Flink-1.5.0-源码分析系列10：接收任务

最新推荐文章于 2022-07-29 21:43:11 发布
最新推荐文章于 2022-07-29 21:43:11 发布
阅读量439 收藏
点赞数
文章标签： runtime java netty
原文链接：https://my.oschina.net/qiangzigege/blog/2208746
版权

2019独角兽企业重金招聘Python工程师标准>>> hot3.png

下面我们看TM是如何接收任务并处理的！

jdb -sourcepath /root/kafka-clients-0.10.2.1-sources:/root/flink-1.5.0-src/flink-connectors/flink-connector-kafka-0.10/src/main/java:/root/flink-1.5.0-src/flink-connectors/flink-connector-kafka-0.9/src/main/java:/root/flink-1.5.0-src/flink-connectors/flink-connector-kafka-base/src/main/java:/root/flink-1.5.0-src/flink-runtime/src/main/java:/root/flink-release-1.5.0/flink-release-1.5.0/flink-connectors/flink-connector-kafka-base:/root/flink-release-1.5.0/flink-release-1.5.0/flink-core/src/main/java:/root/flink-release-1.5.0/flink-release-1.5.0/flink-streaming-java/src/main/java:/root/flink-release-1.5.0/flink-connectors/flink-connector-kafka-0.10/src/main/java:/root/flink-release-1.5.0/flink-connectors/flink-connector-kafka-base/src/main/java:/root/flink-release-1.5.0/flink-streaming-java/src/main/java:/root/yarn_study/hadoop-2.2.0-src/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-ct/src/main/java:/root/yarn_study/hadoop-2.2.0-src/hadoop-common-project/hadoop-common/src/main/java:/root/yarn_study/hadoop-2.2.0-src/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-client/src/main/java:/root/yarn_study/hadoop-2.2.0-src/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-client/src/main/java:/root/flink-release-1.5.0/flink-runtime-web/src/main/java:/root/flink-release-1.5.0/flink-core/src/main/java:/root/yarn_study/hadoop-2.2.0-src/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java:/root/flink-release-1.5.0/flink-yarn/src/main/java:/root/flink-release-1.5.0/flink-runtime/src/main/java -attach localhost:10000

会看到

eaee684e474d0dd690ea97db9d5b5edeea7.jpg

这里的调用栈就是

 [1] org.apache.flink.runtime.deployment.TaskDeploymentDescriptor.getTargetSlotNumber (TaskDeploymentDescriptor.java:256)
  [2] org.apache.flink.runtime.taskmanager.TaskManager.submitTask (TaskManager.scala:1,141)
  [3] org.apache.flink.runtime.taskmanager.TaskManager.org$apache$flink$runtime$taskmanager$TaskManager$$handleTaskMessage (TaskManager.scala:457)
  [4] org.apache.flink.runtime.taskmanager.TaskManager$$anonfun$handleMessage$1.applyOrElse (TaskManager.scala:279)
  [5] scala.runtime.AbstractPartialFunction.apply (AbstractPartialFunction.scala:36)
  [6] org.apache.flink.runtime.LeaderSessionMessageFilter$$anonfun$receive$1.applyOrElse (LeaderSessionMessageFilter.scala:38)
  [7] scala.runtime.AbstractPartialFunction.apply (AbstractPartialFunction.scala:36)
  [8] org.apache.flink.runtime.LogMessages$$anon$1.apply (LogMessages.scala:33)
  [9] org.apache.flink.runtime.LogMessages$$anon$1.apply (LogMessages.scala:28)
  [10] scala.PartialFunction$class.applyOrElse (PartialFunction.scala:123)
  [11] org.apache.flink.runtime.LogMessages$$anon$1.applyOrElse (LogMessages.scala:28)
  [12] akka.actor.Actor$class.aroundReceive (Actor.scala:502)
  [13] org.apache.flink.runtime.taskmanager.TaskManager.aroundReceive (TaskManager.scala:122)
  [14] akka.actor.ActorCell.receiveMessage (ActorCell.scala:526)
  [15] akka.actor.ActorCell.invoke (ActorCell.scala:495)
  [16] akka.dispatch.Mailbox.processMailbox (Mailbox.scala:257)
  [17] akka.dispatch.Mailbox.run (Mailbox.scala:224)
  [18] akka.dispatch.Mailbox.exec (Mailbox.scala:234)
  [19] scala.concurrent.forkjoin.ForkJoinTask.doExec (ForkJoinTask.java:260)
  [20] scala.concurrent.forkjoin.ForkJoinPool$WorkQueue.runTask (ForkJoinPool.java:1,339)
  [21] scala.concurrent.forkjoin.ForkJoinPool.runWorker (ForkJoinPool.java:1,979)
  [22] scala.concurrent.forkjoin.ForkJoinWorkerThread.run (ForkJoinWorkerThread.java:107)

所以就是在这里接收任务的，

 

 

 

我们打上断点

stop in org.apache.flink.runtime.deployment.TaskDeploymentDescriptor.getTargetSlotNumber

stop in org.apache.flink.runtime.taskmanager.Task.run

stop at org.apache.flink.runtime.taskmanager.Task:691

stop in org.apache.flink.streaming.runtime.tasks.StreamTask.invoke

stop at org.apache.flink.streaming.runtime.tasks.StreamTask:261

stop in org.apache.flink.streaming.runtime.tasks.StreamTask.openAllOperators

stop in org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumerBase.open

stop at org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumer010:220

stop at org.apache.flink.streaming.connectors.kafka.internal.Kafka09PartitionDiscoverer:58

stop at org.apache.kafka.clients.consumer.KafkaConsumer:621

stop at org.apache.flink.streaming.connectors.kafka.internal.Kafka09Fetcher:154

stop in org.apache.flink.streaming.connectors.kafka.internal.Kafka010Fetcher.emitRecord

stop at org.apache.flink.streaming.runtime.tasks.OperatorChain:555

stop in org.apache.flink.table.runtime.CRowOutputProcessRunner.processElement

stop at org.apache.flink.streaming.api.operators.ProcessOperator:66

stop at org.apache.flink.streaming.runtime.partitioner.KeyGroupStreamPartitioner:43
stop at org.apache.flink.streaming.runtime.partitioner.KeyGroupStreamPartitioner:59

stop at org.apache.flink.runtime.io.network.api.writer.RecordWriter:148

 

