Which use case/requirement will be addressed by the proposed feature?

When a connector reads commit logs, it processes them sequentially and inserts them into event queues. Currently, if there are multiple queues, events are distributed among the queues based on the hash of the file name. This behavior can result in messages from subsequent files being inserted into the Kafka broker before messages from the current file, which is not desirable.

Implementation ideas (optional)

I propose to change the hashing strategy to use the value of partition columns to guarantee the order of events within the same partitions.

I will submit a PR myself for this issue.