我的電腦是8線程,當我運行下面這段代碼時,可以輸出結果,但是當把並行度註釋掉時,就不輸出結果了,這是爲什麼呢?輸入數據一樣,watermark應該都是一樣的啊
輸入數據爲:
1585721697000,xiao,8
1585721700000,xiao,10
1585721705000,xiao,4
1585721715000,xiao,9
case class Line(id:Long,name:String,age:Int) object EventTimeWindow { def main(args: Array[String]): Unit = { val env = StreamExecutionEnvironment.getExecutionEnvironment // env.setParallelism(1) //指定時間類型 env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime) val value: DataStream[String] = env.socketTextStream("localhost", 9999) //指定wotermark 分配器 val value1: DataStream[String] = value.assignTimestampsAndWatermarks(new BoundedOutOfOrdernessTimestampExtractor[String](Time.seconds(3)) { override def extractTimestamp(t: String): Long = { t.split(",")(0).trim.toLong } }) val value5: DataStream[Line] = value1.map(line => { val arr: Array[String] = line.split(",") Line(arr(0).toLong, arr(1), arr(2).toInt) }) //根據name分組 val value2: KeyedStream[Line, String] = value5.keyBy(_.name) //10秒一個滾動窗口 val value3: WindowedStream[Line, String, TimeWindow] = value2.timeWindow(Time.seconds(10)) val value4: DataStream[Line] = value3.sum(2) value4.print("sum age") env.execute() } }
最後確定,當從socket讀入數據後,不能先進行map操作,而是先分配時間戳和watermark,這樣就能正確輸出結果了。