What causes VECTOR_SEARCH_INDEX_NOT_FOUND?

The vector search index was deleted manually or as part of a Unity Catalog table cleanup. The index was created but the initial sync has not yet completed, so it is not yet queryable. The three-part name of the index (catalog.schema.index_name) has a typo or incorrect catalog prefix. The vector search endpoint hosting the index was deleted, taking all its indexes offline

How do I fix VECTOR_SEARCH_INDEX_NOT_FOUND?

Step 1: In the Databricks UI, navigate to Catalog Explorer and check whether the vector search index table exists.. Step 2: If the index is missing, recreate it using the Mosaic AI Vector Search API or the Catalog Explorer.. Step 3: After creation, wait for the index to complete its initial sync before querying — check status with the vector search index status API.. Step 4: Verify the vector search endpoint is running and healthy under Compute > Vector Search.. Step 5: Update application code to reference the correct index name with the full three-part Unity Catalog path.

High severityconfiguration

Power BI Refresh Error:
VECTOR_SEARCH_INDEX_NOT_FOUND

What does this error mean?

A query or application call referenced a Databricks Vector Search index that does not exist, was not yet synchronized, or was deleted from Unity Catalog.

Common causes

1The vector search index was deleted manually or as part of a Unity Catalog table cleanup
2The index was created but the initial sync has not yet completed, so it is not yet queryable
3The three-part name of the index (catalog.schema.index_name) has a typo or incorrect catalog prefix
4The vector search endpoint hosting the index was deleted, taking all its indexes offline

How to fix it

1Step 1: In the Databricks UI, navigate to Catalog Explorer and check whether the vector search index table exists.
2Step 2: If the index is missing, recreate it using the Mosaic AI Vector Search API or the Catalog Explorer.
3Step 3: After creation, wait for the index to complete its initial sync before querying — check status with the vector search index status API.
4Step 4: Verify the vector search endpoint is running and healthy under Compute > Vector Search.
5Step 5: Update application code to reference the correct index name with the full three-part Unity Catalog path.

Frequently asked questions

How do I check whether a vector search index sync is complete?

Use the Databricks SDK or REST API endpoint 'GET /api/2.0/vector-search/indexes/{index_name}' and check the status field. Statuses progress through PROVISIONING, ONLINE, and ONLINE_NO_PENDING_UPDATE.

Does deleting a Unity Catalog table also delete its vector search index?

Not automatically. The index is a separate Unity Catalog object. However, deleting the source Delta table makes the index stale and future syncs will fail. Drop the index explicitly before or after dropping the source table.