Hey Flink Community, I am using FlinkSQL Over Aggregation <https://nightlies.apache.org/flink/flink-docs-release-1.14/docs/dev/table/sql/queries/over-agg/> to calculate the number of uuid per client ip during the past 1 hour. The flink sql I am using is something like below: SELECT COUNT(DISTINCT consumer_consumerUuid) OVER w AS feature_value, clientIp as entity_id FROM wide_table WINDOW w AS ( PARTITION BY clientIp ORDER BY ts RANGE BETWEEN INTERVAL '1' HOUR PRECEDING AND CURRENT ROW ) ​From the documentation, we know that the OVER aggregates produce an aggregated value for every input row, which means (in my view) the calculation is triggered by every input event in wide_table not by watermark? However, seeing from my logs, there is always about a 5-60 seconds' delay between the input row and the result calculated by window.
The data volume is small, there are only about 1k records/hour in table wide_table and less than 10 consumer for each clientIp. Is it normal with this delay? Or there is something wrong with the way it is used ? Thanks.
