Java Code Examples for org.apache.tez.runtime.library.common.ConfigUtils#getIntermediateInputValueClass()

public ShuffledUnorderedKVReader(ShuffleManager shuffleManager, Configuration conf,
    CompressionCodec codec, boolean ifileReadAhead, int ifileReadAheadLength, int ifileBufferSize,
    TezCounter inputRecordCounter)
    throws IOException {
  this.shuffleManager = shuffleManager;

  this.codec = codec;
  this.ifileReadAhead = ifileReadAhead;
  this.ifileReadAheadLength = ifileReadAheadLength;
  this.ifileBufferSize = ifileBufferSize;
  this.inputRecordCounter = inputRecordCounter;

  this.keyClass = ConfigUtils.getIntermediateInputKeyClass(conf);
  this.valClass = ConfigUtils.getIntermediateInputValueClass(conf);

  this.keyIn = new DataInputBuffer();
  this.valIn = new DataInputBuffer();

  SerializationFactory serializationFactory = new SerializationFactory(conf);

  this.keyDeserializer = serializationFactory.getDeserializer(keyClass);
  this.keyDeserializer.open(keyIn);
  this.valDeserializer = serializationFactory.getDeserializer(valClass);
  this.valDeserializer.open(valIn);
}
 

@SuppressWarnings({ "rawtypes", "unchecked" })
protected synchronized void createValuesIterator()
    throws IOException {
  // Not used by ReduceProcessor
  RawComparator rawComparator = ConfigUtils.getIntermediateInputKeyComparator(conf);
  Class<?> keyClass = ConfigUtils.getIntermediateInputKeyClass(conf);
  Class<?> valClass = ConfigUtils.getIntermediateInputValueClass(conf);
  LOG.info(getContext().getSourceVertexName() + ": " + "creating ValuesIterator with "
      + "comparator=" + rawComparator.getClass().getName()
      + ", keyClass=" + keyClass.getName()
      + ", valClass=" + valClass.getName());

  vIter = new ValuesIterator(rawIter, rawComparator, keyClass, valClass,
      conf, inputKeyCounter, inputValueCounter);

}
 

public UnorderedKVReader(ShuffleManager shuffleManager, Configuration conf,
    CompressionCodec codec, boolean ifileReadAhead, int ifileReadAheadLength, int ifileBufferSize,
    TezCounter inputRecordCounter, InputContext context)
    throws IOException {
  this.shuffleManager = shuffleManager;
  this.context = context;
  this.codec = codec;
  this.ifileReadAhead = ifileReadAhead;
  this.ifileReadAheadLength = ifileReadAheadLength;
  this.ifileBufferSize = ifileBufferSize;
  this.inputRecordCounter = inputRecordCounter;

  this.keyClass = ConfigUtils.getIntermediateInputKeyClass(conf);
  this.valClass = ConfigUtils.getIntermediateInputValueClass(conf);

  this.keyIn = new DataInputBuffer();
  this.valIn = new DataInputBuffer();

  SerializationFactory serializationFactory = new SerializationFactory(conf);

  this.keyDeserializer = serializationFactory.getDeserializer(keyClass);
  this.keyDeserializer.open(keyIn);
  this.valDeserializer = serializationFactory.getDeserializer(valClass);
  this.valDeserializer.open(valIn);
}
 

@SuppressWarnings({ "rawtypes", "unchecked" })
protected synchronized void createValuesIterator()
    throws IOException {
  // Not used by ReduceProcessor
  vIter = new ValuesIterator(rawIter,
      (RawComparator) ConfigUtils.getIntermediateInputKeyComparator(conf),
      ConfigUtils.getIntermediateInputKeyClass(conf),
      ConfigUtils.getIntermediateInputValueClass(conf), conf, inputKeyCounter, inputValueCounter);

