use of org.apache.carbondata.core.datastore.row.CarbonRow 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 };
}
use of org.apache.carbondata.core.datastore.row.CarbonRow 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 };
}
use of org.apache.carbondata.core.datastore.row.CarbonRow in project carbondata by apache.
the class CarbonRowDataWriterProcessorStepImpl method processBatch.
private void processBatch(CarbonRowBatch batch, CarbonFactHandler dataHandler, int iteratorIndex) throws CarbonDataLoadingException {
try {
while (batch.hasNext()) {
CarbonRow row = batch.next();
CarbonRow converted = convertRow(row);
dataHandler.addDataToStore(converted);
readCounter[iteratorIndex]++;
}
writeCounter[iteratorIndex] += batch.getSize();
} catch (Exception e) {
throw new CarbonDataLoadingException("unable to generate the mdkey", e);
}
rowCounter.getAndAdd(batch.getSize());
}
use of org.apache.carbondata.core.datastore.row.CarbonRow in project carbondata by apache.
the class CompactionResultSortProcessor method readAndLoadDataFromSortTempFiles.
/**
* This method will read sort temp files, perform merge sort and add it to store for data loading
*/
private void readAndLoadDataFromSortTempFiles() throws Exception {
try {
intermediateFileMerger.finish();
finalMerger.startFinalMerge();
while (finalMerger.hasNext()) {
Object[] row = finalMerger.next();
dataHandler.addDataToStore(new CarbonRow(row));
}
dataHandler.finish();
} catch (CarbonDataWriterException e) {
LOGGER.error(e);
throw new Exception("Problem loading data during compaction.", e);
} catch (Exception e) {
LOGGER.error(e);
throw new Exception("Problem loading data during compaction.", e);
} finally {
if (null != dataHandler) {
try {
dataHandler.closeHandler();
} catch (CarbonDataWriterException e) {
LOGGER.error(e, "Error in close data handler");
throw new Exception("Error in close data handler", e);
}
}
}
}
Aggregations