Hi, Dennis You should send this question to mailbox u...@flink.apache.org, dev is used to discuss development-related issues.
Back to your question: - AFAIK, uid will be set up as a random value if not defined. How can I find the current uid? Regarding the function of uid, you can see [1] and [2] for more detail. The framework generates UIDs automatically if they are not set manually, it is a random value, and it is maybe changed if you modify the job. There is currently no way to get the UIDs for each operator directly. - What will happen if I set up a new uid in the code above? If you set a new uid, and restore from savepoint, the restart may fail because the old UID can't be found because Kafka source is stateful - When I want to migrate 'FlinkKafkaConsumer' to 'KafkaSource', is it okay not to set up 'uid'? I thinks it is okay. [1] https://github.com/apache/flink/blob/dfb9cb851dc1f0908ea6c3ce1230dd8ca2b48733/flink-core/src/main/java/org/apache/flink/configuration/PipelineOptions.java#L70 [2] https://github.com/apache/flink/blob/dfb9cb851dc1f0908ea6c3ce1230dd8ca2b48733/flink-core/src/main/java/org/apache/flink/api/dag/Transformation.java#L168 Best, Ron Dennis Jung <inylov...@gmail.com> 于2023年8月8日周二 10:44写道： > Hello people, > As suggested in the following, it seems it is recommended to set 'uid' for > operators. > > > https://ververica.zendesk.com/hc/en-us/articles/360010248879-Should-I-call-uid-after-addSource-or-addSink- > > Currently there is an existing personal project which uses > 'FlinkKafkaConsumer'(it is a deprecated feature) to consume data from > Kafka, which 'uid' is not configured. > > ''' > ... > FlinkKafkaConsumer kc = ... ; > DataStream<T> ds = env.addSource(kc, "flink-kafka-source"); > ''' > > - AFAIK, uid will be set up as a random value if not defined. How can I > find the current uid? > - What will happen if I set up a new uid in the code above? > - When I want to migrate 'FlinkKafkaConsumer' to 'KafkaSource', is it okay > not to set up 'uid'? > > Thanks. >