}
 

public MRCombiner(TezTaskContext taskContext) throws IOException {
  this.conf = TezUtils.createConfFromUserPayload(taskContext.getUserPayload());

  assert(taskContext instanceof TezInputContext || taskContext instanceof TezOutputContext);
  if (taskContext instanceof TezOutputContext) {
    this.keyClass = ConfigUtils.getIntermediateOutputKeyClass(conf);
    this.valClass = ConfigUtils.getIntermediateOutputValueClass(conf);
    this.comparator = ConfigUtils.getIntermediateOutputKeyComparator(conf);
    this.reporter = new MRTaskReporter((TezOutputContext)taskContext);
  } else {
    this.keyClass = ConfigUtils.getIntermediateInputKeyClass(conf);
    this.valClass = ConfigUtils.getIntermediateInputValueClass(conf);
    this.comparator = ConfigUtils.getIntermediateInputKeyComparator(conf);
    this.reporter = new MRTaskReporter((TezInputContext)taskContext);
  }

  this.useNewApi = ConfigUtils.useNewApi(conf);
  
  combineInputKeyCounter = taskContext.getCounters().findCounter(TaskCounter.COMBINE_INPUT_RECORDS);
  combineInputValueCounter = taskContext.getCounters().findCounter(TaskCounter.COMBINE_OUTPUT_RECORDS);
  
  boolean isMap = conf.getBoolean(MRConfig.IS_MAP_PROCESSOR,false);
  this.mrTaskAttemptID = new TaskAttemptID(
      new TaskID(String.valueOf(taskContext.getApplicationId()
          .getClusterTimestamp()), taskContext.getApplicationId().getId(),
          isMap ? TaskType.MAP : TaskType.REDUCE,
          taskContext.getTaskIndex()), taskContext.getTaskAttemptNumber());
  
  LOG.info("Using combineKeyClass: " + keyClass + ", combineValueClass: " + valClass + ", combineComparator: " +comparator + ", useNewApi: " + useNewApi);
}
 

public MRCombiner(TaskContext taskContext) throws IOException {
  final Configuration userConf = TezUtils.createConfFromUserPayload(taskContext.getUserPayload());
  useNewApi = ConfigUtils.useNewApi(userConf);
  if (useNewApi) {
    conf = new JobConf(userConf);
  } else {
    conf = userConf;
  }

  assert(taskContext instanceof InputContext || taskContext instanceof OutputContext);
  if (taskContext instanceof OutputContext) {
    this.keyClass = ConfigUtils.getIntermediateOutputKeyClass(conf);
    this.valClass = ConfigUtils.getIntermediateOutputValueClass(conf);
    this.comparator = ConfigUtils.getIntermediateOutputKeyComparator(conf);
    this.reporter = new MRTaskReporter((OutputContext)taskContext);
  } else {
    this.keyClass = ConfigUtils.getIntermediateInputKeyClass(conf);
    this.valClass = ConfigUtils.getIntermediateInputValueClass(conf);
    this.comparator = ConfigUtils.getIntermediateInputKeyComparator(conf);
    this.reporter = new MRTaskReporter((InputContext)taskContext);
  }

  combineInputRecordsCounter = taskContext.getCounters().findCounter(TaskCounter.COMBINE_INPUT_RECORDS);
  combineOutputRecordsCounter = taskContext.getCounters().findCounter(TaskCounter.COMBINE_OUTPUT_RECORDS);
  
  boolean isMap = conf.getBoolean(MRConfig.IS_MAP_PROCESSOR,false);
  this.mrTaskAttemptID = new TaskAttemptID(
      new TaskID(String.valueOf(taskContext.getApplicationId()
          .getClusterTimestamp()), taskContext.getApplicationId().getId(),
          isMap ? TaskType.MAP : TaskType.REDUCE,
          taskContext.getTaskIndex()), taskContext.getTaskAttemptNumber());
  
  LOG.info("Using combineKeyClass: " + keyClass + ", combineValueClass: " + valClass + ", combineComparator: " +comparator + ", useNewApi: " + useNewApi);
}
 

public LocalShuffle(TezInputContext inputContext, Configuration conf, int numInputs) throws IOException {
   this.inputContext = inputContext;
   this.conf = conf;
   this.numInputs = numInputs;
   
   this.keyClass = ConfigUtils.getIntermediateInputKeyClass(conf);
   this.valClass = ConfigUtils.getIntermediateInputValueClass(conf);
   this.comparator = ConfigUtils.getIntermediateInputKeyComparator(conf);
   
   this.sortFactor =
       conf.getInt(
           TezJobConfig.TEZ_RUNTIME_IO_SORT_FACTOR, 
           TezJobConfig.TEZ_RUNTIME_IO_SORT_FACTOR_DEFAULT);
   
   this.rfs = FileSystem.getLocal(conf).getRaw();

   this.spilledRecordsCounter = inputContext.getCounters().findCounter(TaskCounter.SPILLED_RECORDS);
   