注意：真实取数据是通过KafkaConsumerThread来的，中转，然后Kafka09Fetcher.runFetchLoop会从这个中转拿到数据！

这个通过下面的调用栈可见一斑

  [1] org.apache.flink.streaming.runtime.streamrecord.StreamRecord.getValue (StreamRecord.java:67)
  [2] org.apache.flink.streaming.api.operators.ProcessOperator.processElement (ProcessOperator.java:66)
  [3] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator (OperatorChain.java:560)
  [4] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:535)
  [5] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:515)
  [6] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [7] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [8] org.apache.flink.streaming.api.operators.StreamSourceContexts$ManualWatermarkContext.processAndCollectWithTimestamp (StreamSourceContexts.java:310)
  [9] org.apache.flink.streaming.api.operators.StreamSourceContexts$WatermarkContext.collectWithTimestamp (StreamSourceContexts.java:409)
  [10] org.apache.flink.streaming.connectors.kafka.internals.AbstractFetcher.emitRecordWithTimestamp (AbstractFetcher.java:398)
  [11] org.apache.flink.streaming.connectors.kafka.internal.Kafka010Fetcher.emitRecord (Kafka010Fetcher.java:89)
  [12] org.apache.flink.streaming.connectors.kafka.internal.Kafka09Fetcher.runFetchLoop (Kafka09Fetcher.java:154)
  [13] org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumerBase.run (FlinkKafkaConsumerBase.java:721)
  [14] org.apache.flink.streaming.api.operators.StreamSource.run (StreamSource.java:87)
  [15] org.apache.flink.streaming.api.operators.StreamSource.run (StreamSource.java:56)
  [16] org.apache.flink.streaming.runtime.tasks.SourceStreamTask.run (SourceStreamTask.java:99)
  [17] org.apache.flink.streaming.runtime.tasks.StreamTask.invoke (StreamTask.java:306)
  [18] org.apache.flink.runtime.taskmanager.Task.run (Task.java:703)

然后，郁闷的是，debug一段scala代码后，跟丢了，scala代码真不好，搞什么scala，擦， 

回家想了下，翻回客户端的debug笔记，下一段应该是KeyGroupStreamPartitioner的selectChannels方法

于是打断点

stop at org.apache.flink.streaming.runtime.partitioner.KeyGroupStreamPartitioner:43
stop at org.apache.flink.streaming.runtime.partitioner.KeyGroupStreamPartitioner:59

我们来看看是否可以hit breakpoint,果然可以命中

Source: MyKafka010JsonTableSource -> from: (metric, time, value, nested) -> where: (AND(IS NOT NULL(metric), IS NOT NULL(value))), select: (metric, time, nested.number AS $f2, CAST(value) AS value, value AS value0, PROCTIME(time) AS $f5) (1/4)[1] where
  [1] org.apache.flink.streaming.runtime.partitioner.KeyGroupStreamPartitioner.selectChannels (KeyGroupStreamPartitioner.java:59)
  [2] org.apache.flink.streaming.runtime.partitioner.KeyGroupStreamPartitioner.selectChannels (KeyGroupStreamPartitioner.java:32)
  [3] org.apache.flink.runtime.io.network.api.writer.RecordWriter.emit (RecordWriter.java:104)
  [4] org.apache.flink.streaming.runtime.io.StreamRecordWriter.emit (StreamRecordWriter.java:81)
  [5] org.apache.flink.streaming.runtime.io.RecordWriterOutput.pushToRecordWriter (RecordWriterOutput.java:107)
  [6] org.apache.flink.streaming.runtime.io.RecordWriterOutput.collect (RecordWriterOutput.java:89)
  [7] org.apache.flink.streaming.runtime.io.RecordWriterOutput.collect (RecordWriterOutput.java:45)
  [8] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [9] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [10] org.apache.flink.streaming.api.operators.TimestampedCollector.collect (TimestampedCollector.java:51)
  [11] org.apache.flink.table.runtime.CRowWrappingCollector.collect (CRowWrappingCollector.scala:37)
  [12] org.apache.flink.table.runtime.CRowWrappingCollector.collect (CRowWrappingCollector.scala:28)
  [13] DataStreamCalcRule$32.processElement (null)
  [14] org.apache.flink.table.runtime.CRowProcessRunner.processElement (CRowProcessRunner.scala:66)
  [15] org.apache.flink.table.runtime.CRowProcessRunner.processElement (CRowProcessRunner.scala:35)
  [16] org.apache.flink.streaming.api.operators.ProcessOperator.processElement (ProcessOperator.java:66)
  [17] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator (OperatorChain.java:560)
  [18] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:535)
  [19] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:515)
  [20] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [21] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [22] org.apache.flink.streaming.api.operators.TimestampedCollector.collect (TimestampedCollector.java:51)
  [23] org.apache.flink.table.runtime.CRowWrappingCollector.collect (CRowWrappingCollector.scala:37)
  [24] org.apache.flink.table.runtime.CRowWrappingCollector.collect (CRowWrappingCollector.scala:28)
  [25] DataStreamSourceConversion$8.processElement (null)
  [26] org.apache.flink.table.runtime.CRowOutputProcessRunner.processElement (CRowOutputProcessRunner.scala:67)
  [27] org.apache.flink.streaming.api.operators.ProcessOperator.processElement (ProcessOperator.java:66)
  [28] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator (OperatorChain.java:560)
  [29] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:535)
  [30] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:515)
  [31] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [32] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [33] org.apache.flink.streaming.api.operators.StreamSourceContexts$ManualWatermarkContext.processAndCollectWithTimestamp (StreamSourceContexts.java:310)
  [34] org.apache.flink.streaming.api.operators.StreamSourceContexts$WatermarkContext.collectWithTimestamp (StreamSourceContexts.java:409)
  [35] org.apache.flink.streaming.connectors.kafka.internals.AbstractFetcher.emitRecordWithTimestamp (AbstractFetcher.java:398)
  [36] org.apache.flink.streaming.connectors.kafka.internal.Kafka010Fetcher.emitRecord (Kafka010Fetcher.java:89)
  [37] org.apache.flink.streaming.connectors.kafka.internal.Kafka09Fetcher.runFetchLoop (Kafka09Fetcher.java:154)
  [38] org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumerBase.run (FlinkKafkaConsumerBase.java:721)
  [39] org.apache.flink.streaming.api.operators.StreamSource.run (StreamSource.java:87)
  [40] org.apache.flink.streaming.api.operators.StreamSource.run (StreamSource.java:56)
  [41] org.apache.flink.streaming.runtime.tasks.SourceStreamTask.run (SourceStreamTask.java:99)
  [42] org.apache.flink.streaming.runtime.tasks.StreamTask.invoke (StreamTask.java:306)
  [43] org.apache.flink.runtime.taskmanager.Task.run (Task.java:703)
  [44] java.lang.Thread.run (Thread.java:748)

然后发现一个RecordWriter的一个缓存问题，flush相关，感觉数据不会立刻写，很奇怪，抓到调用栈如下

)[1] where
  [1] org.apache.flink.runtime.io.network.api.writer.RecordWriter.<init> (RecordWriter.java:84)
  [2] org.apache.flink.streaming.runtime.io.StreamRecordWriter.<init> (StreamRecordWriter.java:58)
  [3] org.apache.flink.streaming.runtime.tasks.StreamTask.createStreamRecordWriter (StreamTask.java:1,202)
  [4] org.apache.flink.streaming.runtime.tasks.StreamTask.createStreamRecordWriters (StreamTask.java:1,171)
  [5] org.apache.flink.streaming.runtime.tasks.StreamTask.<init> (StreamTask.java:214)
  [6] org.apache.flink.streaming.runtime.tasks.StreamTask.<init> (StreamTask.java:193)
  [7] org.apache.flink.streaming.runtime.tasks.SourceStreamTask.<init> (SourceStreamTask.java:51)
  [8] sun.reflect.NativeConstructorAccessorImpl.newInstance0 (native method)
  [9] sun.reflect.NativeConstructorAccessorImpl.newInstance (NativeConstructorAccessorImpl.java:62)
  [10] sun.reflect.DelegatingConstructorAccessorImpl.newInstance (DelegatingConstructorAccessorImpl.java:45)
  [11] java.lang.reflect.Constructor.newInstance (Constructor.java:423)
  [12] org.apache.flink.runtime.taskmanager.Task.loadAndInstantiateInvokable (Task.java:1,445)
  [13] org.apache.flink.runtime.taskmanager.Task.run (Task.java:680)
  [14] java.lang.Thread.run (Thread.java:748)

