --format squashfs -n python-nest

--compression-level 9 is needed to use xz compression. The only one supported by RHEL8
--dest-prefix /local/conda/envs/python-nest is needed as the environment gets mounted under /local/conda/envs/python-nest in the initramfs.
--format squashfs The environment needs to be packed in a squashfs.

Once conda-pack is finished, we have a file python-nest.squashfs containing the python-nest environment. This packed environment needs to be available in the dracut module directory of dracut-incubator. This is /usr/lib/dracut/modules.d/94incubator

sudo mv python-nest.squashfs /usr/lib/dracut/modules.d/94incubator

Now we're ready to build the initramfs. See: dracut-incubator