I am experimenting with a Flink pipeline which I've configured to read change data events from PostgreSQL database on AWS RDS. The configuration works as expected for the most part, I am able to see change data with before and after properties set properly. Problem I am experiencing is that, if I force stop the Flink pipeline and start it again, it starts to read the entire table from start to end.
Also, I experimented with Debezium IO connector without Flink (using kafka connect) and it is behaving as expected, even if I force stop the connector itself, it is able to continue reading from where left off when I start it back up. Also I noticed in Debezium, it uses offset.storage with Kafka. Wish I could have the same in Flink Debezium.
My requirement from Flink Debezium connector is for it to be able to continue from the last place it read, appreciate it if you could advice me on this as I have pretty much exhausted all my options.
The link to Flink Debezium: https://ververica.github.io/flink-cdc-connectors/master/content/connectors/postgres-cdc.html
I looked in to possibility of getting offset.storage and config.storage to be configurable in Flink Debezium connector but I couldn't find a way to do that. It appeared to me that, the connector itself overrides what I try to set at Flink initialization time.