然后跟踪到chainConfigs

chainedConfigs = "{1=
=======================Stream Config=======================
Number of non-chained inputs: 0
Number of non-chained outputs: 0
Output names: []
Partitioning:
Chained subtasks: [(Source: MyKafka010JsonTableSource-1 -> from: (metric, time, value, nested)-2, typeNumber=0, selectedNames=[], outputPartitioner=FORWARD, outputTag=null)]
Operator: StreamSource
Buffer timeout: 100
State Monitoring: true


---------------------
Chained task configs
---------------------
{2=
=======================Stream Config=======================
Number of non-chained inputs: 0
Number of non-chained outputs: 0
Output names: []
Partitioning:
Chained subtasks: [(from: (metric, time, value, nested)-2 -> where: (AND(IS NOT NULL(metric), IS NOT NULL(value))), select: (metric, time, nested.number AS $f2, CAST(value) AS value, value AS value0, PROCTIME(time) AS $f5)-3, typeNumber=0, selectedNames=[], outputPartitioner=FORWARD, outputTag=null)]
Operator: ProcessOperator
Buffer timeout: 100
State Monitoring: true, 3=
=======================Stream Config=======================
Number of non-chained inputs: 0
Number of non-chained outputs: 1
Output names: [(where: (AND(IS NOT NULL(metric), IS NOT NULL(value))), select: (metric, time, nested.number AS $f2, CAST(value) AS value, value AS value0, PROCTIME(time) AS $f5)-3 -> groupBy: (metric), window: (TumblingGroupWindow('w$, 'time, 3000.millis)), select: (metric, SUM($f2) AS nestedNumber, SUM(value) AS sum, COUNT(value0) AS count, AVG(value) AS avg, MAX(value) AS max, MIN(value) AS min, MAX($f5) AS time)-5, typeNumber=0, selectedNames=[], outputPartitioner=HASH, outputTag=null)]
Partitioning:
	5: HASH
Chained subtasks: []
Operator: ProcessOperator
Buffer timeout: 100
State Monitoring: true}, 2=
=======================Stream Config=======================
Number of non-chained inputs: 0
Number of non-chained outputs: 0
Output names: []
Partitioning:
Chained subtasks: [(from: (metric, time, value, nested)-2 -> where: (AND(IS NOT NULL(metric), IS NOT NULL(value))), select: (metric, time, nested.number AS $f2, CAST(value) AS value, value AS value0, PROCTIME(time) AS $f5)-3, typeNumber=0, selectedNames=[], outputPartitioner=FORWARD, outputTag=null)]
Operator: ProcessOperator
Buffer timeout: 100
State Monitoring: true, 3=
=======================Stream Config=======================
Number of non-chained inputs: 0
Number of non-chained outputs: 1
Output names: [(where: (AND(IS NOT NULL(metric), IS NOT NULL(value))), select: (metric, time, nested.number AS $f2, CAST(value) AS value, value AS value0, PROCTIME(time) AS $f5)-3 -> groupBy: (metric), window: (TumblingGroupWindow('w$, 'time, 3000.millis)), select: (metric, SUM($f2) AS nestedNumber, SUM(value) AS sum, COUNT(value0) AS count, AVG(value) AS avg, MAX(value) AS max, MIN(value) AS min, MAX($f5) AS time)-5, typeNumber=0, selectedNames=[], outputPartitioner=HASH, outputTag=null)]
Partitioning:
	5: HASH
Chained subtasks: []
Operator: ProcessOperator
Buffer timeout: 100
State Monitoring: true}"
Source: MyKafka010JsonTableSource -> from: (metric, time, value, nested) -> where: (AND(IS NOT NULL(metric), IS NOT NULL(value))), select: (metric, time, nested.number AS $f2, CAST(value) AS value, value AS value0, PROCTIME(time) AS $f5) (1/4)[1]

断点stop at org.apache.flink.streaming.runtime.tasks.StreamTask:213

看到的是

taskConfiguration = "{checkpointing=true, vertexID=1, outStreamEdges=[B@1938463b, nonChainedOutputs=[B@53b0d4e0, userFunction=org.apache.flink.streaming.api.operators.StreamSource, numberOfOutputs=0, operatorName=Source: MyKafka010JsonTableSource, timechar=2, statekeyser=[B@6879c7d9, serializedUDF=[B@59eea41a, statePartitioner0=[B@23c2882, statePartitioner1=[B@57feebf1, typeSerializer_in_2=[B@719fb9d8, typeSerializer_in_1=[B@5dfd1d85, typeSerializer_out=[B@3e665680, bufferTimeout=100, chainedTaskConfig_=[B@1ee783de, chainIndex=0, chainedOutputs=[B@12bab8a3, edgesInOrder=[B@271e8e44, outputSelectorWrapper=[B@2b400e71, isChainedSubtask=true, checkpointMode=0, operatorID=[B@91b91b4}"
Source: MyKafka010JsonTableSource -> from: (metric, time, value, nested) -> where: (AND(IS NOT NULL(metric), IS NOT NULL(value))), select: (metric, time, nested.number AS $f2, CAST(value) AS value, value AS value0, PROCTIME(time) AS $f5) (3/4)[1]

