批处理执行模式下的 Apache Flink FileSink：进行中的文件不会转换为完成状态答案

【问题标题】：Apache Flink FileSink in BATCH execution mode: in-progress files are not transitioned to finished state批处理执行模式下的 Apache Flink FileSink：进行中的文件不会转换为完成状态
【发布时间】：2021-07-13 08:56:32
【问题描述】：

我们正在尝试做的事情：我们正在评估 Flink 以使用 DataStream API in BATCH mode 执行批处理。

重现问题的最小应用程序：

public class FlinkS3ProcessingDemoApplication {
    public static void main(String[] args) throws Exception {
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.setRuntimeMode(RuntimeExecutionMode.BATCH);

        DataStreamSource<String> source = env.readTextFile("file:///Users/user1/any-text-file.txt");

        source.sinkTo(FileSink.forRowFormat(new Path("file:///Users/user1/output/"), new SimpleStringEncoder<String>("UTF-8")).build());

        env.execute("Test Flink application");
    }
}

Flink 版本：1.12.2 或 1.13.0

预期结果：/Users/user1/output/ 文件夹中的“最终”文件。

根据FileSinkdocumentation：

鉴于 Flink 接收器和 UDF 通常不会区分正常作业终止（例如有限输入流）和因故障而终止，因此在作业正常终止时，最后一个正在进行的文件不会转换到“完成”状态。

BATCH 模式的具体说明：

待处理的文件被提交，即在整个输入处理完毕后转换到完成状态。

实际结果：

.
└── 2021-07-13--10
    ├── .part-707a8590-04cb-4c2d-97b2-5652697d9c76-0.inprogress.7e99df6f-703d-44b3-875a-283e12b31c8e
    ├── .part-a82bcabd-065d-4263-bee0-72f8673f3fd3-0.inprogress.65067b75-ef6c-4185-ae87-fe59de95c86a
    ├── .part-c7c36fd5-fb31-4d55-b783-5373ce69e216-0.inprogress.3e953235-09f1-487b-8229-2cdfa0e2daf4
    └── .part-e66b004a-271f-4aae-9604-e035b2c2cfe3-0.inprogress.add8b0d9-aa89-491e-9a9d-f07b73ab8256

