Skip to content

installation.md

Latest commit

 

History

History
48 lines (30 loc) · 1.58 KB

installation.md

File metadata and controls

48 lines (30 loc) · 1.58 KB

Installation

Prerequisites

  • Databricks Workspace: Access to a Databricks workspace with an active Spark session
  • Python Dependencies: Ability to install packages via %pip in Databricks notebooks
  • CDF Credentials: Access to CDF credentials (client_id, client_secret, tenant_id, cdf_cluster, project)
  • CDF Data Model: A CDF Data Model with Views (for Data Model UDTFs) or Time Series (for Time Series UDTFs)

Install Required Packages

Option 1: Install from PyPI (Recommended for Production)

In a Databricks notebook, install the required dependencies from PyPI:

%pip install cognite-sdk cognite-databricks

Important: After installing packages, restart the Python kernel when prompted. This ensures all dependencies are properly loaded.

Option 2: Install from Wheel Files (For Testing Local Builds)

If you're testing locally built wheel files, install them directly:

%pip install --force-reinstall \
  /Workspace/Users/user@example.com/wheels/cognite_pygen_spark-0.1.0-py3-none-any.whl \
  /Workspace/Users/user@example.com/wheels/cognite_databricks-0.1.0-py3-none-any.whl

Note: Replace the paths with your actual wheel file locations. After installation, restart the Python kernel.

Verify Installation

from cognite.databricks import generate_udtf_notebook, register_udtf_from_file
from cognite.pygen import load_cognite_client_from_toml

# Verify imports work
print("✓ All imports successful")

Next Steps

Once installation is complete, proceed to Registration to register your UDTFs.