Bug report

For bug reports, provide this information, please:

What Debezium connector do you use and what version?

PostgreSQL connector. Version 2.5.4-Final

What is the captured database version and mode of deployment?

Local, docker container database

What behaviour do you expect?

When the EmbeddedEngine is shutdown, the KafkaOffsetBackingStore should be shutdown as well.

What behaviour do you see?

The producer and consumer threads of the KafkaOffsetBackingStore is not shut down and show that they're still trying to connect to Kafka brokers (which have already been shutdown, after a test concluded).

Do you see the same behaviour using the latest relesead Debezium version?

Yes, I am on 2.5.4-Final

More Information

I have tracked the code down and here are the details:

1. In the main loop, run method of the EmbeddedEngine, we initialise the offset store using the initializeOffsetStore method.
2. In that method, we call KafkaConnectUtil.kafkaOffsetBackingStore to create a backingstore object. 
3. Prior to creating the KafkaOffsetBackingStore, we first create the SharedTopicAdmin instance and pass it to the KafkaOffsetBackingStore. 
4. During the shutdown sequence, the KafkaOffsetBackingStore only shutdown the topic admin if it's an admin that the object initialise. However, in our case, it wasn't. 
5. This leaves the SharedTopicAdmin instance daggling