Problem

Your serverless SQL warehouse is configured to auto-stop after five minutes of inactivity and does not have any visible queries in the query history or queries executing. Despite this, the warehouse repeatedly activates. 

Cause

A remote JDBC server is connecting to your serverless SQL warehouse using the Databricks JDBC driver.

JDBC connections are typically used for monitoring, polling, or maintaining a connection to the warehouse. While these connections do not execute queries visible in the query history, they can still cause the warehouse to remain active or spin up. 

Solution

Review your JDBC connections and adjust the connection settings.

Review JDBC connections

Investigate the source and purpose of the JDBC connections to your serverless SQL warehouse. Identify if these connections are necessary for your operations or if they can be optimized.

• Platforms such as Power BI, Tableau, and Looker frequently use JDBC or ODBC connections to query the warehouse and fetch data for reports and dashboards.
• Internal or third-party applications may use the Databricks JDBC driver to connect programmatically, for purposes such as ETL processes, data transformation tasks (for example, dbt), or analytics platforms.
• Your organization may implement scripts or services to check warehouse status and uptime or collect telemetry, which can also trigger JDBC sessions.

Adjust JDBC Connection Settings 

If the JDBC connections are used for monitoring or maintaining a connection, check if the connection settings can be adjusted to reduce the connection frequency. 

For more information, refer to the Data warehousing on Databricks (AWS | Azure | GCP) documentation.