This happens if the producer for DB history topic hangs upon send(), e.g. when it can't reach the Kafka server. In that case neither the snaphshot thread nor the producer thread will be shut down.