以及以下例外情况：

Exception in thread "main" org.apache.flink.runtime.client.JobExecutionException: Job execution failed.
    at org.apache.flink.runtime.jobmaster.JobResult.toJobExecutionResult(JobResult.java:144)
    at org.apache.flink.runtime.minicluster.MiniClusterJobClient.lambda$getJobExecutionResult$2(MiniClusterJobClient.java:117)
    at java.util.concurrent.CompletableFuture.uniApply(CompletableFuture.java:616)
    at java.util.concurrent.CompletableFuture$UniApply.tryFire(CompletableFuture.java:591)
    at java.util.concurrent.CompletableFuture.postComplete(CompletableFuture.java:488)
    at java.util.concurrent.CompletableFuture.complete(CompletableFuture.java:1975)
    at org.apache.flink.runtime.rpc.akka.AkkaInvocationHandler.lambda$invokeRpc$0(AkkaInvocationHandler.java:237)
    at java.util.concurrent.CompletableFuture.uniWhenComplete(CompletableFuture.java:774)
    at java.util.concurrent.CompletableFuture$UniWhenComplete.tryFire(CompletableFuture.java:750)
    at java.util.concurrent.CompletableFuture.postComplete(CompletableFuture.java:488)
    at java.util.concurrent.CompletableFuture.complete(CompletableFuture.java:1975)
    at org.apache.flink.runtime.concurrent.FutureUtils$1.onComplete(FutureUtils.java:1046)
    at akka.dispatch.OnComplete.internal(Future.scala:264)
    at akka.dispatch.OnComplete.internal(Future.scala:261)
    at akka.dispatch.japi$CallbackBridge.apply(Future.scala:191)
    at akka.dispatch.japi$CallbackBridge.apply(Future.scala:188)
    at scala.concurrent.impl.CallbackRunnable.run(Promise.scala:36)
    at org.apache.flink.runtime.concurrent.Executors$DirectExecutionContext.execute(Executors.java:73)
    at scala.concurrent.impl.CallbackRunnable.executeWithValue(Promise.scala:44)
    at scala.concurrent.impl.Promise$DefaultPromise.tryComplete(Promise.scala:252)
    at akka.pattern.PromiseActorRef.$bang(AskSupport.scala:572)
    at akka.pattern.PipeToSupport$PipeableFuture$$anonfun$pipeTo$1.applyOrElse(PipeToSupport.scala:22)
    at akka.pattern.PipeToSupport$PipeableFuture$$anonfun$pipeTo$1.applyOrElse(PipeToSupport.scala:21)
    at scala.concurrent.Future$$anonfun$andThen$1.apply(Future.scala:436)
    at scala.concurrent.Future$$anonfun$andThen$1.apply(Future.scala:435)
    at scala.concurrent.impl.CallbackRunnable.run(Promise.scala:36)
    at akka.dispatch.BatchingExecutor$AbstractBatch.processBatch(BatchingExecutor.scala:55)
    at akka.dispatch.BatchingExecutor$BlockableBatch$$anonfun$run$1.apply$mcV$sp(BatchingExecutor.scala:91)
    at akka.dispatch.BatchingExecutor$BlockableBatch$$anonfun$run$1.apply(BatchingExecutor.scala:91)
    at akka.dispatch.BatchingExecutor$BlockableBatch$$anonfun$run$1.apply(BatchingExecutor.scala:91)
    at scala.concurrent.BlockContext$.withBlockContext(BlockContext.scala:72)
    at akka.dispatch.BatchingExecutor$BlockableBatch.run(BatchingExecutor.scala:90)
    at akka.dispatch.TaskInvocation.run(AbstractDispatcher.scala:40)
    at akka.dispatch.ForkJoinExecutorConfigurator$AkkaForkJoinTask.exec(ForkJoinExecutorConfigurator.scala:44)
    at akka.dispatch.forkjoin.ForkJoinTask.doExec(ForkJoinTask.java:260)
    at akka.dispatch.forkjoin.ForkJoinPool$WorkQueue.runTask(ForkJoinPool.java:1339)
    at akka.dispatch.forkjoin.ForkJoinPool.runWorker(ForkJoinPool.java:1979)
    at akka.dispatch.forkjoin.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:107)
Caused by: org.apache.flink.runtime.JobException: Recovery is suppressed by NoRestartBackoffTimeStrategy
    at org.apache.flink.runtime.executiongraph.failover.flip1.ExecutionFailureHandler.handleFailure(ExecutionFailureHandler.java:118)
    at org.apache.flink.runtime.executiongraph.failover.flip1.ExecutionFailureHandler.getFailureHandlingResult(ExecutionFailureHandler.java:80)
    at org.apache.flink.runtime.scheduler.DefaultScheduler.handleTaskFailure(DefaultScheduler.java:233)
    at org.apache.flink.runtime.scheduler.DefaultScheduler.maybeHandleTaskFailure(DefaultScheduler.java:224)
