Databricks SQL alerts

Databricks SQL alerts run queries on a schedule and notify you when a condition that you define is met against the query result. When you schedule an alert, its associated query runs and the condition is evaluated. You can also view an alert history to review the results of past evaluations.

To learn how to work with legacy alerts instead, see What are legacy alerts?.

What you can do with alerts

Alerts let you monitor any SQL query result on a schedule. Use them to track business KPIs, monitor data quality, watch cost trends, and catch operational issues across your Databricks workloads. Common patterns across Databricks include:

• Monitor business metrics in metric views: Reference a Unity Catalog metric view by its fully qualified name in the alert query to monitor governed business metrics. See Alert on metric views.
• Detect data quality issues and anomalies: Pair alerts with Unity Catalog data quality monitors and anomaly detection so an unexpected metric, distribution shift, or profile change sends a notification. See Alerts for anomaly detection and Profile alerts.
• Track usage and cost: Build alerts on system tables for serverless billing or ingestion to catch unexpected spend. See Monitor the cost of serverless compute and Monitor managed ingestion pipeline cost.
• Watch SQL warehouse and query health: Alert on warehouse events or query history to catch slow queries, failed sessions, or capacity issues. See Example queries for monitoring SQL warehouse activity and Warehouse events system table reference.
• Audit access and security events: Alert on audit log queries to flag unusual workspace activity. See Monitor Genie Spaces usage with audit logs and alerts.
• Catch failures in AI agents: Alert on agent quality metrics so failures and emerging issues surface during development and operation. See Guide: Agents development workflow.
• Run an alert as a task in Lakeflow Jobs: Add an alert as a task so condition checks run on a pipeline trigger and downstream tasks can branch on the result. See SQL alert task for jobs.

Get started with alerts

The following pages cover the most common alert tasks, from authoring a new alert to ongoing management:

• • Create an alert
  • Walk through the alert editor end-to-end. Includes advanced settings and notification template customization.
• • Manage alerts
  • Find alerts on the listing page, share them, transfer ownership, and track changes with Databricks Git folders.
• • Choose a SQL warehouse
  • Select and size the warehouse that runs your alert query for reliable, cost-efficient evaluations.
• • Alert query patterns
  • SQL patterns for aggregations, multicolumn conditions, and metric views.

Differences from legacy alerts

The latest version of Databricks SQL alerts behaves differently from legacy alerts in a few key ways:

• Query reuse: An existing saved SQL query cannot be reused when creating an alert. Each alert owns its query definition, which can be authored directly in the new alert editor.
• Alert status values: Alert states are simplified and alerts no longer support the UNKNOWN status from legacy alerts. Evaluations resolve to OK, TRIGGERED, or ERROR.

You can continue to use both the latest alerts and legacy alerts side by side while you transition.