// compression
   if (ConfigUtils.isIntermediateInputCompressed(conf)) {
     Class<? extends CompressionCodec> codecClass =
         ConfigUtils.getIntermediateInputCompressorClass(conf, DefaultCodec.class);
     this.codec = ReflectionUtils.newInstance(codecClass, conf);
   } else {
     this.codec = null;
   }
   this.ifileReadAhead = conf.getBoolean(
       TezJobConfig.TEZ_RUNTIME_IFILE_READAHEAD,
       TezJobConfig.TEZ_RUNTIME_IFILE_READAHEAD_DEFAULT);
   if (this.ifileReadAhead) {
     this.ifileReadAheadLength = conf.getInt(
         TezJobConfig.TEZ_RUNTIME_IFILE_READAHEAD_BYTES,
         TezJobConfig.TEZ_RUNTIME_IFILE_READAHEAD_BYTES_DEFAULT);
   } else {
     this.ifileReadAheadLength = 0;
   }
   this.ifileBufferSize = conf.getInt("io.file.buffer.size",
       TezJobConfig.TEZ_RUNTIME_IFILE_BUFFER_SIZE_DEFAULT);
   
   // Always local
   this.mapOutputFile = new TezLocalTaskOutputFiles(conf, inputContext.getUniqueIdentifier());
 }
 

@Override
public void merge(List<MapOutput> inputs) throws IOException, InterruptedException {
  if (inputs == null || inputs.size() == 0) {
    return;
  }
  
  numMemToDiskMerges.increment(1);
  
  //name this output file same as the name of the first file that is 
  //there in the current list of inmem files (this is guaranteed to
  //be absent on the disk currently. So we don't overwrite a prev. 
  //created spill). Also we need to create the output file now since
  //it is not guaranteed that this file will be present after merge
  //is called (we delete empty files as soon as we see them
  //in the merge method)

  //figure out the mapId 
  InputAttemptIdentifier srcTaskIdentifier = inputs.get(0).getAttemptIdentifier();

  List<Segment> inMemorySegments = new ArrayList<Segment>();
  long mergeOutputSize = 
    createInMemorySegments(inputs, inMemorySegments,0);
  int noInMemorySegments = inMemorySegments.size();

  // TODO Maybe track serialized vs deserialized bytes.
  
  // All disk writes done by this merge are overhead - due to the lac of
  // adequate memory to keep all segments in memory.
  Path outputPath = mapOutputFile.getInputFileForWrite(
      srcTaskIdentifier.getInputIdentifier().getInputIndex(),
      mergeOutputSize).suffix(Constants.MERGED_OUTPUT_PREFIX);

  Writer writer = null;
  try {
    writer =
        new Writer(conf, rfs, outputPath,
            (Class)ConfigUtils.getIntermediateInputKeyClass(conf),
            (Class)ConfigUtils.getIntermediateInputValueClass(conf),
            codec, null, null);

    TezRawKeyValueIterator rIter = null;
    LOG.info("Initiating in-memory merge with " + noInMemorySegments + 
        " segments...");

    // Nothing actually materialized to disk - controlled by setting sort-factor to #segments.
    rIter = TezMerger.merge(conf, rfs,
        (Class)ConfigUtils.getIntermediateInputKeyClass(conf),
        (Class)ConfigUtils.getIntermediateInputValueClass(conf),
        inMemorySegments, inMemorySegments.size(),
        new Path(inputContext.getUniqueIdentifier()),
        (RawComparator)ConfigUtils.getIntermediateInputKeyComparator(conf),
        nullProgressable, spilledRecordsCounter, null, additionalBytesRead, null);
    // spilledRecordsCounter is tracking the number of keys that will be
    // read from each of the segments being merged - which is essentially
    // what will be written to disk.

    if (null == combiner) {
      TezMerger.writeFile(rIter, writer, nullProgressable, TezJobConfig.TEZ_RUNTIME_RECORDS_BEFORE_PROGRESS_DEFAULT);
    } else {
      // TODO Counters for Combine
      runCombineProcessor(rIter, writer);
    }
    writer.close();
    additionalBytesWritten.increment(writer.getCompressedLength());
    writer = null;

