Between approximately 3:00 PM ET on April 6 and 11:30 AM ET on April 7, the Discovery API was returning stale model applied state data. Model metadata such as last execution timestamps and related attributes were not updating correctly after successful dbt Cloud job runs, causing incorrect data to be displayed in dbt Explorer and via the Discovery API.
A fix was deployed to the metadata ingestion pipeline at 11:30 AM ET on April 7 and the incident was resolved. Going forward, metadata will update correctly after each successful run. Models affected during this period will not be retroactively corrected - users will need to trigger a new dbt Cloud run to refresh their metadata.
Resolved
Between approximately 3:00 PM ET on April 6 and 11:30 AM ET on April 7, the Discovery API was returning stale model applied state data. Model metadata such as last execution timestamps and related attributes were not updating correctly after successful dbt Cloud job runs, causing incorrect data to be displayed in dbt Explorer and via the Discovery API.
A fix was deployed to the metadata ingestion pipeline at 11:30 AM ET on April 7 and the incident was resolved. Going forward, metadata will update correctly after each successful run. Models affected during this period will not be retroactively corrected - users will need to trigger a new dbt Cloud run to refresh their metadata.
Monitoring
Started: April 6, 2026 at approximately 3:00 PM ET Resolved: April 7, 2026 at approximately 11:30 AM ET
Summary: We identified an issue where the Discovery API was returning stale model applied state data. As a result, model metadata such as last execution timestamps and related attributes were not updating correctly after successful dbt Cloud job runs. This caused incorrect data to be displayed in dbt Explorer and via the Discovery API.
Impact: Users may have seen outdated "last updated" timestamps and stale execution metadata on their models in dbt Explorer and via the Discovery API, despite jobs completing successfully.
Resolution: At 11:30 AM ET on April 7, we deployed a fix to our metadata ingestion pipeline. Going forward, metadata will update correctly after each successful run.
Note: Models that were affected during this period will not be retroactively corrected. Running your dbt Cloud jobs again will refresh the metadata to the correct state.
We apologise for any disruption this caused to your workflows.
