这个是因为你的over window的 order by字段不是具有时间属性的字段。 而之所以两个query串起来执行不成功，是因为第一个hop window之后，时间属性就丢掉了。 关于如何保持时间属性，可以参考文档[1]，具体到你这里，就是用HOP_ROWTIME或者HOP_PROCTIME， 具体用哪个取决于你的ts的类型。
[1] https://ci.apache.org/projects/flink/flink-docs-master/dev/table/sql/queries.html#time-attributes 111 <[email protected]> 于2020年4月7日周二 下午1:07写道： > HI, > 我这边想要在滑动窗口之后做一个topN，于是就有下面的SQL： > -- 窗口聚合 > create view window_v as > select > member_id, > category_id, > HOP_END(ts, INTERVAL '10' SECOND, INTERVAL '30' MINUTE) as `time`, > count(1) as c > from stream_t > group by member_id, category_id, > HOP(ts, INTERVAL '10' SECOND, INTERVAL '30' MINUTE); > > > -- 排序取top 20 > SELECT > member_id, > category_id, > c, > row_number() over (PARTITION BY member_id ORDER BY c) AS rn > FROM window_v; > > > 但是执行异常： > Caused by: org.apache.flink.table.api.TableException: OVER windows' > ordering in stream mode must be defined on a time attribute. at > org.apache.flink.table.planner.plan.nodes.physical.stream.StreamExecOverAggregate.translateToPlanInternal(StreamExecOverAggregate.scala:195) > at > org.apache.flink.table.planner.plan.nodes.physical.stream.StreamExecOverAggregate.translateToPlanInternal(StreamExecOverAggregate.scala:56) > at > org.apache.flink.table.planner.plan.nodes.exec.ExecNode$class.translateToPlan(ExecNode.scala:58) > 我看官网的TOPN例子中，order by 后面也可以是long型。难道row_number不能作用在窗口结果上？ > Best,Xinghalo -- Benchao Li School of Electronics Engineering and Computer Science, Peking University Tel:+86-15650713730 Email: [email protected]; [email protected]
