Installing DBND

From the command line, run the following command:

pip install databand

The DBND PyPI basic package installs only packages required for getting started. Behind the scenes, DBND does conditional imports of operators that require extra dependencies.

Whether you are looking to track your pipeline metadata, or if you want to orchestrate pipelines, you may want to install DBND plugins for integrating with third-party tools.

| See Connecting DBND to Databand to learn how to connect SDK integrations into your Databand Application.

Installing Plugins

Run the following command to install any of the plugins listed in the tables below. For example:

pip install dbnd-spark dbnd-airflow

You can use bundled installation via databand[plugin-slug]

pip install databand[spark,airflow]

Plugins for tracking

Plugin name	Description
dbnd-airflow	Enables monitoring of Airflow DAGs by DBND. Supported versions are 1.10.7-2.2.3
dbnd-airflow-auto-tracking	Enables automatic tracking for Airflow DAGs.
dbnd-airflow-export	Enables exporting of Airflow DAGs metadata from Airflow Web UI (used by dbnd-airflow-monitor service).
dbnd-luigi	Enables integration with Luigi. Monitors Luigi pipelines execution.
dbnd-spark	Required for Spark DataFrame observability features.
dbnd-mlflow	Enables integration with MLflow (submitting all metrics via MLFlow bindings).
dbnd-postgres	Enables integration with the Postgres database.
dbnd-redshift	Enables integration with the Redshift database.
dbnd-snowflake	Enables integration with the Snowflake database.

SDK version in different parts of the system.

It's strongly advised that you use the same SDK version across all components in communication - for example, an Airflow DAG Python environment and a Spark cluster environment.

System Requirements

You can install DBND on the following operating systems:

Linux (Recommended for production)
macOS
Windows

Software requirements

Before installing DBND, make sure you have the following software installed:

Mandatory	Software	Additional information
Mandatory	Python 3.6 and later	We recommend using Python 3.7 as it's our main supported version.
Strongly recommended	Virtualenv or * * Conda	Each of these virtual environments has its own Pip installation that is required. If you are an Anaconda user, run `conda install pip`, and follow the installation instructions for pip users.
Windows only	Microsoft Build Tools 2019
Linux only	setproctitle 1.1.10	Run `sudo apt-get install python-setproctitle`

Supported platforms and libraries

Platform/Library	Supported versions	Notes
Apache Airflow	1.10.7-2.2.3
Strongly recommended	0.24-1.1.5