Hello guys, one question about kafka offset and savepoint. Can you confirm that I change kafka consumer group name of a flink job and start with the savepoint generated by old flink job. I can't re-use the offset information saved in that savepoint, right? Or we don't really keep kafka offsets at all in savepoint ?

Hi! Kafka offsets are stored in a savepoint. Consumer group name is pretty irrelevant, operator uid is the only important thing for recovery. So, you should be able to use the same savepoint with a different consumer group name as long as the operator uid hasn’t changed.

Hello @sap1ens, I am really not sure, I didn't change uuid of my kafka source operator. I have just created a savepoint in flink 16, and use it to start a new flink job with a new consumer group. But I see clearly it does not use the offset of old consumer group

Do you use Table API or DataStream API? If it’s the latter, did you define

uid

for the source?

I have used KafkaSource API https://nightlies.apache.org/flink/flink-docs-master/docs/connectors/datastream/kafka/ , I have defined uid for this source, the thing is , if I don't change consumer group, I think kafka consumer offsets will kept But when I change consumer group (ex: a consumer group not exist yet). My flink job can not set consumer offsets itself I think