Please ignore this message. The issue was that a different timestamp
extractor was used when the kafka source was setup. That caused the issue.
On Tue, Jun 9, 2020 at 2:58 PM Yu Yang wrote:
> Hi,
>
>
> We implement a flink application that uses TumblingWindow, and uses even
> time as time characteristics. In the TumblingWindow's process function, we
> has the implementation below that checks whether the event's timestamp is
> in the tumbling window's timestamp range. We expected that all events
> shall be in the range. However, the application reports events with
> out-of-range timestamps. Any insights on how this happens?
>
>
> @Override
> public void process(EventStreamPartitionKey key,
> Context context, Iterable elements,
> Collector out) {
>
> for(Event event : elements) {
> if ( event.getTimestamp() >= context.window().getEnd() ||
>event.getTimestamp() < context.window().getStart() )
>
> System.out.println("NOT in RANGE: " + context.window().getStart()
>
> + ", " + event.getTimestamp() + ", " + context.window().getEnd());
> ...
>
> }
> out.collect(res);
> }
>
>
> Thanks!
>
>
> Regards,
>
> -Yu
>
