Extend DefaultCodec to support Zip compression for Hadoop file

Question

I've got some Spark code that reads two files from HDFS (a header file and a body file), reduces the RDD[String] to a single partition, then writes the result as a compressed file using the GZip codec:

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

This works 100% as expected. We're now being asked to support zip compression for Windows users who are unable to natively decompress *.gzip files. Obviously, zip format isn't natively supported, so I'm attempting to roll my own compression codec.

I'm running into a "ZipException: no current ZIP entry" exception when running the code however:

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)

I've created a ZipCodec class that extends DefaultCodec:

public class ZipCodec extends DefaultCodec {

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

As well as a ZipCompressorStream which extends CompressorStream:

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);
   }

We're currently using Spark 1.6.0 and Hadoop 2.6.0-cdh5.8.2

Any thoughts at all?

Thanks in advance!

Mariusz Mariusz · Accepted Answer · 2018-07-14T12:02:53

ZIP is a container format, when GZip is just a stream-like format (used to store one file). That's why when crating a new ZIP file you need to start an entry first (giving a name), then after writing close that entry before closing a container. See example here: https://www.programcreek.com/java-api-examples/?class=java.util.zip.ZipOutputStream&method=putNextEntry

Extend DefaultCodec to support Zip compression for Hadoop file

1 Answers