About the Databricks connector

Integration Service connector for the Databricks workspace REST API in Automation Cloud, supporting jobs, notebooks, Unity Catalog volumes, Genie, and Vector Search.

Remarque :

The Databricks connector is available as a preview feature.

Automate a Databricks workspace over its REST API: submit and monitor notebook job runs, manage Unity Catalog volumes and functions, import and export notebooks, converse with Genie, and more.

Version Cloud prise en charge

The Databricks connector supports Databricks in the AWS, Google Cloud Platform and Azure clouds.

Remarque :

This connector is separate from the Databricks Agent connector, which is used to trigger Databricks agents from UiPath Maestro^TM. Use the Databricks connector for general workspace automation.

Authentification

Before automating processes, you need to establish a connection with your Databricks workspace. The connector authenticates as a service principal using OAuth 2.0 client credentials (machine-to-machine), with either a Databricks OAuth service principal or a Microsoft Entra ID service principal.

Step-by-step instructions are available in the Databricks authentication page.

Events

Ce connecteur ne prend pas en charge les événements.

Activités

You can use the connection to build automations that query data, orchestrate jobs and notebooks, manage volume files, and converse with Genie. The connector also includes a Databricks HTTP Request activity for reaching workspace API endpoints that the predefined activities do not cover.

For the full list, visit the Databricks activities page.

Documentation du fournisseur

Find more information in the official Databricks documentation:

• Documentation de Databricks
• Databricks REST API reference
• Databricks OAuth machine-to-machine (M2M) authentication
• Unity Catalog
• AI/BI Genie

Problèmes connus

• Insert Record on the Function object — Input Parameters is an object, but the default fields view sends it as a string, so the call fails. The workaround is to use Switch to object view and pass Input Parameters as an object. For more information on object view, see the Data Mapping editor.

Questions fréquemment posées

Which Databricks integration should I use?

UiPath offers three ways to work with Databricks. Choose based on what you are automating.

Intégration	Utilisez-le lorsque
Databricks (this connector)	You are automating the Databricks workspace itself: triggering and monitoring jobs, importing, exporting and running notebooks, managing Unity Catalog objects and volume files, asking questions through Genie, or querying Vector Search indexes.
Database Hub, with Databricks as the database type	You are doing table-level data work: reading, inserting, updating and deleting rows, and running custom SQL. Database Hub treats Databricks as one of several JDBC databases, so the same activities also work against SQL Server, PostgreSQL, Oracle and Redshift. It is also the only option that supports Databricks triggers, through the Record Created and Record Updated events.
Agent Databricks	You are calling a Databricks agent served from a model serving endpoint, from an agentic process in UiPath MaestroTM, using the Start and Wait for External Agent task.

The following table compares what each integration supports.

	Database Hub	Databricks (Preview)	Agent Databricks
Idéal pour	Table data	Jobs, notebooks, Genie	Calling agents
SQL and tables	✅	✅	❌
Lakebase	✅	Read only 1	❌
Data Fabric source	✅	❌	❌
Jobs, notebooks, files, Unity Catalog, Genie, Vector Search	❌	✅	❌
Agents and serving endpoints	❌	✅ 2	✅
Relais	SQL Server only	❌	❌
Authentification	OAuth, PAT	OAuth, Microsoft Entra ID	OAuth, PAT

¹ Lakebase databases registered in Unity Catalog can be queried through a SQL warehouse with Execute a SQL Query. This access is read only.

² Through the Databricks HTTP Request activity, which can call a serving endpoint's invocations API.

The three are independent and can be used side by side in the same automation.

My SQL query times out. What should I do?

Execute a SQL Query waits synchronously for the duration set in Wait Timeout. For statements that run longer than that, set Wait Timeout to 0s and On Wait Timeout to CONTINUE. The activity then returns a statement ID immediately, which you can poll with Get SQL Statement Result until the status is SUCCEEDED.

Why does an activity fail with a permission error even though the connection works?

The connection authenticates as a service principal. Creating the connection only proves the credentials are valid — the service principal also needs to be granted access to each object it touches, such as the SQL warehouse, job, Unity Catalog securable, volume, Genie space, or Vector Search index. Grant those permissions in Databricks and retry.

The endpoint I need is not covered by any activity. What are my options?

Use the Databricks HTTP Request activity, which calls any Databricks workspace REST API endpoint using the same connection. It accepts application/json only and does not support file upload or download.