What is the recommended practice to avoid duplicate rows when using the Kafka connector?

• A. Ensure each topic is processed by only one connector instance
• B. Run as many connectors as possible per topic
• C. Disable duplicate checks
• D. Use multiple topics per connector

Answer: (A)

The practice hinges on preventing concurrent consumers fromreading the same data. When a topic is assigned to only one connector instance, that topic’s partitions are handled by a single processing path, so each message is ingested once and pushed downstream without duplication. If multiple connectors were allowed to process the same topic, they could both pull the same records and forward duplicates, leading to repeated rows in the sink. Other options don’t solve the root cause: disabling duplicate checks hides duplicates, running many connectors per topic increases the chance of overlap, and using multiple topics per connector can still allow another connector to handle the same topic, reintroducing duplicates. So, ensuring each topic is processed by a single connector instance is the clean, effective way to avoid duplicate rows.