Strengthen your Kafka Connect vocabulary covering source/sink connectors, SMTs, converters, and distributed vs standalone modes.
0 / 5 completed
1 / 5
During a PR review, a teammate asks whether to use a source connector or a sink connector to pull data from a database into Kafka. The correct choice is:
Source connectors read from external systems (databases, APIs) and produce records into Kafka. Sink connectors consume from Kafka and write to external destinations.
2 / 5
A Kafka Connect pipeline corrupts JSON records. In a post-mortem, you identify the issue as a misconfigured converter. What do converters do in Kafka Connect?
Converters (e.g. JsonConverter, AvroConverter) handle serialisation between Kafka Connect's SinkRecord/SourceRecord objects and the raw bytes in Kafka topics.
3 / 5
In a design review, you add a Single Message Transform (SMT) to mask a PII field before it reaches the sink. A colleague asks where SMTs run. The correct answer is:
SMTs run inside the Kafka Connect worker process, transforming records in the connector's pipeline between the source/sink and the Kafka topic — before the converter serialises the output.
4 / 5
A standup ticket: a connector task keeps failing and restarting. You check worker.properties and find errors.tolerance=none. Changing it to all means:
errors.tolerance=all lets the connector skip records that fail to process, logging them and optionally writing them to a dead-letter queue topic for later inspection.
5 / 5
In a platform review, the team asks about the difference between distributed and standalone Kafka Connect modes. The key distinction is:
Distributed mode runs multiple workers that coordinate via Kafka topics (config, offset, status), providing fault tolerance and horizontal scaling. Standalone is single-process with local file offsets.