Choose a Bad Row Handling Policy

Bad rows need policy, not panic. Fail fast Fail the whole load when one invalid row makes the dataset unsafe, such as a financial file that must reconcile exactly or a schema break that means columns shifted. Repair or reject Repair only when the rule is deterministic and approved, such as trimming whitespace or mapping a known code. Reject and continue when valid rows can still be trusted and invalid rows need source-owner action. Quarantine with evidence A quarantine table should keep rejected rows, rule names, run IDs, and owners. That evidence lets the source fix the defect without silently…

Sign up free — one personalized lesson every day, matched to your role and goals.

Already have an account? Sign in