这个其实很简单，java代码里设置env就行了！

---继续debug


[1] org.apache.flink.runtime.io.network.netty.PartitionRequestQueue.notifyReaderNonEmpty (PartitionRequestQueue.java:92)
  [2] org.apache.flink.runtime.io.network.netty.CreditBasedSequenceNumberingViewReader.notifyDataAvailable (CreditBasedSequenceNumberingViewReader.java:197)
  [3] org.apache.flink.runtime.io.network.partition.PipelinedSubpartitionView.notifyDataAvailable (PipelinedSubpartitionView.java:56)
  [4] org.apache.flink.runtime.io.network.partition.PipelinedSubpartition.notifyDataAvailable (PipelinedSubpartition.java:290)
  [5] org.apache.flink.runtime.io.network.partition.PipelinedSubpartition.flush (PipelinedSubpartition.java:76)
  [6] org.apache.flink.runtime.io.network.partition.ResultPartition.flush (ResultPartition.java:269)
  [7] org.apache.flink.runtime.io.network.api.writer.RecordWriter.sendToTarget (RecordWriter.java:149)
  [8] org.apache.flink.runtime.io.network.api.writer.RecordWriter.emit (RecordWriter.java:105)
  [9] org.apache.flink.streaming.runtime.io.StreamRecordWriter.emit (StreamRecordWriter.java:81)
  [10] org.apache.flink.streaming.runtime.io.RecordWriterOutput.pushToRecordWriter (RecordWriterOutput.java:107)
  [11] org.apache.flink.streaming.runtime.io.RecordWriterOutput.collect (RecordWriterOutput.java:89)
  [12] org.apache.flink.streaming.runtime.io.RecordWriterOutput.collect (RecordWriterOutput.java:45)
  [13] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [14] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [15] org.apache.flink.streaming.api.operators.TimestampedCollector.collect (TimestampedCollector.java:51)
  [16] org.apache.flink.table.runtime.RowtimeProcessFunction.processElement (RowtimeProcessFunction.scala:45)
  [17] org.apache.flink.table.runtime.RowtimeProcessFunction.processElement (RowtimeProcessFunction.scala:32)
  [18] org.apache.flink.streaming.api.operators.ProcessOperator.processElement (ProcessOperator.java:66)
  [19] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator (OperatorChain.java:560)
  [20] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:535)
  [21] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:515)
  [22] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [23] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [24] org.apache.flink.streaming.api.operators.TimestampedCollector.collect (TimestampedCollector.java:51)
  [25] org.apache.flink.table.runtime.CRowWrappingCollector.collect (CRowWrappingCollector.scala:37)
  [26] org.apache.flink.table.runtime.CRowWrappingCollector.collect (CRowWrappingCollector.scala:28)
  [27] DataStreamCalcRule$32.processElement (null)
  [28] org.apache.flink.table.runtime.CRowProcessRunner.processElement (CRowProcessRunner.scala:66)
  [29] org.apache.flink.table.runtime.CRowProcessRunner.processElement (CRowProcessRunner.scala:35)
  [30] org.apache.flink.streaming.api.operators.ProcessOperator.processElement (ProcessOperator.java:66)
  [31] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator (OperatorChain.java:560)
  [32] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:535)
  [33] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:515)
  [34] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [35] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [36] org.apache.flink.streaming.runtime.operators.TimestampsAndPeriodicWatermarksOperator.processElement (TimestampsAndPeriodicWatermarksOperator.java:67)
  [37] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator (OperatorChain.java:560)
  [38] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:535)
  [39] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:515)
  [40] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [41] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [42] org.apache.flink.streaming.api.operators.TimestampedCollector.collect (TimestampedCollector.java:51)
  [43] org.apache.flink.table.runtime.CRowWrappingCollector.collect (CRowWrappingCollector.scala:37)
  [44] org.apache.flink.table.runtime.CRowWrappingCollector.collect (CRowWrappingCollector.scala:28)
  [45] DataStreamSourceConversion$9.processElement (null)
  [46] org.apache.flink.table.runtime.CRowOutputProcessRunner.processElement (CRowOutputProcessRunner.scala:67)
  [47] org.apache.flink.streaming.api.operators.ProcessOperator.processElement (ProcessOperator.java:66)
  [48] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator (OperatorChain.java:560)
  [49] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:535)
  [50] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:515)
  [51] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [52] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [53] org.apache.flink.streaming.api.operators.StreamSourceContexts$ManualWatermarkContext.processAndCollectWithTimestamp (StreamSourceContexts.java:310)
  [54] org.apache.flink.streaming.api.operators.StreamSourceContexts$WatermarkContext.collectWithTimestamp (StreamSourceContexts.java:409)
  [55] org.apache.flink.streaming.connectors.kafka.internals.AbstractFetcher.emitRecordWithTimestamp (AbstractFetcher.java:398)
  [56] org.apache.flink.streaming.connectors.kafka.internal.Kafka010Fetcher.emitRecord (Kafka010Fetcher.java:89)
  [57] org.apache.flink.streaming.connectors.kafka.internal.Kafka09Fetcher.runFetchLoop (Kafka09Fetcher.java:154)
  [58] org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumerBase.run (FlinkKafkaConsumerBase.java:721)
  [59] org.apache.flink.streaming.api.operators.StreamSource.run (StreamSource.java:87)
  [60] org.apache.flink.streaming.api.operators.StreamSource.run (StreamSource.java:56)
  [61] org.apache.flink.streaming.runtime.tasks.SourceStreamTask.run (SourceStreamTask.java:99)
  [62] org.apache.flink.streaming.runtime.tasks.StreamTask.invoke (StreamTask.java:306)
  [63] org.apache.flink.runtime.taskmanager.Task.run (Task.java:703)
  [64] java.lang.Thread.run (Thread.java:748)

看来是通过一个线程池执行异步通知，handler是啥呢？？？

stop at org.apache.flink.runtime.io.network.netty.PartitionRequestQueue:75

找到了下面2个调用栈

[1] org.apache.flink.runtime.io.network.netty.PartitionRequestQueue.channelRegistered (PartitionRequestQueue.java:76)
  [2] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeChannelRegistered (AbstractChannelHandlerContext.java:162)
  [3] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireChannelRegistered (AbstractChannelHandlerContext.java:148)
  [4] org.apache.flink.shaded.netty4.io.netty.channel.ChannelInboundHandlerAdapter.channelRegistered (ChannelInboundHandlerAdapter.java:42)
  [5] org.apache.flink.runtime.io.network.netty.PartitionRequestServerHandler.channelRegistered (PartitionRequestServerHandler.java:68)
  [6] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeChannelRegistered (AbstractChannelHandlerContext.java:162)
  [7] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireChannelRegistered (AbstractChannelHandlerContext.java:148)
  [8] org.apache.flink.shaded.netty4.io.netty.channel.ChannelInboundHandlerAdapter.channelRegistered (ChannelInboundHandlerAdapter.java:42)
  [9] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeChannelRegistered (AbstractChannelHandlerContext.java:162)
  [10] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireChannelRegistered (AbstractChannelHandlerContext.java:148)
  [11] org.apache.flink.shaded.netty4.io.netty.channel.ChannelInboundHandlerAdapter.channelRegistered (ChannelInboundHandlerAdapter.java:42)
  [12] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeChannelRegistered (AbstractChannelHandlerContext.java:162)
  [13] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireChannelRegistered (AbstractChannelHandlerContext.java:148)
  [14] org.apache.flink.shaded.netty4.io.netty.channel.ChannelInitializer.channelRegistered (ChannelInitializer.java:71)
  [15] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeChannelRegistered (AbstractChannelHandlerContext.java:162)
  [16] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireChannelRegistered (AbstractChannelHandlerContext.java:148)
  [17] org.apache.flink.shaded.netty4.io.netty.channel.DefaultChannelPipeline.fireChannelRegistered (DefaultChannelPipeline.java:734)
  [18] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannel$AbstractUnsafe.register0 (AbstractChannel.java:450)
  [19] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannel$AbstractUnsafe.register (AbstractChannel.java:418)
  [20] org.apache.flink.shaded.netty4.io.netty.channel.SingleThreadEventLoop.register (SingleThreadEventLoop.java:60)
  [21] org.apache.flink.shaded.netty4.io.netty.channel.SingleThreadEventLoop.register (SingleThreadEventLoop.java:48)
  [22] org.apache.flink.shaded.netty4.io.netty.channel.MultithreadEventLoopGroup.register (MultithreadEventLoopGroup.java:64)
  [23] org.apache.flink.shaded.netty4.io.netty.bootstrap.ServerBootstrap$ServerBootstrapAcceptor.channelRead (ServerBootstrap.java:251)
  [24] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead (AbstractChannelHandlerContext.java:339)
  [25] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireChannelRead (AbstractChannelHandlerContext.java:324)
  [26] org.apache.flink.shaded.netty4.io.netty.channel.DefaultChannelPipeline.fireChannelRead (DefaultChannelPipeline.java:847)
  [27] org.apache.flink.shaded.netty4.io.netty.channel.nio.AbstractNioMessageChannel$NioMessageUnsafe.read (AbstractNioMessageChannel.java:93)
  [28] org.apache.flink.shaded.netty4.io.netty.channel.nio.NioEventLoop.processSelectedKey (NioEventLoop.java:511)
  [29] org.apache.flink.shaded.netty4.io.netty.channel.nio.NioEventLoop.processSelectedKeysOptimized (NioEventLoop.java:468)
  [30] org.apache.flink.shaded.netty4.io.netty.channel.nio.NioEventLoop.processSelectedKeys (NioEventLoop.java:382)
  [31] org.apache.flink.shaded.netty4.io.netty.channel.nio.NioEventLoop.run (NioEventLoop.java:354)
  [32] org.apache.flink.shaded.netty4.io.netty.util.concurrent.SingleThreadEventExecutor$2.run (SingleThreadEventExecutor.java:111)
  [33] java.lang.Thread.run (Thread.java:748)

