Examples with UnsafeSortDataRows org.apache.carbondata.processing.loading.sort.unsafe.UnsafeSortDataRows used on opensource projects

Example 1 with UnsafeSortDataRows

use of org.apache.carbondata.processing.loading.sort.unsafe.UnsafeSortDataRows in project carbondata by apache.

the class UnsafeParallelReadMergeSorterWithColumnRangeImpl method sort.

@Override
public Iterator<CarbonRowBatch>[] sort(Iterator<CarbonRowBatch>[] iterators) throws CarbonDataLoadingException {
    UnsafeSortDataRows[] sortDataRows = new UnsafeSortDataRows[columnRangeInfo.getNumOfRanges()];
    intermediateFileMergers = new UnsafeIntermediateMerger[columnRangeInfo.getNumOfRanges()];
    SortParameters[] sortParameterArray = new SortParameters[columnRangeInfo.getNumOfRanges()];
    try {
        for (int i = 0; i < columnRangeInfo.getNumOfRanges(); i++) {
            SortParameters parameters = originSortParameters.getCopy();
            parameters.setPartitionID(i + "");
            parameters.setRangeId(i);
            sortParameterArray[i] = parameters;
            setTempLocation(parameters);
            intermediateFileMergers[i] = new UnsafeIntermediateMerger(parameters);
            sortDataRows[i] = new UnsafeSortDataRows(parameters, intermediateFileMergers[i], inMemoryChunkSizeInMB);
            sortDataRows[i].initialize();
        }
    } catch (Exception e) {
        throw new CarbonDataLoadingException(e);
    }
    ExecutorService executorService = Executors.newFixedThreadPool(iterators.length);
    this.threadStatusObserver = new ThreadStatusObserver(executorService);
    final int batchSize = CarbonProperties.getInstance().getBatchSize();
    try {
        for (int i = 0; i < iterators.length; i++) {
            executorService.execute(new SortIteratorThread(iterators[i], sortDataRows, rowCounter, this.insideRowCounterList, this.threadStatusObserver));
        }
        executorService.shutdown();
        executorService.awaitTermination(2, TimeUnit.DAYS);
        processRowToNextStep(sortDataRows, originSortParameters);
    } 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 (Exception e) {
        throw new CarbonDataLoadingException(e);
    }
    Iterator<CarbonRowBatch>[] batchIterator = new Iterator[columnRangeInfo.getNumOfRanges()];
    for (int i = 0; i < sortDataRows.length; i++) {
        batchIterator[i] = new MergedDataIterator(sortParameterArray[i], batchSize, intermediateFileMergers[i]);
    }
    return batchIterator;
}

Example 2 with UnsafeSortDataRows

use of org.apache.carbondata.processing.loading.sort.unsafe.UnsafeSortDataRows in project carbondata by apache.

the class UnsafeParallelReadMergeSorterImpl method sort.

@Override
public Iterator<CarbonRowBatch>[] sort(Iterator<CarbonRowBatch>[] iterators) throws CarbonDataLoadingException {
    int inMemoryChunkSizeInMB = CarbonProperties.getInstance().getSortMemoryChunkSizeInMB();
    final int batchSize = CarbonProperties.getInstance().getBatchSize();
    this.executorService = Executors.newFixedThreadPool(sortParameters.getNumberOfCores(), new CarbonThreadFactory("UnsafeParallelSorterPool:" + sortParameters.getTableName(), true));
    this.threadStatusObserver = new ThreadStatusObserver(executorService);
    try {
        for (int i = 0; i < iterators.length; i++) {
            UnsafeSortDataRows sortDataRows = new UnsafeSortDataRows(sortParameters, unsafeIntermediateFileMerger, inMemoryChunkSizeInMB);
            sortDataRows.setInstanceId(i);
            executorService.execute(new SortIteratorThread(iterators[i], sortDataRows, batchSize, rowCounter, this.threadStatusObserver));
        }
        executorService.shutdown();
        executorService.awaitTermination(2, TimeUnit.DAYS);
        if (!sortParameters.getObserver().isFailed()) {
            LOGGER.info("Record Processed For table: " + sortParameters.getTableName());
            CarbonTimeStatisticsFactory.getLoadStatisticsInstance().recordSortRowsStepTotalTime(sortParameters.getPartitionID(), System.currentTimeMillis());
            CarbonTimeStatisticsFactory.getLoadStatisticsInstance().recordDictionaryValuesTotalTime(sortParameters.getPartitionID(), System.currentTimeMillis());
        }
    } catch (Exception e) {
        checkError();
        throw new CarbonDataLoadingException("Problem while shutdown the server ", e);
    }
    checkError();
    try {
        unsafeIntermediateFileMerger.finish();
        List<UnsafeCarbonRowPage> rowPages = unsafeIntermediateFileMerger.getRowPages();
        finalMerger.startFinalMerge(rowPages.toArray(new UnsafeCarbonRowPage[rowPages.size()]), unsafeIntermediateFileMerger.getMergedPages());
    } catch (CarbonDataWriterException e) {
        throw new CarbonDataLoadingException(e);
    } catch (CarbonSortKeyAndGroupByException e) {
        throw new CarbonDataLoadingException(e);
    }
    // Creates the iterator to read from merge sorter.
    Iterator<CarbonRowBatch> batchIterator = new CarbonIterator<CarbonRowBatch>() {

        @Override
        public boolean hasNext() {
            return finalMerger.hasNext();
        }

        @Override
        public CarbonRowBatch next() {
            int counter = 0;
            CarbonRowBatch rowBatch = new CarbonRowBatch(batchSize);
            while (finalMerger.hasNext() && counter < batchSize) {
                rowBatch.addRow(new CarbonRow(finalMerger.next()));
                counter++;
            }
            return rowBatch;
        }
    };
    return new Iterator[] { batchIterator };
}