Examples with SortIntermediateFileMerger org.apache.carbondata.processing.sortandgroupby.sortdata.SortIntermediateFileMerger used on opensource projects

Example 1 with SortIntermediateFileMerger

use of org.apache.carbondata.processing.sortandgroupby.sortdata.SortIntermediateFileMerger in project carbondata by apache.

the class ParallelReadMergeSorterImpl method initialize.

@Override
public void initialize(SortParameters sortParameters) {
    this.sortParameters = sortParameters;
    intermediateFileMerger = new SortIntermediateFileMerger(sortParameters);
    String storeLocation = CarbonDataProcessorUtil.getLocalDataFolderLocation(sortParameters.getDatabaseName(), sortParameters.getTableName(), String.valueOf(sortParameters.getTaskNo()), sortParameters.getPartitionID(), sortParameters.getSegmentId() + "", false);
    // Set the data file location
    String dataFolderLocation = storeLocation + File.separator + CarbonCommonConstants.SORT_TEMP_FILE_LOCATION;
    finalMerger = new SingleThreadFinalSortFilesMerger(dataFolderLocation, sortParameters.getTableName(), sortParameters.getDimColCount(), sortParameters.getComplexDimColCount(), sortParameters.getMeasureColCount(), sortParameters.getNoDictionaryCount(), sortParameters.getMeasureDataType(), sortParameters.getNoDictionaryDimnesionColumn(), sortParameters.getNoDictionarySortColumn());
}

Example 2 with SortIntermediateFileMerger

use of org.apache.carbondata.processing.sortandgroupby.sortdata.SortIntermediateFileMerger in project carbondata by apache.

the class ParallelReadMergeSorterWithBucketingImpl method sort.

@Override
public Iterator<CarbonRowBatch>[] sort(Iterator<CarbonRowBatch>[] iterators) throws CarbonDataLoadingException {
    SortDataRows[] sortDataRows = new SortDataRows[bucketingInfo.getNumberOfBuckets()];
    intermediateFileMergers = new SortIntermediateFileMerger[sortDataRows.length];
    try {
        for (int i = 0; i < bucketingInfo.getNumberOfBuckets(); i++) {
            SortParameters parameters = sortParameters.getCopy();
            parameters.setPartitionID(i + "");
            setTempLocation(parameters);
            parameters.setBufferSize(sortBufferSize);
            intermediateFileMergers[i] = new SortIntermediateFileMerger(parameters);
            sortDataRows[i] = new SortDataRows(parameters, intermediateFileMergers[i]);
            sortDataRows[i].initialize();
        }
    } catch (CarbonSortKeyAndGroupByException e) {
        throw new CarbonDataLoadingException(e);
    }
    this.executorService = Executors.newFixedThreadPool(iterators.length);
    this.threadStatusObserver = new ThreadStatusObserver(this.executorService);
    final int batchSize = CarbonProperties.getInstance().getBatchSize();
    try {
        for (int i = 0; i < iterators.length; i++) {
            executorService.submit(new SortIteratorThread(iterators[i], sortDataRows, rowCounter, this.threadStatusObserver));
        }
        executorService.shutdown();
        executorService.awaitTermination(2, TimeUnit.DAYS);
        processRowToNextStep(sortDataRows, sortParameters);
    } catch (Exception e) {
        checkError();
        throw new CarbonDataLoadingException("Problem while shutdown the server ", e);
    }
    checkError();
    try {
        for (int i = 0; i < intermediateFileMergers.length; i++) {
            intermediateFileMergers[i].finish();
        }
    } catch (CarbonDataWriterException e) {
        throw new CarbonDataLoadingException(e);
    } catch (CarbonSortKeyAndGroupByException e) {
        throw new CarbonDataLoadingException(e);
    }
    Iterator<CarbonRowBatch>[] batchIterator = new Iterator[bucketingInfo.getNumberOfBuckets()];
    for (int i = 0; i < bucketingInfo.getNumberOfBuckets(); i++) {
        batchIterator[i] = new MergedDataIterator(String.valueOf(i), batchSize);
    }
    return batchIterator;
}