A JDBC sink connector moves records from Kafka into a relational database through JDBC. It fits integrations where Kafka carries events and a database provides queryable or operational storage. section 11.10
Configure the connector around the failure and transformation behavior the integration requires:
- Set connection.attempts and connection.backoff.ms to control database reconnection attempts. The default behavior is three attempts separated by 10 seconds. section 11.6.2
- Use max.retries and retry.backoff.ms to retry failed sink writes, such as writes blocked by temporarily locked database entries. The default is 10 attempts with three-second intervals. section 11.6.2
- Combine connector retries with errors.retry.timeout and errors.retry.delay.max.ms for broader Kafka Connect error handling; route unprocessable sink records to a dead-letter queue. section 11.10 section 11.6.2
- Apply Single Message Transformations for simple changes such as renaming fields, masking data, or moving values to keys; reserve complex ETL for other mechanisms. section 11.10
Handling a temporary database outage
During database maintenance or a temporary network outage, extend the JDBC connector's connection retry window rather than relying on the default period. A connector that fails may require a manual restart before its tasks resume, while max.retries and retry.backoff.ms handle individual failed writes. section 11.6.2