以及

90    		PartitionRequestServerHandler serverHandler = new PartitionRequestServerHandler(

Flink Netty Server (0) Thread 0[1] where
  [1] org.apache.flink.runtime.io.network.netty.NettyProtocol.getServerChannelHandlers (NettyProtocol.java:90)
  [2] org.apache.flink.runtime.io.network.netty.NettyServer$1.initChannel (NettyServer.java:161)
  [3] org.apache.flink.runtime.io.network.netty.NettyServer$1.initChannel (NettyServer.java:151)
  [4] org.apache.flink.shaded.netty4.io.netty.channel.ChannelInitializer.channelRegistered (ChannelInitializer.java:69)
  [5] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeChannelRegistered (AbstractChannelHandlerContext.java:162)
  [6] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireChannelRegistered (AbstractChannelHandlerContext.java:148)
  [7] org.apache.flink.shaded.netty4.io.netty.channel.DefaultChannelPipeline.fireChannelRegistered (DefaultChannelPipeline.java:734)
  [8] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannel$AbstractUnsafe.register0 (AbstractChannel.java:450)
  [9] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannel$AbstractUnsafe.register (AbstractChannel.java:418)
  [10] org.apache.flink.shaded.netty4.io.netty.channel.SingleThreadEventLoop.register (SingleThreadEventLoop.java:60)
  [11] org.apache.flink.shaded.netty4.io.netty.channel.SingleThreadEventLoop.register (SingleThreadEventLoop.java:48)
  [12] org.apache.flink.shaded.netty4.io.netty.channel.MultithreadEventLoopGroup.register (MultithreadEventLoopGroup.java:64)
  [13] org.apache.flink.shaded.netty4.io.netty.bootstrap.ServerBootstrap$ServerBootstrapAcceptor.channelRead (ServerBootstrap.java:251)
  [14] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead (AbstractChannelHandlerContext.java:339)
  [15] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireChannelRead (AbstractChannelHandlerContext.java:324)
  [16] org.apache.flink.shaded.netty4.io.netty.channel.DefaultChannelPipeline.fireChannelRead (DefaultChannelPipeline.java:847)
  [17] org.apache.flink.shaded.netty4.io.netty.channel.nio.AbstractNioMessageChannel$NioMessageUnsafe.read (AbstractNioMessageChannel.java:93)
  [18] org.apache.flink.shaded.netty4.io.netty.channel.nio.NioEventLoop.processSelectedKey (NioEventLoop.java:511)
  [19] org.apache.flink.shaded.netty4.io.netty.channel.nio.NioEventLoop.processSelectedKeysOptimized (NioEventLoop.java:468)
  [20] org.apache.flink.shaded.netty4.io.netty.channel.nio.NioEventLoop.processSelectedKeys (NioEventLoop.java:382)
  [21] org.apache.flink.shaded.netty4.io.netty.channel.nio.NioEventLoop.run (NioEventLoop.java:354)
  [22] org.apache.flink.shaded.netty4.io.netty.util.concurrent.SingleThreadEventExecutor$2.run (SingleThreadEventExecutor.java:111)
  [23] java.lang.Thread.run (Thread.java:748)

这让我发现了2个handler

	public ChannelHandler[] getServerChannelHandlers() {
		PartitionRequestQueue queueOfPartitionQueues = new PartitionRequestQueue();
		PartitionRequestServerHandler serverHandler = new PartitionRequestServerHandler(
			partitionProvider, taskEventDispatcher, queueOfPartitionQueues, creditBasedEnabled);

		return new ChannelHandler[] {
			messageEncoder,
			createFrameLengthDecoder(),
			messageDecoder,
			serverHandler,
			queueOfPartitionQueues
		};
	}
		bootstrap.childHandler(new ChannelInitializer<SocketChannel>() {
			@Override
			public void initChannel(SocketChannel channel) throws Exception {
				if (serverSSLContext != null) {
					SSLEngine sslEngine = serverSSLContext.createSSLEngine();
					config.setSSLVerAndCipherSuites(sslEngine);
					sslEngine.setUseClientMode(false);
					channel.pipeline().addLast("ssl", new SslHandler(sslEngine));
				}

				channel.pipeline().addLast(protocol.getServerChannelHandlers());
			}
		});

从名字来看，还挺吻合，尝试看看

先看最下面2个handler的初始化，断点就是

stop in org.apache.flink.runtime.io.network.netty.NettyProtocol.getServerChannelHandlers

初始化完毕后，我们看怎么触发的，断点在下面2个

stop in org.apache.flink.runtime.io.network.netty.PartitionRequestQueue.userEventTriggered

确实触发了

Breakpoint hit: "thread=Flink Netty Server (0) Thread 0", org.apache.flink.runtime.io.network.netty.PartitionRequestQueue.userEventTriggered(), line=175 bci=0
175    		if (msg instanceof NetworkSequenceViewReader) {

