Examples with CarbonRowBatch org.apache.carbondata.processing.loading.row.CarbonRowBatch used on opensource projects

Example 1 with CarbonRowBatch

use of org.apache.carbondata.processing.loading.row.CarbonRowBatch in project carbondata by apache.

the class DataWriterBatchProcessorStepImpl method execute.

@Override
public Iterator<CarbonRowBatch>[] execute() throws CarbonDataLoadingException {
    Iterator<CarbonRowBatch>[] iterators = child.execute();
    CarbonTableIdentifier tableIdentifier = configuration.getTableIdentifier().getCarbonTableIdentifier();
    String tableName = tableIdentifier.getTableName();
    try {
        CarbonTimeStatisticsFactory.getLoadStatisticsInstance().recordDictionaryValue2MdkAdd2FileTime(CarbonTablePath.DEPRECATED_PATITION_ID, System.currentTimeMillis());
        int i = 0;
        String[] storeLocation = getStoreLocation(tableIdentifier);
        CarbonDataProcessorUtil.createLocations(storeLocation);
        for (Iterator<CarbonRowBatch> iterator : iterators) {
            int k = 0;
            while (iterator.hasNext()) {
                CarbonRowBatch next = iterator.next();
                // If no rows from merge sorter, then don't create a file in fact column handler
                if (next.hasNext()) {
                    CarbonFactDataHandlerModel model = CarbonFactDataHandlerModel.createCarbonFactDataHandlerModel(configuration, storeLocation, 0, k++);
                    CarbonFactHandler dataHandler = CarbonFactHandlerFactory.createCarbonFactHandler(model, CarbonFactHandlerFactory.FactHandlerType.COLUMNAR);
                    dataHandler.initialise();
                    processBatch(next, dataHandler);
                    finish(tableName, dataHandler);
                }
            }
            i++;
        }
    } catch (Exception e) {
        LOGGER.error(e, "Failed for table: " + tableName + " in DataWriterBatchProcessorStepImpl");
        if (e.getCause() instanceof BadRecordFoundException) {
            throw new BadRecordFoundException(e.getCause().getMessage());
        }
        throw new CarbonDataLoadingException("There is an unexpected error: " + e.getMessage());
    }
    return null;
}

Example 2 with CarbonRowBatch

use of org.apache.carbondata.processing.loading.row.CarbonRowBatch in project carbondata by apache.

the class DataWriterProcessorStepImpl method execute.

@Override
public Iterator<CarbonRowBatch>[] execute() throws CarbonDataLoadingException {
    Iterator<CarbonRowBatch>[] iterators = child.execute();
    CarbonTableIdentifier tableIdentifier = configuration.getTableIdentifier().getCarbonTableIdentifier();
    String tableName = tableIdentifier.getTableName();
    try {
        CarbonTimeStatisticsFactory.getLoadStatisticsInstance().recordDictionaryValue2MdkAdd2FileTime(CarbonTablePath.DEPRECATED_PATITION_ID, System.currentTimeMillis());
        ExecutorService rangeExecutorService = Executors.newFixedThreadPool(iterators.length, new CarbonThreadFactory("WriterForwardPool: " + tableName));
        List<Future<Void>> rangeExecutorServiceSubmitList = new ArrayList<>(iterators.length);
        int i = 0;
        // do this concurrently
        for (Iterator<CarbonRowBatch> iterator : iterators) {
            String[] storeLocation = getStoreLocation(tableIdentifier);
            CarbonFactDataHandlerModel model = CarbonFactDataHandlerModel.createCarbonFactDataHandlerModel(configuration, storeLocation, i, 0);
            CarbonFactHandler dataHandler = null;
            boolean rowsNotExist = true;
            while (iterator.hasNext()) {
                if (rowsNotExist) {
                    rowsNotExist = false;
                    dataHandler = CarbonFactHandlerFactory.createCarbonFactHandler(model, CarbonFactHandlerFactory.FactHandlerType.COLUMNAR);
                    dataHandler.initialise();
                }
                processBatch(iterator.next(), dataHandler);
            }
            if (!rowsNotExist) {
                finish(dataHandler);
            }
            rangeExecutorServiceSubmitList.add(rangeExecutorService.submit(new WriterForwarder(iterator, tableIdentifier, i)));
            i++;
        }
        try {
            rangeExecutorService.shutdown();
            rangeExecutorService.awaitTermination(2, TimeUnit.DAYS);
            for (int j = 0; j < rangeExecutorServiceSubmitList.size(); j++) {
                rangeExecutorServiceSubmitList.get(j).get();
            }
        } catch (InterruptedException | ExecutionException e) {
            throw new CarbonDataWriterException(e);
        }
    } catch (CarbonDataWriterException e) {
        LOGGER.error(e, "Failed for table: " + tableName + " in DataWriterProcessorStepImpl");
        throw new CarbonDataLoadingException("Error while initializing data handler : " + e.getMessage());
    } catch (Exception e) {
        LOGGER.error(e, "Failed for table: " + tableName + " in DataWriterProcessorStepImpl");
        throw new CarbonDataLoadingException("There is an unexpected error: " + e.getMessage(), e);
    }
    return null;
}

Example 3 with CarbonRowBatch

use of org.apache.carbondata.processing.loading.row.CarbonRowBatch 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 (MemoryException e) {
        throw new CarbonDataLoadingException(e);
    }
    this.executorService = Executors.newFixedThreadPool(iterators.length, new CarbonThreadFactory("UnsafeParallelSorterPool:" + sortParameters.getTableName()));
    this.threadStatusObserver = new ThreadStatusObserver(executorService);
    try {
        for (int i = 0; i < iterators.length; i++) {
            executorService.execute(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 };
}

Example 4 with CarbonRowBatch

use of org.apache.carbondata.processing.loading.row.CarbonRowBatch 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, new CarbonThreadFactory("SafeParallelSorterPool:" + sortParameters.getTableName()));
    this.threadStatusObserver = new ThreadStatusObserver(executorService);
    try {
        for (int i = 0; i < iterators.length; i++) {
            executorService.execute(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 };
}