apache-spark - 扩展 DefaultCodec 以支持 Hadoop 文件的 Zip 压缩

Question

我有一些 Spark 代码从 HDFS 读取两个文件(一个头文件和一个正文文件)，将 RDD[String] 减少到单个分区，然后使用 GZip 编解码器将结果作为压缩文件写入:

spark.sparkContext.textFile("path_to_header.txt,path_to_body.txt")
.coalesce(1)
.saveAsTextFile("output_path", classOf[GzipCodec])

这按预期 100% 有效。我们现在被要求为无法本地解压缩 *.gzip 文件的 Windows 用户支持 zip 压缩。显然，zip 格式不受 native 支持，因此我正在尝试推出自己的压缩编解码器。

我在运行代码时遇到了“ ZipException: no current ZIP entry ”异常:

Exception occured while exporting org.apache.spark.SparkException: Job aborted due to stage failure: Task 0 in stage 16.0 failed 2 times, most recent failure: Lost task 0.1 in stage 16.0 (TID 675, xxxxxxx.xxxxx.xxx, executor 16): java.util.zip.ZipException: no current ZIP entry
    at java.util.zip.ZipOutputStream.write(Unknown Source)
    at io.ZipCompressorStream.write(ZipCompressorStream.java:23)
    at java.io.DataOutputStream.write(Unknown Source)
    at org.apache.hadoop.mapred.TextOutputFormat$LineRecordWriter.writeObject(TextOutputFormat.java:81)
    at org.apache.hadoop.mapred.TextOutputFormat$LineRecordWriter.write(TextOutputFormat.java:102)
    at org.apache.spark.SparkHadoopWriter.write(SparkHadoopWriter.scala:95)
    at org.apache.spark.rdd.PairRDDFunctions$$anonfun$saveAsHadoopDataset$1$$anonfun$13$$anonfun$apply$7.apply$mcV$sp(PairRDDFunctions.scala:1205)
    at org.apache.spark.rdd.PairRDDFunctions$$anonfun$saveAsHadoopDataset$1$$anonfun$13$$anonfun$apply$7.apply(PairRDDFunctions.scala:1203)
    at org.apache.spark.rdd.PairRDDFunctions$$anonfun$saveAsHadoopDataset$1$$anonfun$13$$anonfun$apply$7.apply(PairRDDFunctions.scala:1203)
    at org.apache.spark.util.Utils$.tryWithSafeFinallyAndFailureCallbacks(Utils.scala:1348)
    at org.apache.spark.rdd.PairRDDFunctions$$anonfun$saveAsHadoopDataset$1$$anonfun$13.apply(PairRDDFunctions.scala:1211)
    at org.apache.spark.rdd.PairRDDFunctions$$anonfun$saveAsHadoopDataset$1$$anonfun$13.apply(PairRDDFunctions.scala:1190)
    at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:70)
    at org.apache.spark.scheduler.Task.run(Task.scala:86)
    at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
    at java.lang.Thread.run(Unknown Source)

我创建了一个扩展 DefaultCodec 的 ZipCodec 类:

public class ZipCodec extends DefaultCodec {

   @Override
   public CompressionOutputStream createOutputStream(final OutputStream out, final Compressor compressor) throws IOException {
      return new ZipCompressorStream(new ZipOutputStream(out));
   }

以及扩展 CompressorStream 的 ZipCompressorStream:

public class ZipCompressorStream extends CompressorStream {

   public ZipCompressorStream(final ZipOutputStream out) {
      super(out);
   }

   @Override
   public void write(final int b) throws IOException {
      out.write(b);
   }

   @Override
   public void write(final byte[] data, final int offset, final int length) throws IOException {
      out.write(data, offset, length);
   }

我们目前正在使用 Spark 1.6.0 和 Hadoop 2.6.0-cdh5.8.2

有什么想法吗？

提前致谢!

Answer 1

ZIP 是一种容器格式，而 GZip 只是一种类似流的格式(用于存储一个文件)。这就是为什么在创建一个新的 ZIP 文件时需要先启动一个条目(给出一个名称)，然后在关闭该条目之后再关闭容器。请参阅此处的示例:https://www.programcreek.com/java-api-examples/?class=java.util.zip.ZipOutputStream&method=putNextEntry