Hello everyone, how are you? I have some doubts about the savepoint process and TableAPIs. The question is how

uid

are assigned when having sources declared using the TableAPI? Is there a way of forcing the

uid

? The context is that I have multiple Kafka sources declared inside my application and I want to reset some of them to earliest. I already changed the reset configuration to earliest but It looks like even if I change the reset configuration and the table name it still uses the savepoint information somehow.

If you’re starting from a savepoint, it will just continue where the application ended. It won’t suddenly restart from earliest. That would only change if you start it without a savepoint

I though that Flink used the table name as some sort of unique identifier for the table. But it seams that the only way is deleting the savepoint.

SQL/Table API is declarative, so an optimized plan and code are generated

The problem that i have is that some sources I want to reset and some I dont.

Indeed

Noob question: is there anyway of editing those offsets in the savepoint file? I know this looks like a workaround but would do the job.

You can try with the State Processor API, but it’s definitely not supported or documented

Many thanks @Martijn Visser!