Relates to this discussion: https://debezium.zulipchat.com/#narrow/channel/350571-community-dbz-server/topic/Pub.2Fsub.20error.20handling

We are using Debezium Server with Pub/Sub running in Docker. We have a Docker Run command that we use to start up Debezium and we have retries set so that DBZ will start up again on failure. We've noticed the retry does not work when we encounter certain errors in the sink, in particular the timeout error. This is the error generated if this value is hit: debezium.sink.pubsub.wait.message.delivery.timeout.ms.

I believe this is being caused by DBZ not shutting down correctly in this situation

What Debezium connector do you use and what version?

I've reproduced what I believe to be the issue in Debezium Server 3.0.7

To test, I set debezium.sink.pubsub.wait.message.delivery.timeout.ms to a very low value in order to intentionally hit this value. You'll see from the attached container logs that there is a "java.lang.IllegalStateException","message":"Engine has been already shut down."

I can see also that after this exception there is still a lot of grpc logging being generated