Search in sources :

Example 1 with SortDataRows

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

the class ParallelReadMergeSorterImpl method sort.

@Override
public Iterator<CarbonRowBatch>[] sort(Iterator<CarbonRowBatch>[] iterators) throws CarbonDataLoadingException {
    SortDataRows sortDataRow = new SortDataRows(sortParameters, intermediateFileMerger);
    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(executorService);
    try {
        for (int i = 0; i < iterators.length; i++) {
            executorService.submit(new SortIteratorThread(iterators[i], sortDataRow, batchSize, rowCounter, 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 {
        intermediateFileMerger.finish();
        intermediateFileMerger = null;
        finalMerger.startFinalMerge();
    } 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) 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) SortDataRows(org.apache.carbondata.processing.sortandgroupby.sortdata.SortDataRows) CarbonIterator(org.apache.carbondata.common.CarbonIterator) CarbonIterator(org.apache.carbondata.common.CarbonIterator) Iterator(java.util.Iterator) CarbonSortKeyAndGroupByException(org.apache.carbondata.processing.sortandgroupby.exception.CarbonSortKeyAndGroupByException)

Example 2 with SortDataRows

use of org.apache.carbondata.processing.sortandgroupby.sortdata.SortDataRows 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;
}
Also used : CarbonDataLoadingException(org.apache.carbondata.processing.newflow.exception.CarbonDataLoadingException) CarbonDataWriterException(org.apache.carbondata.processing.store.writer.exception.CarbonDataWriterException) SortIntermediateFileMerger(org.apache.carbondata.processing.sortandgroupby.sortdata.SortIntermediateFileMerger) 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) SortDataRows(org.apache.carbondata.processing.sortandgroupby.sortdata.SortDataRows) 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)

Aggregations

Iterator (java.util.Iterator)2 CarbonIterator (org.apache.carbondata.common.CarbonIterator)2 CarbonDataLoadingException (org.apache.carbondata.processing.newflow.exception.CarbonDataLoadingException)2 CarbonSortKeyAndGroupByException (org.apache.carbondata.processing.sortandgroupby.exception.CarbonSortKeyAndGroupByException)2 SortDataRows (org.apache.carbondata.processing.sortandgroupby.sortdata.SortDataRows)2 CarbonDataWriterException (org.apache.carbondata.processing.store.writer.exception.CarbonDataWriterException)2 CarbonRow (org.apache.carbondata.processing.newflow.row.CarbonRow)1 CarbonRowBatch (org.apache.carbondata.processing.newflow.row.CarbonRowBatch)1 SortIntermediateFileMerger (org.apache.carbondata.processing.sortandgroupby.sortdata.SortIntermediateFileMerger)1 SortParameters (org.apache.carbondata.processing.sortandgroupby.sortdata.SortParameters)1