Configuring alerts on a job

Databricks Jobs support configurable notifications that alert stakeholders via email or webhook destinations when a job or individual task starts, succeeds, fails, or (for streaming jobs) exceeds a backlog threshold. Alerts are set on the job (and optionally per-task) in the job's Notifications settings, letting teams route failure alerts to on-call channels without custom code.

1 · Learn the must-know

• 1Notifications can be configured at the job level and independently overridden or added at the individual task level for finer-grained alerting.
• 2Supported trigger events include job/task start, success, failure, and streaming backlog threshold exceeded (for streaming tasks).
• 3Email notifications are sent to addresses entered directly in the job settings, while system destinations (e.g., Slack, PagerDuty, Teams, generic webhooks) must first be configured as a notification destination by a workspace admin before they appear as selectable targets.
• 4You can suppress notifications for skipped runs and for runs canceled due to another concurrent run, avoiding alert fatigue from non-failure events.
• 5Notifications configured via the UI are stored in the job's JSON definition (email_notifications/webhook_notifications) and can be set identically through the Jobs API or Terraform for automated deployment.
• 6Failure alerts fire per run of the job/task regardless of retry policy, so a job with automatic retries can still send multiple failure notifications before eventually succeeding unless notifications are scoped appropriately.

2 · Check your understanding