What causes EmptySourceWarningElevatedToError?

A source table was truncated upstream (e.g., by a failed ETL load) and the dbt freshness check found zero rows newer than the warn threshold. A new data source was configured in dbt but the first load has not yet run, leaving the source table empty while dbt freshness checks are already active. The error_if threshold is more aggressive than the underlying load frequency, causing a legitimate weekend or holiday data gap to be treated as an error

How do I fix EmptySourceWarningElevatedToError?

Step 1: Check the actual source table row count and the maximum loaded_at timestamp to determine whether the data is genuinely missing or just delayed.. Step 2: If the upstream ETL failed, investigate and re-run the load job. For ADF or Databricks pipelines, check MetricSign's job_failed incidents for the upstream connector.. Step 3: If the error threshold is too aggressive for the load frequency, adjust the error_if value in the source YAML to match the expected load cadence.. Step 4: Use dbt source freshness --select source: to re-check freshness after the upstream load is fixed.

High severitydata_qualitydbt Cloud →

dbt Error:
EmptySourceWarningElevatedToError

What does this error mean?

A dbt source freshness check found a source table with no rows newer than the configured threshold. The project's error_if setting escalated this from a warning to a hard error, blocking all downstream model runs.

Common causes

1A source table was truncated upstream (e.g., by a failed ETL load) and the dbt freshness check found zero rows newer than the warn threshold
2A new data source was configured in dbt but the first load has not yet run, leaving the source table empty while dbt freshness checks are already active
3The error_if threshold is more aggressive than the underlying load frequency, causing a legitimate weekend or holiday data gap to be treated as an error

How to fix it

1Step 1: Check the actual source table row count and the maximum loaded_at timestamp to determine whether the data is genuinely missing or just delayed.
2Step 2: If the upstream ETL failed, investigate and re-run the load job. For ADF or Databricks pipelines, check MetricSign's job_failed incidents for the upstream connector.
3Step 3: If the error threshold is too aggressive for the load frequency, adjust the error_if value in the source YAML to match the expected load cadence.
4Step 4: Use dbt source freshness --select source:<source_name> to re-check freshness after the upstream load is fixed.

Frequently asked questions

What is the difference between warn_if and error_if in dbt source freshness?

warn_if sets a threshold at which dbt emits a warning but still exits with code 0. error_if sets a stricter threshold at which dbt exits with code 1, blocking downstream model runs in most orchestration setups.

Can I skip source freshness checks for a specific source without removing the config?

Yes. Pass --exclude source:<source_name> when running dbt source freshness, or temporarily set the source's freshness config to null to disable it without removing the definition.

Source · docs.getdbt.com/reference/exceptions