Flink 如何让数据延迟 10 分钟输出

可以使用 ProcessFunction 的 定时器实现
这里我们直接上代码
核心代码是，水位 + 10 分钟，然后注册定时器
在 OnTimer 方法里，时间到了的时候，就输出

package com.dsj361.flinksql

import java.sql.Timestamp
import java.util.Properties

import com.dsj361.common.enums.ModeEnum
import com.dsj361.common.lang.DateUtils
import com.dsj361.config.KafkaConfiguration
import org.apache.flink.formats.json.JsonNodeDeserializationSchema
import org.apache.flink.shaded.jackson2.com.fasterxml.jackson.databind.node.ObjectNode
import org.apache.flink.streaming.api.TimeCharacteristic
import org.apache.flink.streaming.api.functions.KeyedProcessFunction
import org.apache.flink.streaming.api.scala._
import org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumer010
import org.apache.flink.table.api.scala._
import org.apache.flink.table.functions.ScalarFunction
import org.apache.flink.util.Collector

/**
 *
 *
 * @author wangkai
 * @date 2019/12/13 10:40
 */
object FlinkSqlWindowTest5 extends App {

  class Utc2Local extends ScalarFunction {

    def eval(timestamp: Timestamp): Timestamp = {
      new Timestamp(timestamp.getTime + 28800000)
    }
  }

  case class LogData(userName:String,ts:Long,age:Int)


  val env = StreamExecutionEnvironment.getExecutionEnvironment
  env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime)

  val tEnv = StreamTableEnvironment.create(env)

  tEnv.registerFunction("Utc2Local", new Utc2Local)
  
  val kafkaConf = KafkaConfiguration.getConfig(ModeEnum.DEV)
  val prop: Properties = new Properties
  prop.put("bootstrap.servers", "192.168.17.26:9092,192.168.17.27:9092,192.168.17.28:9092")
  prop.put("group.id", "Flink_kafka_test_5")
  prop.put("enable.auto.commit", "true")
  prop.put("auto.commit.interval.ms", "100")
  val consumer = new FlinkKafkaConsumer010[ObjectNode]("flink-test-05", new JsonNodeDeserializationSchema, prop)

  class MyProcessFunction() extends KeyedProcessFunction[LogData,(LogData,Long),LogData] {
    override def processElement(value: (LogData, Long), ctx: KeyedProcessFunction[LogData, (LogData, Long), LogData]#Context, out: Collector[LogData]): Unit = {
      println("\n\n--------------process-------")
      val watermark = value._1.ts + 60000L
      ctx.timerService().registerEventTimeTimer(watermark)
    }

    override def onTimer(timestamp: Long, ctx: KeyedProcessFunction[LogData, (LogData, Long), LogData]#OnTimerContext, out: Collector[LogData]): Unit = {
      println("\n\n--------------onTimer-----------")
      out.collect(ctx.getCurrentKey)
    }
  }

  val ds = env.addSource(consumer)
    // 数据结构为：user_name:String, ts:TIMESTAMP
    .map(f => {
      (LogData (f.get("user_name").asText(), DateUtils.parseDateNewFormat(f.get("ts").asText()).getTime, f.get("age").asInt()),1L)
    })
    .assignAscendingTimestamps(f => f._1.ts)
    .keyBy(_._1)
    .process(new MyProcessFunction)
    .print()

  env.execute("job_name")


}