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 <yuyan...@gmail.com> 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<Event> elements,
> Collector<EventResult> 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
>

Reply via email to