Flink Netty Server (0) Thread 0[1] where
  [1] org.apache.flink.runtime.io.network.netty.PartitionRequestQueue.userEventTriggered (PartitionRequestQueue.java:175)
  [2] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeUserEventTriggered (AbstractChannelHandlerContext.java:308)
  [3] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireUserEventTriggered (AbstractChannelHandlerContext.java:294)
  [4] org.apache.flink.shaded.netty4.io.netty.channel.ChannelInboundHandlerAdapter.userEventTriggered (ChannelInboundHandlerAdapter.java:108)
  [5] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeUserEventTriggered (AbstractChannelHandlerContext.java:308)
  [6] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireUserEventTriggered (AbstractChannelHandlerContext.java:294)
  [7] org.apache.flink.shaded.netty4.io.netty.channel.ChannelInboundHandlerAdapter.userEventTriggered (ChannelInboundHandlerAdapter.java:108)
  [8] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeUserEventTriggered (AbstractChannelHandlerContext.java:308)
  [9] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireUserEventTriggered (AbstractChannelHandlerContext.java:294)
  [10] org.apache.flink.shaded.netty4.io.netty.channel.ChannelInboundHandlerAdapter.userEventTriggered (ChannelInboundHandlerAdapter.java:108)
  [11] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.invokeUserEventTriggered (AbstractChannelHandlerContext.java:308)
  [12] org.apache.flink.shaded.netty4.io.netty.channel.AbstractChannelHandlerContext.fireUserEventTriggered (AbstractChannelHandlerContext.java:294)
  [13] org.apache.flink.shaded.netty4.io.netty.channel.DefaultChannelPipeline.fireUserEventTriggered (DefaultChannelPipeline.java:841)
  [14] org.apache.flink.runtime.io.network.netty.PartitionRequestQueue$1.run (PartitionRequestQueue.java:95)
  [15] org.apache.flink.shaded.netty4.io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks (SingleThreadEventExecutor.java:357)
  [16] org.apache.flink.shaded.netty4.io.netty.channel.nio.NioEventLoop.run (NioEventLoop.java:357)
  [17] org.apache.flink.shaded.netty4.io.netty.util.concurrent.SingleThreadEventExecutor$2.run (SingleThreadEventExecutor.java:111)
  [18] java.lang.Thread.run (Thread.java:748)

继续打断点

stop in org.apache.flink.runtime.io.network.netty.PartitionRequestQueue.writeAndFlushNextMessageIfPossible

最后发出去的消息是

264    					channel.writeAndFlush(msg).addListener(writeListener);

Flink Netty Server (0) Thread 0[1] print msg
 msg = "org.apache.flink.runtime.io.network.netty.NettyMessage$BufferResponse@1ae5086"

打印这个channel,

channel = "[id: 0x4f18cfc9, /192.168.255.184:57834 => /192.168.255.185:35974]"

是别的机器过来连自己，发给客户端那边的，客户端那边怎么处理的？？？

 

 

 

数据通过group发出去了，那么谁来取呢？断点就是

stop in org.apache.flink.runtime.io.network.partition.consumer.SingleInputGate.getNextBufferOrEvent

 

 

确实有，不过是在groupby线程里的，调用栈如下：

  [1] org.apache.flink.runtime.io.network.partition.consumer.SingleInputGate.getNextBufferOrEvent (SingleInputGate.java:540)
  [2] org.apache.flink.runtime.io.network.partition.consumer.SingleInputGate.getNextBufferOrEvent (SingleInputGate.java:502)
  [3] org.apache.flink.streaming.runtime.io.BarrierBuffer.getNextNonBlocked (BarrierBuffer.java:165)
  [4] org.apache.flink.streaming.runtime.io.StreamInputProcessor.processInput (StreamInputProcessor.java:209)
  [5] org.apache.flink.streaming.runtime.tasks.OneInputStreamTask.run (OneInputStreamTask.java:103)
  [6] org.apache.flink.streaming.runtime.tasks.StreamTask.invoke (StreamTask.java:306)
  [7] org.apache.flink.runtime.taskmanager.Task.run (Task.java:703)
  [8] java.lang.Thread.run (Thread.java:748)

取完之后，剩下的就是被window operator处理了，调用栈是

  [1] org.apache.flink.streaming.runtime.operators.windowing.WindowOperator.processElement (WindowOperator.java:295)
  [2] org.apache.flink.streaming.runtime.io.StreamInputProcessor.processInput (StreamInputProcessor.java:202)
  [3] org.apache.flink.streaming.runtime.tasks.OneInputStreamTask.run (OneInputStreamTask.java:103)
  [4] org.apache.flink.streaming.runtime.tasks.StreamTask.invoke (StreamTask.java:306)
  [5] org.apache.flink.runtime.taskmanager.Task.run (Task.java:703)
  [6] java.lang.Thread.run (Thread.java:748)

对应的断点就是：

stop in org.apache.flink.streaming.runtime.operators.windowing.WindowOperator.processElement

结果发现一个大问题，这里的

print windowAssigner.getClass()
 windowAssigner.getClass() = "class org.apache.flink.streaming.api.windowing.assigners.TumblingProcessingTimeWindows"

这就恐怖了，我命名的是EventTime,为什么是ProcessingTime呢？定位到初始化函数

stop at org.apache.flink.streaming.runtime.operators.windowing.WindowOperator:193

最后检查本地java代码，搞定，验证如下

print windowAssigner.getClass()
 windowAssigner.getClass() = "class org.apache.flink.streaming.api.windowing.assigners.TumblingEventTimeWindows"

 

关于时间窗口，我们放到第12节讲，我们将继续的处理

stop in org.apache.flink.streaming.runtime.operators.windowing.WindowOperator.emitWindowContents

后面会把结果算出来，然后继续处理，所谓的调用栈如下

[1] DataStreamCalcRule$49.processElement (null)
  [2] org.apache.flink.table.runtime.CRowProcessRunner.processElement (CRowProcessRunner.scala:66)
  [3] org.apache.flink.table.runtime.CRowProcessRunner.processElement (CRowProcessRunner.scala:35)
  [4] org.apache.flink.streaming.api.operators.ProcessOperator.processElement (ProcessOperator.java:66)
  [5] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator (OperatorChain.java:560)
  [6] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:535)
  [7] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:515)
  [8] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [9] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [10] org.apache.flink.streaming.api.operators.TimestampedCollector.collect (TimestampedCollector.java:51)
  [11] org.apache.flink.table.runtime.aggregate.TimeWindowPropertyCollector.collect (TimeWindowPropertyCollector.scala:65)
  [12] org.apache.flink.table.runtime.aggregate.IncrementalAggregateWindowFunction.apply (IncrementalAggregateWindowFunction.scala:74)
  [13] org.apache.flink.table.runtime.aggregate.IncrementalAggregateTimeWindowFunction.apply (IncrementalAggregateTimeWindowFunction.scala:72)
  [14] org.apache.flink.table.runtime.aggregate.IncrementalAggregateTimeWindowFunction.apply (IncrementalAggregateTimeWindowFunction.scala:39)
  [15] org.apache.flink.streaming.runtime.operators.windowing.functions.InternalSingleValueWindowFunction.process (InternalSingleValueWindowFunction.java:46)
  [16] org.apache.flink.streaming.runtime.operators.windowing.WindowOperator.emitWindowContents (WindowOperator.java:550)
  [17] org.apache.flink.streaming.runtime.operators.windowing.WindowOperator.onEventTime (WindowOperator.java:457)
  [18] org.apache.flink.streaming.api.operators.HeapInternalTimerService.advanceWatermark (HeapInternalTimerService.java:288)
  [19] org.apache.flink.streaming.api.operators.InternalTimeServiceManager.advanceWatermark (InternalTimeServiceManager.java:108)
  [20] org.apache.flink.streaming.api.operators.AbstractStreamOperator.processWatermark (AbstractStreamOperator.java:734)
  [21] org.apache.flink.streaming.runtime.io.StreamInputProcessor$ForwardingValveOutputHandler.handleWatermark (StreamInputProcessor.java:262)
  [22] org.apache.flink.streaming.runtime.streamstatus.StatusWatermarkValve.findAndOutputNewMinWatermarkAcrossAlignedChannels (StatusWatermarkValve.java:189)
  [23] org.apache.flink.streaming.runtime.streamstatus.StatusWatermarkValve.inputWatermark (StatusWatermarkValve.java:111)
  [24] org.apache.flink.streaming.runtime.io.StreamInputProcessor.processInput (StreamInputProcessor.java:184)
  [25] org.apache.flink.streaming.runtime.tasks.OneInputStreamTask.run (OneInputStreamTask.java:103)
  [26] org.apache.flink.streaming.runtime.tasks.StreamTask.invoke (StreamTask.java:306)
  [27] org.apache.flink.runtime.taskmanager.Task.run (Task.java:703)
  [28] java.lang.Thread.run (Thread.java:748)

