Databricks CLI

0.18.0 · deprecated · verified Sun Apr 05

The `databricks-cli` Python package provided a command-line interface for Databricks. As of version 0.18.0, this package is deprecated. Users are advised to migrate to the new Databricks CLI (an independent executable) for command-line operations and to the `databricks-sdk-py` Python SDK for programmatic interactions. Databricks plans no further support or new feature work for this legacy package.

Warnings

breaking This `databricks-cli` Python package is deprecated. It receives no new features or support. For programmatic use, migrate to `databricks-sdk-py`. For command-line use, migrate to the new standalone `databricks/cli` executable. Failure to migrate will result in outdated functionality and potential compatibility issues.
Fix: For Python SDK usage: `pip install databricks-sdk` and update imports/code to use `databricks.sdk.WorkspaceClient` or `databricks.sdk.AccountClient`. For CLI usage: Install the new Databricks CLI executable via `curl -fsSL https://raw.githubusercontent.com/databricks/setup-cli/main/install.sh | sh` and adapt scripts to use the new `databricks` commands.
deprecated Personal Access Token (PAT) authentication for Databricks is deprecated, and basic authentication (username/password) reached End-of-Life on July 10, 2024. New projects should use OAuth machine-to-machine (M2M) or user-to-machine (U2M) authentication.
Fix: Transition authentication methods to Databricks unified authentication, prioritizing OAuth M2M/U2M. Update `DATABRICKS_TOKEN` usage to alternative methods where possible. Refer to Databricks authentication documentation for the latest best practices.
gotcha Version 0.18.0 of `databricks-cli` removed the explicit pin on `urllib3 < 2`. While this unblocks newer `urllib3` versions, older environments might encounter compatibility issues if they upgrade `databricks-cli` but retain an old `urllib3` that has breaking changes with other dependencies.
Fix: Ensure `urllib3` is updated to a compatible version alongside `databricks-cli` (or its recommended replacements, the `databricks-sdk` or new CLI) to avoid unexpected errors. If encountering issues, check the `urllib3` version in your environment and upgrade it or pin a known working version.

Install

pip install databricks-cli Install legacy databricks-cli
pip install databricks-sdk Recommended: Install new Databricks SDK for Python
curl -fsSL https://raw.githubusercontent.com/databricks/setup-cli/main/install.sh | sh Recommended: Install new Databricks CLI executable

Imports

ApiClient, service
```
from databricks.sdk import WorkspaceClient
# Or, for account-level operations:
from databricks.sdk import AccountClient
```
The programmatic interface (SDK-like usage) of `databricks-cli` is deprecated. Migrate to the official `databricks-sdk` for Python, which offers `WorkspaceClient` for workspace-level APIs and `AccountClient` for account-level APIs.

Quickstart

This quickstart first demonstrates how to use the recommended `databricks-sdk` for Python to programmatically list Databricks clusters, assuming authentication via environment variables (`DATABRICKS_HOST`, `DATABRICKS_TOKEN`). It also highlights typical command-line usage for both the legacy `databricks-cli` and the recommended new `databricks` CLI executable. Users should migrate programmatic logic to `databricks-sdk` and command-line scripts to the new `databricks` CLI.

# Authenticate via environment variables (recommended for quickstart)
# export DATABRICKS_HOST="https://<your-workspace-url>"
# export DATABRICKS_TOKEN="dapiXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX"

# --- Old databricks-cli usage (deprecated) ---
# The primary use was via the command line, e.g.:
# databricks clusters list

# --- New Databricks SDK for Python (recommended for programmatic access) ---
import os
from databricks.sdk import WorkspaceClient

# The SDK automatically uses DATABRICKS_HOST and DATABRICKS_TOKEN environment variables
w = WorkspaceClient(
    host=os.environ.get('DATABRICKS_HOST', ''),
    token=os.environ.get('DATABRICKS_TOKEN', '')
)

# Example: List clusters using the new SDK
print("Listing Databricks clusters (using new SDK):")
for c in w.clusters.list():
    print(f"  - {c.cluster_name} ({c.cluster_id})")

# --- New Databricks CLI executable (recommended for command-line access) ---
# After installing the new CLI executable (not this Python package):
# databricks auth login --host https://<your-workspace-url>
# databricks clusters list

view raw JSON →