Search in sources :

Example 1 with UnsafeSortDataRows

use of org.apache.carbondata.processing.newflow.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();
    UnsafeSortDataRows sortDataRow = new UnsafeSortDataRows(sortParameters, unsafeIntermediateFileMerger, inMemoryChunkSizeInMB);
    final int batchSize = CarbonProperties.getInstance().getBatchSize();
    try {
        sortDataRow.initialize();
    } catch (CarbonSortKeyAndGroupByException e) {
        throw new CarbonDataLoadingException(e);
    }
    this.executorService = Executors.newFixedThreadPool(iterators.length);
    this.threadStatusObserver = new ThreadStatusObserver(this.executorService);
    try {
        for (int i = 0; i < iterators.length; i++) {
            executorService.submit(new SortIteratorThread(iterators[i], sortDataRow, batchSize, rowCounter, this.threadStatusObserver));
        }
        executorService.shutdown();
        executorService.awaitTermination(2, TimeUnit.DAYS);
        processRowToNextStep(sortDataRow, sortParameters);
    } 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 };
}
Also used : CarbonRowBatch(org.apache.carbondata.processing.newflow.row.CarbonRowBatch) CarbonDataLoadingException(org.apache.carbondata.processing.newflow.exception.CarbonDataLoadingException) CarbonRow(org.apache.carbondata.processing.newflow.row.CarbonRow) UnsafeCarbonRowPage(org.apache.carbondata.processing.newflow.sort.unsafe.UnsafeCarbonRowPage) CarbonDataWriterException(org.apache.carbondata.processing.store.writer.exception.CarbonDataWriterException) CarbonSortKeyAndGroupByException(org.apache.carbondata.processing.sortandgroupby.exception.CarbonSortKeyAndGroupByException) CarbonDataWriterException(org.apache.carbondata.processing.store.writer.exception.CarbonDataWriterException) CarbonDataLoadingException(org.apache.carbondata.processing.newflow.exception.CarbonDataLoadingException) CarbonIterator(org.apache.carbondata.common.CarbonIterator) CarbonIterator(org.apache.carbondata.common.CarbonIterator) Iterator(java.util.Iterator) CarbonSortKeyAndGroupByException(org.apache.carbondata.processing.sortandgroupby.exception.CarbonSortKeyAndGroupByException) UnsafeSortDataRows(org.apache.carbondata.processing.newflow.sort.unsafe.UnsafeSortDataRows)

Example 2 with UnsafeSortDataRows

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

the class UnsafeParallelReadMergeSorterWithBucketingImpl method sort.

@Override
public Iterator<CarbonRowBatch>[] sort(Iterator<CarbonRowBatch>[] iterators) throws CarbonDataLoadingException {
    UnsafeSortDataRows[] sortDataRows = new UnsafeSortDataRows[bucketingInfo.getNumberOfBuckets()];
    UnsafeIntermediateMerger[] intermediateFileMergers = new UnsafeIntermediateMerger[sortDataRows.length];
    int inMemoryChunkSizeInMB = CarbonProperties.getInstance().getSortMemoryChunkSizeInMB();
    inMemoryChunkSizeInMB = inMemoryChunkSizeInMB / bucketingInfo.getNumberOfBuckets();
    if (inMemoryChunkSizeInMB < 5) {
        inMemoryChunkSizeInMB = 5;
    }
    try {
        for (int i = 0; i < bucketingInfo.getNumberOfBuckets(); i++) {
            SortParameters parameters = sortParameters.getCopy();
            parameters.setPartitionID(i + "");
            setTempLocation(parameters);
            intermediateFileMergers[i] = new UnsafeIntermediateMerger(parameters);
            sortDataRows[i] = new UnsafeSortDataRows(parameters, intermediateFileMergers[i], inMemoryChunkSizeInMB);
            sortDataRows[i].initialize();
        }
    } catch (CarbonSortKeyAndGroupByException e) {
        throw new CarbonDataLoadingException(e);
    }
    this.executorService = Executors.newFixedThreadPool(iterators.length);
    final int batchSize = CarbonProperties.getInstance().getBatchSize();
    try {
        for (int i = 0; i < iterators.length; i++) {
            executorService.submit(new SortIteratorThread(iterators[i], sortDataRows));
        }
        executorService.shutdown();
        executorService.awaitTermination(2, TimeUnit.DAYS);
        processRowToNextStep(sortDataRows, sortParameters);
    } catch (Exception e) {
        throw new CarbonDataLoadingException("Problem while shutdown the server ", e);
    }
    try {
        for (int i = 0; i < intermediateFileMergers.length; i++) {
            intermediateFileMergers[i].finish();
        }
    } catch (Exception e) {
        throw new CarbonDataLoadingException(e);
    }
    Iterator<CarbonRowBatch>[] batchIterator = new Iterator[bucketingInfo.getNumberOfBuckets()];
    for (int i = 0; i < sortDataRows.length; i++) {
        batchIterator[i] = new MergedDataIterator(String.valueOf(i), batchSize, intermediateFileMergers[i]);
    }
    return batchIterator;
}
Also used : CarbonDataLoadingException(org.apache.carbondata.processing.newflow.exception.CarbonDataLoadingException) UnsafeIntermediateMerger(org.apache.carbondata.processing.newflow.sort.unsafe.merger.UnsafeIntermediateMerger) CarbonSortKeyAndGroupByException(org.apache.carbondata.processing.sortandgroupby.exception.CarbonSortKeyAndGroupByException) CarbonDataLoadingException(org.apache.carbondata.processing.newflow.exception.CarbonDataLoadingException) SortParameters(org.apache.carbondata.processing.sortandgroupby.sortdata.SortParameters) CarbonIterator(org.apache.carbondata.common.CarbonIterator) Iterator(java.util.Iterator) CarbonSortKeyAndGroupByException(org.apache.carbondata.processing.sortandgroupby.exception.CarbonSortKeyAndGroupByException) UnsafeSortDataRows(org.apache.carbondata.processing.newflow.sort.unsafe.UnsafeSortDataRows)

Aggregations

Iterator (java.util.Iterator)2 CarbonIterator (org.apache.carbondata.common.CarbonIterator)2 CarbonDataLoadingException (org.apache.carbondata.processing.newflow.exception.CarbonDataLoadingException)2 UnsafeSortDataRows (org.apache.carbondata.processing.newflow.sort.unsafe.UnsafeSortDataRows)2 CarbonSortKeyAndGroupByException (org.apache.carbondata.processing.sortandgroupby.exception.CarbonSortKeyAndGroupByException)2 CarbonRow (org.apache.carbondata.processing.newflow.row.CarbonRow)1 CarbonRowBatch (org.apache.carbondata.processing.newflow.row.CarbonRowBatch)1 UnsafeCarbonRowPage (org.apache.carbondata.processing.newflow.sort.unsafe.UnsafeCarbonRowPage)1 UnsafeIntermediateMerger (org.apache.carbondata.processing.newflow.sort.unsafe.merger.UnsafeIntermediateMerger)1 SortParameters (org.apache.carbondata.processing.sortandgroupby.sortdata.SortParameters)1 CarbonDataWriterException (org.apache.carbondata.processing.store.writer.exception.CarbonDataWriterException)1