Schema Drift Detection Alert
Description
The Schema Drift Detection alert fires when data flowing through a monitored pipeline edge deviates from the learned schema, specifically when a record introduces a new field that was not present during learning, or when a field carries a type that was not previously observed.
This alert is system-managed: Monad creates and associates a Schema Drift Detection alert rule automatically for your organization. You do not need to create or configure it manually.
See the Schema Detection guide for how to enable schema detection on a pipeline edge and how the learning and detecting modes work. When this alert fires, see Handling drift alerts for how to triage and respond to each drift event.
When it fires
A Schema Drift Detection alert fires once per drift event on a given edge. Drift events are:
| Event type | Description |
|---|---|
new_field | A field appeared in a record that was not part of the learned schema |
type_change | A field carried a type not observed during learning |
An edge can fire multiple alerts in a single flush if several fields drifted in the same batch of records.
Alert severity
| Event type | Severity |
|---|---|
new_field | low |
type_change | medium |
When a single batch contains both new fields and type changes, the alert takes the highest severity (medium).
Viewing alerts in the UI
Fired alerts appear in the Alert Stream on the Alerts page. Each schema drift event shows:
- The affected pipeline and edge
- Severity badge
- A diff table of every changed field, with before and after types
- Links to view the edge in the pipeline and to open the schema viewer
Clicking View edge navigates directly to the pipeline with the affected edge selected.
Routing schema drift alerts
Like all Monad alerts, schema drift events can be routed to any destination through a pipeline using the Monad Alerts input. This lets you send drift notifications to Slack, PagerDuty, a SIEM, or any other output your organization uses.
Resetting after intentional schema changes
If you intentionally changed the schema of your source data (for example, added a new field to an upstream service), and want to reset the edge schema to make the new shape the new baseline, see Resetting the schema in the Schema Detection guide.