是否可以使用输入流中的时间戳在 Flink 中创建 Tumbling Windows。 如果是，那么如何？

Question

I have been trying to use Tumbling window on a Flink-Kafka consumer. Flink will be taking input from KafkaStream in Json format :

>{"eventGroup":"1121212" , "eventTime":"2020-01-02T13:43:43"}
>{"eventGroup":"1121212" , "eventTime":"2020-01-02T13:43:44"}
>{"eventGroup":"1121213" , "eventTime":"2020-01-02T13:43:48"}

I want to group this stream by eventGroup for all values of eventTime lying between say 2020-01-02T13:40:00 to 2020-01-02T13:44:59 . Is such functionality supported by flink Tumbling Windows ? If yes then a Sample code / procedure will be great help.

Thanks in Advance.

Answer 1

Your use case is very common. You want to add a timestamp extractor that makes the timestamp of your events known to Flink. If you use event time processing , your tumbling window will behave as you expect it.