[Bug] [connector-kafka] Kafka start_made handles exceptions

[CosmosNi]

Search before asking

• I had searched in the issues and found no similar issues.

What happened

In the scenario where the current topic has not been consumed, in batch mode, if start_mode is set to earliest, the job will consume from the beginning after restarting, rather than continuing to consume from the last consumed offset. If start_mode is set to latest, no data can be read at the beginning.

SeaTunnel Version

2.3.10

SeaTunnel Config

source {
  Kafka {
    topic = "test_topic_source"
    bootstrap.servers = "kafkaCluster:9092"
    start_mode = "earliest"
    format_error_handle_way = skip
  }
}

Running Command

./seatunnel.sh run

Error Exception

Repeated consumption

Zeta or Flink or Spark Version

No response

Java or Scala Version

No response

Screenshots

No response

Are you willing to submit PR?

• Yes I am willing to submit a PR!

Code of Conduct

• I agree to follow this project's Code of Conduct

[litiliu]

the doc says it support the group_offsets as start_mode, have you tried that? https://seatunnel.apache.org/docs/2.3.0/connector-v2/source/kafka/#start_mode