What causes INVALID_CURSOR_OPERATION?

Attempting to FETCH from a cursor after it was already CLOSEd. Cursor declared in one session or connection being referenced in another. SQL warehouse session timeout closing the cursor before the fetch loop completed. Application code not handling cursor lifecycle correctly across retries. Using cursor operations in a SQL warehouse that does not support session-persistent cursors

How do I fix INVALID_CURSOR_OPERATION?

Ensure the cursor lifecycle (DECLARE → OPEN → FETCH → CLOSE) is completed within a single session. Add error handling to detect and re-open cursors that were closed due to a session timeout. Increase the SQL warehouse idle timeout if cursors are timing out mid-fetch. For large result sets, consider using the Databricks statement execution REST API with pagination instead of cursors. Check that retry logic in application code re-declares and re-opens the cursor on retry rather than resuming from a closed state

Medium severityquery

Power BI Refresh Error:
INVALID_CURSOR_OPERATION

What does this error mean?

A cursor operation (FETCH, CLOSE, MOVE) was attempted on a cursor that is not open, has already been closed, or does not exist in the current session.

Common causes

1Attempting to FETCH from a cursor after it was already CLOSEd
2Cursor declared in one session or connection being referenced in another
3SQL warehouse session timeout closing the cursor before the fetch loop completed
4Application code not handling cursor lifecycle correctly across retries
5Using cursor operations in a SQL warehouse that does not support session-persistent cursors

How to fix it

1Ensure the cursor lifecycle (DECLARE → OPEN → FETCH → CLOSE) is completed within a single session
2Add error handling to detect and re-open cursors that were closed due to a session timeout
3Increase the SQL warehouse idle timeout if cursors are timing out mid-fetch
4For large result sets, consider using the Databricks statement execution REST API with pagination instead of cursors
5Check that retry logic in application code re-declares and re-opens the cursor on retry rather than resuming from a closed state

Frequently asked questions

Is cursor-based pagination recommended for SQL warehouses?

For large result sets, the Databricks statement execution API with `nextChunkIndex` pagination is more reliable than SQL cursor operations, which depend on session continuity.