Hi all, I have a streaming job that essentially looks like this: KafkaSource -> Map -> EventTimeWindow -> RichFlatMap -> CustomSink. The RichFlatMap part does some heavy lifting in open(), so that the open() call blocks for several minutes. I assumed that until open() returns the backpressure mechanism would slow down the entire upstream up to the KafkaSource, so that no new records would be emitted to the pipeline until the RichFlatMap is ready. What I actually observe is that the KafkaSource, Map and EventTimeWindow continue processing - the in/out records, in/out MB counters keep increasing. The RichFlatMap and its downstream CustomSink have 0 as expected, until the RichFlatMap is actually done with open(). The backpressure monitor in Flink UI shows "OK" for all operators.
Why doesn't backpressure mechanism work in this case? Thanks, Yury
