更多内容详见:https://github.com/pierre94/flink-notes
主要是两种处理模式:
DataStream,DataStream → ConnectedStreams
连接两个保持他们类型的数据流,两个数据流被Connect之后,只是被放在了一个同一个流中,内部依然保持各自的数据和形式不发生任何变化,两个流相互独立。
Connect后使用CoProcessFunction、CoMap、CoFlatMap、KeyedCoProcessFunction等API 对两个流分别处理。如CoMap:
val warning = high.map( sensorData => (sensorData.id, sensorData.temperature) )
val connected = warning.connect(low)
val coMap = connected.map(
warningData => (warningData._1, warningData._2, "warning"),
lowData => (lowData.id, "healthy")
)
(ConnectedStreams → DataStream 功能与 map 一样,对 ConnectedStreams 中的每一个流分别进行 map 和 flatMap 处理。)
疑问,既然两个流内部独立,那Connect 后有什么意义呢?
Connect后的两条流可以共享状态,在对账等场景具有重大意义!
DataStream → DataStream:对两个或者两个以上的 DataStream 进行 union 操作,产生一个包含所有 DataStream 元素的新 DataStream。
val unionStream: DataStream[StartUpLog] = appStoreStream.union(otherStream) unionStream.print("union:::")
注意:Union 可以操作多个流,而Connect只能对两个流操作
Join是基于Connect更高层的一个实现,结合Window实现。
相关知识点比较多,详细文档见: https://ci.apache.org/projects/flink/flink-docs-release-1.10/dev/stream/operators/joining.html
有两个时间Event1、Event2,第一个字段是时间id,第二个字段是时间戳,需要对两者进行实时对账。当其中一个事件缺失、延迟时要告警出来。
类似之前的订单超时告警需求。之前数据源是一个流,我们在function里面进行一些改写。这里我们分别使用Event1和Event2两个流进行Connect处理。
// 事件1
case class Event1(id: Long, eventTime: Long)
// 事件2
case class Event2(id: Long, eventTime: Long)
// 输出结果
case class Result(id: Long, warnings: String)
scala实现
涉及知识点:
启动两个TCP服务:
nc -lh 9999
nc -lk 9998
注意:nc启动的是服务端、flink启动的是客户端
import java.text.SimpleDateFormat
import org.apache.flink.api.common.state.{ValueState, ValueStateDescriptor}
import org.apache.flink.streaming.api.TimeCharacteristic
import org.apache.flink.streaming.api.functions.co.KeyedCoProcessFunction
import org.apache.flink.streaming.api.scala.{StreamExecutionEnvironment, _}
import org.apache.flink.util.Collector
object CoTest {
val simpleDateFormat = new SimpleDateFormat("dd/MM/yyyy:HH:mm:ss")
val txErrorOutputTag = new OutputTag[Result]("txErrorOutputTag")
def main(args: Array[String]): Unit = {
val env = StreamExecutionEnvironment.getExecutionEnvironment
env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime)
env.setParallelism(1)
val event1Stream = env.socketTextStream("127.0.0.1", 9999)
.map(data => {
val dataArray = data.split(",")
Event1(dataArray(0).trim.toLong, simpleDateFormat.parse(dataArray(1).trim).getTime)
}).assignAscendingTimestamps(_.eventTime * 1000L)
.keyBy(_.id)
val event2Stream = env.socketTextStream("127.0.0.1", 9998)
.map(data => {
val dataArray = data.split(",")
Event2(dataArray(0).trim.toLong, simpleDateFormat.parse(dataArray(1).trim).getTime)
}).assignAscendingTimestamps(_.eventTime * 1000L)
.keyBy(_.id)
val coStream = event1Stream.connect(event2Stream)
.process(new CoTestProcess())
// union 必须是同一条类型的流
// val unionStream = event1Stream.union(event2Stream)
// unionStream.print()
coStream.print("ok")
coStream.getSideOutput(txErrorOutputTag).print("txError")
env.execute("union test")
}
//共享状态
class CoTestProcess() extends KeyedCoProcessFunction[Long,Event1, Event2, Result] {
lazy val event1State: ValueState[Boolean]
= getRuntimeContext.getState(new ValueStateDescriptor[Boolean]("event1-state", classOf[Boolean]))
lazy val event2State: ValueState[Boolean]
= getRuntimeContext.getState(new ValueStateDescriptor[Boolean]("event2-state", classOf[Boolean]))
override def processElement1(value: Event1, ctx: KeyedCoProcessFunction[Long, Event1, Event2, Result]#Context, out: Collector[Result]): Unit = {
if (event2State.value()) {
event2State.clear()
out.collect(Result(value.id, "ok"))
} else {
event1State.update(true)
//等待一分钟
ctx.timerService().registerEventTimeTimer(value.eventTime + 1000L * 60)
}
}
override def processElement2(value: Event2, ctx: KeyedCoProcessFunction[Long, Event1, Event2, Result]#Context, out: Collector[Result]): Unit = {
if (event1State.value()) {
event1State.clear()
out.collect(Result(value.id, "ok"))
} else {
event2State.update(true)
ctx.timerService().registerEventTimeTimer(value.eventTime + 1000L * 60)
}
}
override def onTimer(timestamp: Long, ctx: KeyedCoProcessFunction[Long, Event1, Event2, Result]#OnTimerContext, out: Collector[Result]): Unit = {
if(event1State.value()){
ctx.output(txErrorOutputTag,Result(ctx.getCurrentKey,s"no event2,timestamp:$timestamp"))
event1State.clear()
}else if(event2State.value()){
ctx.output(txErrorOutputTag,Result(ctx.getCurrentKey,s"no event1,timestamp:$timestamp"))
event2State.clear()
}
}
}
}
《Flink状态编程: 订单超时告警》:
https://blog.csdn.net/u013128262/article/details/104648592
《github:Flink学习笔记》:
https://github.com/pierre94/flink-notes
原创声明,本文系作者授权云+社区发表,未经许可,不得转载。
https://cloud.tencent.com/developer/article/1596145