Share via

Monitor and revoke personal access tokens

  • Article

To authenticate to the Azure Databricks REST API, a user can create a personal access token (PAT) and use it in their REST API request. A user can also create a service principal and use it with a personal access token to call Azure Databricks REST APIs in their CI/CD tools and automation. This article explains how Azure Databricks admins can manage personal access tokens in their workspace. To create a personal access token, see Azure Databricks personal access token authentication.

Use OAuth instead of personal access tokens

Databricks recommends you use OAuth access tokens instead of PATs for greater security and convenience. Databricks continues to support PATs, but due to their greater security risk, it is suggested that you audit your account’s current PAT usage, and migrate your users and service principals to OAuth access tokens. To create an OAuth access token (instead of a PAT) to use with a service principal in automation, see Authenticate access to Azure Databricks with a service principal using OAuth (OAuth M2M).

Databricks recommends you minimize your personal access token exposure with the following steps:

  1. Set a short lifetime for all new tokens created in your workspaces. The lifetime should be less than 90 days.
  2. Work with your Azure Databricks workspace administrators and users to switch to those tokens with shorter lifetimes.
  3. Revoke all long-lived tokens to reduce the risk of these older tokens getting misused over time. Databricks automatically revokes personal access tokens that haven’t been used in 90 or more days.

To assess your own organization’s usage of PATs and plan a migration from PATs to OAuth access tokens, see Assess personal access token usage in your Databricks account.

Requirements

  • You must be an Azure Databricks workspace admin to disable personal access tokens for a workspace, monitor and revoke tokens, control which non-admin users can create tokens and use tokens, and set a maximum lifetime for new tokens.
  • Your Azure Databricks workspace must be on the Premium plan.

Enable or disable personal access token authentication for the workspace

Personal access token authentication is enabled by default for all Azure Databricks workspaces that were created in 2018 or later. You can change this setting in the workspace settings page.

When personal access tokens are disabled for a workspace, personal access tokens cannot be used to authenticate to Azure Databricks and workspace users and service principals cannot create new tokens. No tokens are deleted when you disable personal access token authentication for a workspace. If tokens are re-enabled later, any non-expired tokens are available for use.

If you want to disable token access for a subset of users, you can keep personal access token authentication enabled for the workspace and set fine-grained permissions for users and groups. See Control who can create and use personal access tokens.

Warning

Partner Connect and partner integrations require personal access tokens to be enabled on a workspace.

To disable the ability to create and use personal access tokens for the workspace:

  1. Go to the settings page.

  2. Click the Advanced tab.

  3. Click the Personal Access Tokens toggle.

  4. Click Confirm.

    This change may take a few seconds to take effect.

You can also use the Workspace configuration API to disable personal access tokens for the workspace.

Control who can create and use personal access tokens

Workspace admins can set permissions on personal access tokens to control which users, service principals, and groups can create and use tokens. For details on how to configure personal access token permissions, see Manage personal access token permissions.

Set maximum lifetime of new personal access tokens

You can manage the maximum lifetime of new tokens in your workspace using the Databricks CLI or the Workspace configuration API. This limit applies only to new tokens.

Note

Databricks automatically revokes personal access tokens unused for 90 or more days. Databricks will not revoke tokens with lifetimes exceeding 90 days as long as the tokens are actively used.

As a security best practice, Databricks recommends the use of OAuth tokens over PATs. If you are transitioning your authentication from PATs to OAuth, Databricks recommends using short-lived tokens for stronger security.

Set maxTokenLifetimeDays to the maximum token lifetime of new tokens in days, as an integer. If you set it to zero, new tokens are permitted to have no lifetime limit. For example:

Databricks CLI

databricks workspace-conf set-status --json '{
  "maxTokenLifetimeDays": "90"
}'

Workspace configuration API

curl -n -X PATCH "https://<databricks-instance>/api/2.0/workspace-conf" \
  -d '{
  "maxTokenLifetimeDays": "90"
  }'

To use the Databricks Terraform provider to manage the maximum lifetime for new tokens in a workspace, see databricks_workspace_conf Resource.

Monitor and revoke tokens

This section describes how to use the Databricks CLI to manage existing tokens in the workspace. You can also use the Token Management API. Databricks automatically revokes personal access tokens that haven’t been used in 90 or more days.

Get tokens for the workspace

To get the workspace’s tokens:

Python

from databricks.sdk import WorkspaceClient

w = WorkspaceClient()

spark.createDataFrame([token.as_dict() for token in w.token_management.list()]).createOrReplaceTempView('tokens')

display(spark.sql('select * from tokens order by creation_time'))

Bash

# Filter results by a user by using the `created-by-id` (to filter by the user ID) or `created-by-username` flags.
databricks token-management list

Delete (revoke) a token

To delete a token, replace TOKEN_ID with the id of the token to delete:

databricks token-management delete TOKEN_ID