Conda Environment Manifest File and Metadata

The odsc conda CLI generates the manifest file for each conda environment that contains the metadata for the environment.

The metadata is displayed in each environment card displayed in the Environment Explorer tab. A manifest file, <name>_manifest.yaml, is stored under each conda environment directory in /home/datascience/conda.

An environment's metadata includes these parameters in the manifest section of the manifest YAML file:

manifest_version

Important

Do not modify this parameter.

The version of the current manifest file is updated by the odsc conda CLI after the environment is published so manually setting this value is not needed.

name

The name of the conda environment. This parameter is defined as <name> in the osdc CLI commands.

description

A description of the conda environment, which is limited to 500 characters. Edit the manifest file directly if you want to modify the description.

version

The version of the conda environment. The version parameter doesn't necessarily reflect the version of any of the libraries installed in the conda environment. Versioning a conda environment allows you to track environment lineage over time. This parameter is defined as <version> in the osdc conda CLI commands.

arch_type

The type of hardware architecture supporting the conda environment at runtime. Two values are possible, cpu and gpu. Hardware checks are performed before installing a conda environment to a notebook session. The odsc conda check warns you of a mismatch between the notebook session architecture and the target architecture of the conda environment. Also, odsc conda detects the type of architecture of a notebook sessions, and sets the value to either cpu or gpu when you create a conda environment.

notebooks

Important

Do not modify this parameter.

Instructs odsc conda which notebook examples are associated with this conda environment. Notebook examples are stored in /home/datascience/conda/notebooks and are deleted when you delete an installed conda environment.

libraries

Use this parameter to highlight a series of important libraries that are available in the conda environment. user-defined. You can modify the list if you decide to add new libraries or modify the versions of preinstalled libraries. These libraries are listed under Top Libraries in the environment card.

commands

Use this parameter to write commands to run when a conda environment is installed. For example, some libraries require extra configurations and the storage of configuration files in a user's home directory. The commands parameter allows you to automate the configuration steps needed to run the conda environment. Supported commands include MKDIR to create directory, and COPY to copy files from the conda environment directory to some other path in /home/datascience.

You can look at the Data Science Pyspark conda environment for an example of an environment that applies the COMMANDS field.

Slug

Important

Do not modify this parameter.

A short string that uniquely identifies the conda environment. a combination of the name and version of the conda environment, which the odsc conda toolkit generates. This parameter is defined as <Slug> in the odsc conda CLI commands.

type

Important

Do not modify this parameter.

The odsc conda CLI generates and edits this parameter, which defines the type of conda environment. There are two types, data_science and user. The data_science condas are defined and provided by the Data Science service team. A user-defined conda is a conda environment that a Data Science user creates or modifies.

python

The version of Python installed in the conda environment for documentation purposes. This parameter is user-defined. If you change the version of Python, ensure that you update this parameter in the manifest.

pack_path

Path to Object Storage where this conda environment is stored. This path is in the Data Science service tenancy when the environment is a Data Science environment. If the environment is published, the path is to a bucket in a tenancy. This parameter is updated after the environment is published.

logo

The associated conda environment logo. Not in use.

The remaining parameters defined in the manifest file are standard parameters expected by the conda CLI command. Whenever a conda environment is installed, created, cloned, or published, odsc conda modifies the content of the manifest file to reflect the current changes that you made. It includes changes to the manifest parameters in the preceding list made using odsc conda commands and changes to the list of Python libraries installed in the environment.