    LOG.info(inputContext.getUniqueIdentifier() +  
        " Merge of the " + noInMemorySegments +
        " files in-memory complete." +
        " Local file is " + outputPath + " of size " + 
        localFS.getFileStatus(outputPath).getLen());
  } catch (IOException e) { 
    //make sure that we delete the ondisk file that we created 
    //earlier when we invoked cloneFileAttributes
    localFS.delete(outputPath, true);
    throw e;
  } finally {
    if (writer != null) {
      writer.close();
    }
  }

  // Note the output of the merge
  closeOnDiskFile(outputPath);
}
 

@Override
public void merge(List<Path> inputs) throws IOException {
  // sanity check
  if (inputs == null || inputs.isEmpty()) {
    LOG.info("No ondisk files to merge...");
    return;
  }
  numDiskToDiskMerges.increment(1);
  
  long approxOutputSize = 0;
  int bytesPerSum = 
    conf.getInt("io.bytes.per.checksum", 512);
  
  LOG.info("OnDiskMerger: We have  " + inputs.size() + 
           " map outputs on disk. Triggering merge...");
  
  // 1. Prepare the list of files to be merged. 
  for (Path file : inputs) {
    approxOutputSize += localFS.getFileStatus(file).getLen();
  }

  // add the checksum length
  approxOutputSize += 
    ChecksumFileSystem.getChecksumLength(approxOutputSize, bytesPerSum);

  // 2. Start the on-disk merge process
  Path outputPath = 
    localDirAllocator.getLocalPathForWrite(inputs.get(0).toString(), 
        approxOutputSize, conf).suffix(Constants.MERGED_OUTPUT_PREFIX);
  Writer writer = 
    new Writer(conf, rfs, outputPath, 
                    (Class)ConfigUtils.getIntermediateInputKeyClass(conf), 
                    (Class)ConfigUtils.getIntermediateInputValueClass(conf),
                    codec, null, null);
  TezRawKeyValueIterator iter  = null;
  Path tmpDir = new Path(inputContext.getUniqueIdentifier());
  try {
    iter = TezMerger.merge(conf, rfs,
                        (Class)ConfigUtils.getIntermediateInputKeyClass(conf), 
                        (Class)ConfigUtils.getIntermediateInputValueClass(conf),
                        codec, ifileReadAhead, ifileReadAheadLength, ifileBufferSize,
                        inputs.toArray(new Path[inputs.size()]), true, ioSortFactor, tmpDir, 
                        (RawComparator)ConfigUtils.getIntermediateInputKeyComparator(conf), 
                        nullProgressable, spilledRecordsCounter, null, 
                        mergedMapOutputsCounter, null);

    // TODO Maybe differentiate between data written because of Merges and
    // the finalMerge (i.e. final mem available may be different from
    // initial merge mem)
    TezMerger.writeFile(iter, writer, nullProgressable, TezJobConfig.TEZ_RUNTIME_RECORDS_BEFORE_PROGRESS_DEFAULT);
    writer.close();
    additionalBytesWritten.increment(writer.getCompressedLength());
  } catch (IOException e) {
    localFS.delete(outputPath, true);
    throw e;
  }

  closeOnDiskFile(outputPath);

  LOG.info(inputContext.getUniqueIdentifier() +
      " Finished merging " + inputs.size() + 
      " map output files on disk of total-size " + 
      approxOutputSize + "." + 
      " Local output file is " + outputPath + " of size " +
      localFS.getFileStatus(outputPath).getLen());
}
 

@Override
public void run(Map<String, LogicalInput> inputs,
    Map<String, LogicalOutput> outputs) throws Exception {

  LOG.info("Running reduce: " + processorContext.getUniqueIdentifier());

  if (outputs.size() <= 0 || outputs.size() > 1) {
    throw new IOException("Invalid number of outputs"
        + ", outputCount=" + outputs.size());
  }

  if (inputs.size() <= 0 || inputs.size() > 1) {
    throw new IOException("Invalid number of inputs"
        + ", inputCount=" + inputs.size());
  }

  LogicalInput in = inputs.values().iterator().next();
  in.start();

  List<Input> pendingInputs = new LinkedList<Input>();
  pendingInputs.add(in);
  processorContext.waitForAllInputsReady(pendingInputs);
  LOG.info("Input is ready for consumption. Starting Output");