中间debug几次都是有问题，不过想一想应该是到了producer那块，

断点在

stop in org.apache.flink.streaming.connectors.kafka.FlinkKafkaProducerBase.open

也就是说，先看open的操作

  [1] org.apache.flink.streaming.connectors.kafka.FlinkKafkaProducerBase.open (FlinkKafkaProducerBase.java:215)
  [2] org.apache.flink.api.common.functions.util.FunctionUtils.openFunction (FunctionUtils.java:36)
  [3] org.apache.flink.streaming.api.operators.AbstractUdfStreamOperator.open (AbstractUdfStreamOperator.java:102)
  [4] org.apache.flink.streaming.api.operators.StreamSink.open (StreamSink.java:48)
  [5] org.apache.flink.streaming.runtime.tasks.StreamTask.openAllOperators (StreamTask.java:420)
  [6] org.apache.flink.streaming.runtime.tasks.StreamTask.invoke (StreamTask.java:296)
  [7] org.apache.flink.runtime.taskmanager.Task.run (Task.java:703)
  [8] java.lang.Thread.run (Thread.java:748)

最后看一下invoke的操作 

stop in org.apache.flink.streaming.connectors.kafka.partitioner.FlinkFixedPartitioner.partition

命中，直接看调用栈

[1] org.apache.flink.streaming.connectors.kafka.partitioner.FlinkFixedPartitioner.partition (FlinkFixedPartitioner.java:71)
  [2] org.apache.flink.streaming.connectors.kafka.FlinkKafkaProducer010.invoke (FlinkKafkaProducer010.java:375)
  [3] org.apache.flink.streaming.api.operators.StreamSink.processElement (StreamSink.java:56)
  [4] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator (OperatorChain.java:560)
  [5] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:535)
  [6] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:515)
  [7] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [8] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [9] org.apache.flink.streaming.api.operators.StreamMap.processElement (StreamMap.java:41)
  [10] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator (OperatorChain.java:560)
  [11] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:535)
  [12] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:515)
  [13] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [14] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [15] org.apache.flink.streaming.api.operators.TimestampedCollector.collect (TimestampedCollector.java:51)
  [16] org.apache.flink.table.runtime.CRowWrappingCollector.collect (CRowWrappingCollector.scala:37)
  [17] org.apache.flink.table.runtime.CRowWrappingCollector.collect (CRowWrappingCollector.scala:28)
  [18] DataStreamCalcRule$49.processElement (null)
  [19] org.apache.flink.table.runtime.CRowProcessRunner.processElement (CRowProcessRunner.scala:66)
  [20] org.apache.flink.table.runtime.CRowProcessRunner.processElement (CRowProcessRunner.scala:35)
  [21] org.apache.flink.streaming.api.operators.ProcessOperator.processElement (ProcessOperator.java:66)
  [22] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator (OperatorChain.java:560)
  [23] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:535)
  [24] org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect (OperatorChain.java:515)
  [25] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:679)
  [26] org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.collect (AbstractStreamOperator.java:657)
  [27] org.apache.flink.streaming.api.operators.TimestampedCollector.collect (TimestampedCollector.java:51)
  [28] org.apache.flink.table.runtime.aggregate.TimeWindowPropertyCollector.collect (TimeWindowPropertyCollector.scala:65)
  [29] org.apache.flink.table.runtime.aggregate.IncrementalAggregateWindowFunction.apply (IncrementalAggregateWindowFunction.scala:74)
  [30] org.apache.flink.table.runtime.aggregate.IncrementalAggregateTimeWindowFunction.apply (IncrementalAggregateTimeWindowFunction.scala:72)
  [31] org.apache.flink.table.runtime.aggregate.IncrementalAggregateTimeWindowFunction.apply (IncrementalAggregateTimeWindowFunction.scala:39)
  [32] org.apache.flink.streaming.runtime.operators.windowing.functions.InternalSingleValueWindowFunction.process (InternalSingleValueWindowFunction.java:46)
  [33] org.apache.flink.streaming.runtime.operators.windowing.WindowOperator.emitWindowContents (WindowOperator.java:550)
  [34] org.apache.flink.streaming.runtime.operators.windowing.WindowOperator.onEventTime (WindowOperator.java:457)
  [35] org.apache.flink.streaming.api.operators.HeapInternalTimerService.advanceWatermark (HeapInternalTimerService.java:288)
  [36] org.apache.flink.streaming.api.operators.InternalTimeServiceManager.advanceWatermark (InternalTimeServiceManager.java:108)
  [37] org.apache.flink.streaming.api.operators.AbstractStreamOperator.processWatermark (AbstractStreamOperator.java:734)
  [38] org.apache.flink.streaming.runtime.io.StreamInputProcessor$ForwardingValveOutputHandler.handleWatermark (StreamInputProcessor.java:262)
  [39] org.apache.flink.streaming.runtime.streamstatus.StatusWatermarkValve.findAndOutputNewMinWatermarkAcrossAlignedChannels (StatusWatermarkValve.java:189)
  [40] org.apache.flink.streaming.runtime.streamstatus.StatusWatermarkValve.inputWatermark (StatusWatermarkValve.java:111)
  [41] org.apache.flink.streaming.runtime.io.StreamInputProcessor.processInput (StreamInputProcessor.java:184)
  [42] org.apache.flink.streaming.runtime.tasks.OneInputStreamTask.run (OneInputStreamTask.java:103)
  [43] org.apache.flink.streaming.runtime.tasks.StreamTask.invoke (StreamTask.java:306)
  [44] org.apache.flink.runtime.taskmanager.Task.run (Task.java:703)
  [45] java.lang.Thread.run (Thread.java:748)

 