    at org.apache.flink.runtime.scheduler.DefaultScheduler.updateTaskExecutionStateInternal(DefaultScheduler.java:215)
    at org.apache.flink.runtime.scheduler.SchedulerBase.updateTaskExecutionState(SchedulerBase.java:669)
    at org.apache.flink.runtime.scheduler.SchedulerNG.updateTaskExecutionState(SchedulerNG.java:89)
    at org.apache.flink.runtime.jobmaster.JobMaster.updateTaskExecutionState(JobMaster.java:447)
    at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
    at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
    at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
    at java.lang.reflect.Method.invoke(Method.java:498)
    at org.apache.flink.runtime.rpc.akka.AkkaRpcActor.handleRpcInvocation(AkkaRpcActor.java:305)
    at org.apache.flink.runtime.rpc.akka.AkkaRpcActor.handleRpcMessage(AkkaRpcActor.java:212)
    at org.apache.flink.runtime.rpc.akka.FencedAkkaRpcActor.handleRpcMessage(FencedAkkaRpcActor.java:77)
    at org.apache.flink.runtime.rpc.akka.AkkaRpcActor.handleMessage(AkkaRpcActor.java:158)
    at akka.japi.pf.UnitCaseStatement.apply(CaseStatements.scala:26)
    at akka.japi.pf.UnitCaseStatement.apply(CaseStatements.scala:21)
    at scala.PartialFunction$class.applyOrElse(PartialFunction.scala:123)
    at akka.japi.pf.UnitCaseStatement.applyOrElse(CaseStatements.scala:21)
    at scala.PartialFunction$OrElse.applyOrElse(PartialFunction.scala:170)
    at scala.PartialFunction$OrElse.applyOrElse(PartialFunction.scala:171)
    at scala.PartialFunction$OrElse.applyOrElse(PartialFunction.scala:171)
    at akka.actor.Actor$class.aroundReceive(Actor.scala:517)
    at akka.actor.AbstractActor.aroundReceive(AbstractActor.scala:225)
    at akka.actor.ActorCell.receiveMessage(ActorCell.scala:592)
    at akka.actor.ActorCell.invoke(ActorCell.scala:561)
    at akka.dispatch.Mailbox.processMailbox(Mailbox.scala:258)
    at akka.dispatch.Mailbox.run(Mailbox.scala:225)
    at akka.dispatch.Mailbox.exec(Mailbox.scala:235)
    ... 4 more
Caused by: java.nio.channels.ClosedChannelException
    at sun.nio.ch.FileChannelImpl.ensureOpen(FileChannelImpl.java:110)
    at sun.nio.ch.FileChannelImpl.position(FileChannelImpl.java:253)
    at org.apache.flink.core.fs.local.LocalRecoverableFsDataOutputStream.getPos(LocalRecoverableFsDataOutputStream.java:103)
    at org.apache.flink.streaming.api.functions.sink.filesystem.OutputStreamBasedPartFileWriter.getSize(OutputStreamBasedPartFileWriter.java:74)
    at org.apache.flink.streaming.api.functions.sink.filesystem.rollingpolicies.DefaultRollingPolicy.shouldRollOnCheckpoint(DefaultRollingPolicy.java:71)
    at org.apache.flink.connector.file.sink.writer.FileWriterBucket.prepareCommit(FileWriterBucket.java:196)
    at org.apache.flink.connector.file.sink.writer.FileWriter.prepareCommit(FileWriter.java:200)
    at org.apache.flink.streaming.runtime.operators.sink.AbstractSinkWriterOperator.endInput(AbstractSinkWriterOperator.java:97)
    at org.apache.flink.streaming.runtime.tasks.StreamOperatorWrapper.endOperatorInput(StreamOperatorWrapper.java:91)
    at org.apache.flink.streaming.runtime.tasks.StreamOperatorWrapper.lambda$close$0(StreamOperatorWrapper.java:128)
    at org.apache.flink.streaming.runtime.tasks.StreamTaskActionExecutor$1.runThrowing(StreamTaskActionExecutor.java:50)
    at org.apache.flink.streaming.runtime.tasks.StreamOperatorWrapper.close(StreamOperatorWrapper.java:128)
    at org.apache.flink.streaming.runtime.tasks.StreamOperatorWrapper.close(StreamOperatorWrapper.java:135)
    at org.apache.flink.streaming.runtime.tasks.OperatorChain.closeOperators(OperatorChain.java:439)
    at org.apache.flink.streaming.runtime.tasks.StreamTask.afterInvoke(StreamTask.java:627)
    at org.apache.flink.streaming.runtime.tasks.StreamTask.invoke(StreamTask.java:589)
    at org.apache.flink.runtime.taskmanager.Task.doRun(Task.java:755)
    at org.apache.flink.runtime.taskmanager.Task.run(Task.java:570)
    at java.lang.Thread.run(Thread.java:748)