  LogicalOutput out = outputs.values().iterator().next();
  out.start();

  initTask(out);

  this.statusUpdate();

  Class keyClass = ConfigUtils.getIntermediateInputKeyClass(jobConf);
  Class valueClass = ConfigUtils.getIntermediateInputValueClass(jobConf);
  LOG.info("Using keyClass: " + keyClass);
  LOG.info("Using valueClass: " + valueClass);
  RawComparator comparator =
      ConfigUtils.getInputKeySecondaryGroupingComparator(jobConf);
  LOG.info("Using comparator: " + comparator);

  reduceInputKeyCounter =
      mrReporter.getCounter(TaskCounter.REDUCE_INPUT_GROUPS);
  reduceInputValueCounter =
      mrReporter.getCounter(TaskCounter.REDUCE_INPUT_RECORDS);

  // Sanity check
  if (!(in instanceof ShuffledMergedInputLegacy)) {
    throw new IOException("Illegal input to reduce: " + in.getClass());
  }
  ShuffledMergedInputLegacy shuffleInput = (ShuffledMergedInputLegacy)in;
  KeyValuesReader kvReader = shuffleInput.getReader();

  KeyValueWriter kvWriter = null;
  if((out instanceof MROutputLegacy)) {
    kvWriter = ((MROutputLegacy) out).getWriter();
  } else if ((out instanceof OnFileSortedOutput)) {
    kvWriter = ((OnFileSortedOutput) out).getWriter();
  } else {
    throw new IOException("Illegal output to reduce: " + in.getClass());
  }

  if (useNewApi) {
    try {
      runNewReducer(
          jobConf,
          mrReporter,
          shuffleInput, comparator,  keyClass, valueClass,
          kvWriter);
    } catch (ClassNotFoundException cnfe) {
      throw new IOException(cnfe);
    }
  } else {
    runOldReducer(
        jobConf, mrReporter,
        kvReader, comparator, keyClass, valueClass, kvWriter);
  }

  done();
}
 

@Override
public void merge(List<MapOutput> inputs) throws IOException, InterruptedException {
  if (inputs == null || inputs.size() == 0) {
    return;
  }

  numMemToDiskMerges.increment(1);
  inputContext.notifyProgress();
  
  //name this output file same as the name of the first file that is 
  //there in the current list of inmem files (this is guaranteed to
  //be absent on the disk currently. So we don't overwrite a prev. 
  //created spill). Also we need to create the output file now since
  //it is not guaranteed that this file will be present after merge
  //is called (we delete empty files as soon as we see them
  //in the merge method)

  //figure out the mapId 
  srcTaskIdentifier = inputs.get(0).getAttemptIdentifier();

  List<Segment> inMemorySegments = new ArrayList<Segment>();
  long mergeOutputSize =
    createInMemorySegments(inputs, inMemorySegments,0);
  int noInMemorySegments = inMemorySegments.size();

  // TODO Maybe track serialized vs deserialized bytes.
  
  // All disk writes done by this merge are overhead - due to the lack of
  // adequate memory to keep all segments in memory.
  outputPath = mapOutputFile.getInputFileForWrite(
      srcTaskIdentifier.getInputIdentifier(), srcTaskIdentifier.getSpillEventId(),
      mergeOutputSize).suffix(Constants.MERGED_OUTPUT_PREFIX);

  Writer writer = null;
  long outFileLen = 0;
  try {
    writer =
        new Writer(conf, rfs, outputPath,
            (Class)ConfigUtils.getIntermediateInputKeyClass(conf),
            (Class)ConfigUtils.getIntermediateInputValueClass(conf),
            codec, null, null);

    TezRawKeyValueIterator rIter = null;
    LOG.info("Initiating in-memory merge with " + noInMemorySegments + 
        " segments...");

    tmpDir = new Path(inputContext.getUniqueIdentifier());
    // Nothing actually materialized to disk - controlled by setting sort-factor to #segments.
    rIter = TezMerger.merge(conf, rfs,
        (Class)ConfigUtils.getIntermediateInputKeyClass(conf),
        (Class)ConfigUtils.getIntermediateInputValueClass(conf),
        inMemorySegments, inMemorySegments.size(),
        tmpDir, (RawComparator)ConfigUtils.getIntermediateInputKeyComparator(conf),
        progressable, spilledRecordsCounter, null, additionalBytesRead, null);
    // spilledRecordsCounter is tracking the number of keys that will be
    // read from each of the segments being merged - which is essentially
    // what will be written to disk.