转载于:https://my.oschina.net/qiangzigege/blog/2208746

weixin_33695082
关注
  • 0
    点赞
  • 踩
  • 0
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
1关于flink的时间字段-ProcessTime-重新debug-读kafka->通知观察方(含正确脚本)
weixin_33910434的博客
11-30 566
2019独角兽企业重金招聘Python工程师标准>>> ...
flink任务重启原因分析
L13763338360的博客
08-10 1万+
有个flink实时任务，上周升级了版本，早上过来看下任务，发现任务凌晨4点左右的时候重启了。flink ui查看异常日志如下 异常日志 2020-08-10 04:07:23 org.apache.flink.runtime.io.network.netty.exception.RemoteTransportException: Connection unexpectedly closed by remote task manager '/9.150.12.175:39365'. This m.
Flink 源码解析（十）数据在各个task之间exchange的过程
xingdianp的博客
12-01 398
数据流转过程 上一节讲了各层数据的抽象，这一节讲讲数据在各个task之间exchange的过程。 1 整体过程 看这张图： 第一步必然是准备一个ResultPartition； 通知JobMaster； JobMaster通知下游节点；如果下游节点尚未部署，则部署之； 下游节点向上游请求数据 开始传输数据 .2 数据跨task传递 本节讲一下算子之间具体的数据传输过程。也先上一张图： image_1cfmpba9v15anggtvsba2o1277m.png-357.5kB 数据在task之间传递有如下几
flink-doris-connector-1.16-1.5.0-SNAPSHOT.jar
08-18
Flink-Doris最新发布的1.16驱动jar包，极大的提升了实时同步ODS效率，不仅可以同步DML,还可以同步DDL,并且可以自动创建目标表结构，为数据仓库同学解决了燃眉之急。
flink-1.5.0-bin-hadoop26-scala_2.11.tgz
03-26
flink-1.5.0-bin-hadoop26-scala_2.11.tgz 用于flink安装(1.5版本有全局锁的问题,新版本2.2无锁) 官网下载地址https://archive.apache.org/dist/flink/flink-1.5.0/flink-1.5.0-bin-hadoop26-scala_2.11.tgz
flink-1.5.0-src.tgz
07-07
《Apache Flink 1.5.0 源码解析》 Apache Flink 是一个用于处理无界和有界数据的开源流处理框架，它在大数据处理领域扮演着重要的角色。Flink 1.5.0 版本是其发展历程中的一个重要里程碑，引入了多项改进和新特性，...
iceberg-flink-runtime-1.16-1.5.0.jar
最新发布
06-05
flink、iceberg
flink-clickhouse-sink:Flink水槽的Clickhouse
05-17
Flink-ClickHouse-Sink 描述 用于数据库的器。 由。 用于将数据加载到ClickHouse的...flink-clickhouse-sink使用配置属性的两个部分：common和操作员链中的每个接收器。 共同的部分（像全局一样使用）： clickhouse
Flink使用Rocksdb状态后端任务运行一段时间后挂掉
一个不安分的程序员
04-11 2979
问题描述： 由于需要存储大量状态信息，在使用Flink Sql开发实时计算作业时配置RocksDB做为状态后端， 任务起初运行良好，运行一段时间后有任务偶发性出现失败，（使用yarn logs -applicaitonId）查看作业日志发现以下异常信息： org.apache.flink.runtime.io.network.netty.exception.RemoteTransportException: Connection unexpectedly closed by remote task m
flink-Connecting to remote task manager ‘/192.168.1.*:*‘ has failed.
Number_1412的专栏
07-29 1494
flink taskmanager报错Connection refused
ctp join java 崩溃_有时会出现运行Flink内置程序的情况异常：java.io.IOException： join频道失败...
weixin_39616798的博客
02-27 377
我已经建立了一个flink Standalone cluster，有一个主集群和三个从集群，都是sesulinux机器。在主仪表板中http://flink master：8081/我可以看到我设置的3个任务管理器和3个任务槽taskmanager.numberOfTaskSlots：1英寸火石-确认.yaml所有的奴隶。当我运行flink内置程序时，比如示例/流/迭代.jar，我经常例外：jav...
flink 问题记录
Jhon_yh的博客
06-09 1826
文章目录1. Caused by: java.lang.UnsatisfiedLinkError: org.apache.hadoop.util.NativeCrc32.nativeComputeChunkedSums(IILjava/nio/ByteBuffer;ILjava/nio/ByteBuffer;IILjava/lang/String;JZ)V原因 1. Caused by: java.lang.UnsatisfiedLinkError: org.apache.hadoop.util.Nat
部署flink集群遇到的问题（记录）
飞朋的博客
01-03 2056
1、flink jdk 版本的问题 org/apache/flink/yarn/entrypoint/YarnSessionClusterEntrypoint : Unsupported major.minor version 52.0 org/apache/flink/runtime/entrypoint/StandaloneSessionClusterEntrypoint : Unsupp...
Flink源码阅读(二)TaskManager
abcdef324的博客
04-22 685
TaskManager简介 通过前面的分析可以知道taskManager的入口是org.apache.flink.runtime.taskexecutor.TaskManagerRunner类，而且类上的注释也指明了这一点。 This class is the executable entry point for the task manager in yarn or standalone mo...
【Flink】Flink RecordWriterOutput pushToRecordWriter InterruptedException
九师兄
07-03 1236
1.美图 2.背景
Flink类型导致的异常：java.lang.NullPointerException
xiaozhaoshigedasb的博客
05-23 5118
代码提交到yarn上后，一直报这个错，但是数据好像没丢，或者丢的少， 没看出来。。 一开始分析的是业务逻辑中匹配的时候，会有一些类型的问题，但是全部排查了一遍，貌似没有，更何况下边这个异常里面并没有指出代码中的哪一行类型异常。。所以就怀疑是不是kafka中的数据进来的时候就有空字段。。 今天无意中翻了一眼官网，他说： Scala Case Class（包括Scala元组）：最多22个字段，不支持空...
【Flink】Flink 报错 flink 1.12.5 启动作业报 partition not found
九师兄
12-13 1463
flink 1.12.5 启动作业报 partition not found的问题， 可能是什么原因导致的启动作业就有这个问题，特别是耗费资源一百多g的大作业出现的更频繁基本每个作业都出现， flink 1.12.5 native k8s模式是不是有些算子state很大或者要load非常大的维表之类的，gc的话会无法响应下游的partition request，到达max backoff之后还连不上就会触发这个？
flink遇到的坑
热门推荐
晴空的博客
01-26 1万+
flink运行一切顺利，在发送数据量快达到1G的时候，突然发生故障，无故重启，最终导致整个flink程序崩溃。 发现问题后，就开始测试，追踪到日志，如下： AsynchronousException{java.lang.Exception: Could not materialize checkpoint 1143 for operator logData -> Sink: host
YuniKorn优化Flink-on-K8s调度：现状、挑战与未来
"1-FFA-YUNIKORN对Flink-on-K8s的调度优化-Final_compressed1" 本文主要探讨了YuniKorn如何优化Flink在Kubernetes(K8s)上的运行效率，以及Flink在K8s环境中的现状、挑战与未来的发展方向。 Flink on K8s现状： ...
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值