我们想知道的：是否可以结合FileSink或StreamingFileSink在批处理模式下使用Flink。

提前致谢！

【问题讨论】：

你能用新的FileSource试试这个吗？这是 readTextFile 的替代品，支持 BATCH 模式执行。 ci.apache.org/projects/flink/flink-docs-release-1.13/api/java/…
谢谢！是的，它可以与DataStreamSource<String> source = env.fromSource(FileSource.forRecordStreamFormat(new TextLineFormat(), new Path("file:///Users/user/file.txt")).build(), WatermarkStrategy.noWatermarks(), "MySourceName"); 一起正常工作。我应该将此添加为答案，还是您将提供更多背景信息，为什么它在您的答案中如此相似？

标签： apache-flink flink-batch

【解决方案1】：

在 FLIP-27 中重新设计的源接口以提供对 DataStream API 中的 BATCH 执行模式的支持。为了让FileSink 在以 BATCH 模式运行时正确地将 PENDING 文件转换为 FINISHED，您需要使用实现 FLIP-27 的源，例如 FileSource（而不是 readTextFile）：https://ci.apache.org/projects/flink/flink-docs-release-1.13/api/java/org/apache/flink/connector/file/src/FileSource.html .

如您所见，它看起来像这样：

DataStreamSource<String> source = 
  env.fromSource(
    FileSource.forRecordStreamFormat(
      new TextLineFormat(),
      new Path("file:///Users/user/file.txt")
    ).build(),
    WatermarkStrategy.noWatermarks(),
    "MySourceName"
  );

如果您需要批量格式，例如镶木地板，那么您可以执行以下操作：

DataStreamSource<String> source = 
  env.fromSource(
    FileSource.forBulkFileFormat(
      new ParquetColumnarRowInputFormat(...),
      new Path("file:///Users/me/data.parquet")
    ).build(),
    WatermarkStrategy.noWatermarks(),
    "MySourceName"
  );

parquet 还有ParquetVectorizedInputFormat，还有orc 等格式。

【讨论】：

嗨，大卫，读取 Parquet 数据时的逻辑是否相同？镶木地板的“TextLineFormat”相当于什么？并且可以自动创建读取文件的标题吗？
见ci.apache.org/projects/flink/flink-docs-stable/api/java/org/…
嗨，大卫，我正在尝试建议（使用 FileSource 和 FileSink），但该解决方案对我不起作用。看起来我缺少一些概念上的差异和理解何时将确切的数据发布到文件系统。是否有一些文档可供我阅读以了解其中的机制？
数据仅在检查点期间最终确定到文件系统——启用检查点对于使用文件接收器的流式作业是强制性的。
我已明确将运行时模式设置为 BATCH，现在文件正在完成！好吧 :) 现在很酷。不过，看起来输出数据存储在每小时的文件夹中。这看起来像一个默认的存储桶分配器。

【解决方案2】：

我无法使用上述方法，因为我正在读取 AVRO 文件并且找不到 AVRO 行格式。但是，如果您像这样为输入流分配时间戳和水印，它可以与 env.readTextFile 一起使用：

public class FlinkS3ProcessingDemoApplication {
public static void main(String[] args) throws Exception {
    StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
    env.setRuntimeMode(RuntimeExecutionMode.BATCH);

    DataStream<String> source = env.readTextFile("file:///Users/user1/any-text-file.txt")
                    .assignTimestampsAndWatermarks(
                            WatermarkStrategy.<String>noWatermarks().withTimestampAssigner(new SerializableTimestampAssigner<String>() {
                                @Override
                                public long extractTimestamp(String element, long recordTimestamp) {
                                    return -1;
                                }
                            }));

    source.sinkTo(FileSink.forRowFormat(new Path("file:///Users/artvolk/output/"), new SimpleStringEncoder<String>("UTF-8")).build());

    env.execute("Test Flink application");
}

}

【讨论】：