    if (null == combiner) {
      TezMerger.writeFile(rIter, writer, progressable, TezRuntimeConfiguration.TEZ_RUNTIME_RECORDS_BEFORE_PROGRESS_DEFAULT);
    } else {
      // TODO Counters for Combine
      runCombineProcessor(rIter, writer);
    }
    writer.close();
    additionalBytesWritten.increment(writer.getCompressedLength());
    writer = null;

    outFileLen = localFS.getFileStatus(outputPath).getLen();
    LOG.info(inputContext.getUniqueIdentifier() +
        " Merge of the " + noInMemorySegments +
        " files in-memory complete." +
        " Local file is " + outputPath + " of size " +
        outFileLen);
  } catch (IOException e) { 
    //make sure that we delete the ondisk file that we created 
    //earlier when we invoked cloneFileAttributes
    localFS.delete(outputPath, true);
    throw e;
  } finally {
    if (writer != null) {
      writer.close();
    }
  }

  // Note the output of the merge
  closeOnDiskFile(new FileChunk(outputPath, 0, outFileLen));
}
 

@Override
public void run(Map<String, LogicalInput> _inputs,
    Map<String, LogicalOutput> _outputs) throws Exception {
  this.inputs = _inputs;
  this.outputs = _outputs;
  progressHelper = new ProgressHelper(this.inputs, processorContext, this.getClass().getSimpleName());
  LOG.info("Running reduce: " + processorContext.getUniqueIdentifier());

  if (_outputs.size() <= 0 || _outputs.size() > 1) {
    throw new IOException("Invalid number of _outputs"
        + ", outputCount=" + _outputs.size());
  }

  if (_inputs.size() <= 0 || _inputs.size() > 1) {
    throw new IOException("Invalid number of _inputs"
        + ", inputCount=" + _inputs.size());
  }

  LogicalInput in = _inputs.values().iterator().next();
  in.start();

  List<Input> pendingInputs = new LinkedList<Input>();
  pendingInputs.add(in);
  processorContext.waitForAllInputsReady(pendingInputs);
  LOG.info("Input is ready for consumption. Starting Output");

  LogicalOutput out = _outputs.values().iterator().next();
  out.start();

  initTask(out);
  progressHelper.scheduleProgressTaskService(0, 100);
  this.statusUpdate();

  Class keyClass = ConfigUtils.getIntermediateInputKeyClass(jobConf);
  Class valueClass = ConfigUtils.getIntermediateInputValueClass(jobConf);
  LOG.info("Using keyClass: " + keyClass);
  LOG.info("Using valueClass: " + valueClass);
  RawComparator comparator =
      ConfigUtils.getInputKeySecondaryGroupingComparator(jobConf);
  LOG.info("Using comparator: " + comparator);

  reduceInputKeyCounter =
      mrReporter.getCounter(TaskCounter.REDUCE_INPUT_GROUPS);
  reduceInputValueCounter =
      mrReporter.getCounter(TaskCounter.REDUCE_INPUT_RECORDS);

  // Sanity check
  if (!(in instanceof OrderedGroupedInputLegacy)) {
    throw new IOException("Illegal input to reduce: " + in.getClass());
  }
  OrderedGroupedInputLegacy shuffleInput = (OrderedGroupedInputLegacy)in;
  KeyValuesReader kvReader = shuffleInput.getReader();

  KeyValueWriter kvWriter = null;
  if((out instanceof MROutputLegacy)) {
    kvWriter = ((MROutputLegacy) out).getWriter();
  } else if ((out instanceof OrderedPartitionedKVOutput)) {
    kvWriter = ((OrderedPartitionedKVOutput) out).getWriter();
  } else {
    throw new IOException("Illegal output to reduce: " + in.getClass());
  }

  if (useNewApi) {
    try {
      runNewReducer(
          jobConf,
          mrReporter,
          shuffleInput, comparator,  keyClass, valueClass,
          kvWriter);
    } catch (ClassNotFoundException cnfe) {
      throw new IOException(cnfe);
    }
  } else {
    runOldReducer(
        jobConf, mrReporter,
        kvReader, comparator, keyClass, valueClass, kvWriter);
  }

  done();
}