Search in sources :

Example 66 with AsterixException

use of org.apache.asterix.common.exceptions.AsterixException in project asterixdb by apache.

the class DatasetUtil method compactDatasetJobSpec.

public static JobSpecification compactDatasetJobSpec(Dataverse dataverse, String datasetName, MetadataProvider metadataProvider) throws AlgebricksException {
    String dataverseName = dataverse.getDataverseName();
    Dataset dataset = metadataProvider.findDataset(dataverseName, datasetName);
    if (dataset == null) {
        throw new AsterixException("Could not find dataset " + datasetName + " in dataverse " + dataverseName);
    }
    JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
    Pair<IFileSplitProvider, AlgebricksPartitionConstraint> splitsAndConstraint = metadataProvider.getSplitProviderAndConstraints(dataset);
    IIndexDataflowHelperFactory indexHelperFactory = new IndexDataflowHelperFactory(metadataProvider.getStorageComponentProvider().getStorageManager(), splitsAndConstraint.first);
    LSMTreeIndexCompactOperatorDescriptor compactOp = new LSMTreeIndexCompactOperatorDescriptor(spec, indexHelperFactory);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, compactOp, splitsAndConstraint.second);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, compactOp, splitsAndConstraint.second);
    spec.addRoot(compactOp);
    return spec;
}
Also used : LSMTreeIndexCompactOperatorDescriptor(org.apache.hyracks.storage.am.lsm.common.dataflow.LSMTreeIndexCompactOperatorDescriptor) AsterixException(org.apache.asterix.common.exceptions.AsterixException) Dataset(org.apache.asterix.metadata.entities.Dataset) IFileSplitProvider(org.apache.hyracks.dataflow.std.file.IFileSplitProvider) IIndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory) AlgebricksPartitionConstraint(org.apache.hyracks.algebricks.common.constraints.AlgebricksPartitionConstraint) AMutableString(org.apache.asterix.om.base.AMutableString) AString(org.apache.asterix.om.base.AString) JobSpecification(org.apache.hyracks.api.job.JobSpecification) IIndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory) IndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IndexDataflowHelperFactory)

Example 67 with AsterixException

use of org.apache.asterix.common.exceptions.AsterixException in project asterixdb by apache.

the class InvertedIndexResourceFactoryProvider method getResourceFactory.

@Override
public IResourceFactory getResourceFactory(MetadataProvider mdProvider, Dataset dataset, Index index, ARecordType recordType, ARecordType metaType, ILSMMergePolicyFactory mergePolicyFactory, Map<String, String> mergePolicyProperties, ITypeTraits[] filterTypeTraits, IBinaryComparatorFactory[] filterCmpFactories) throws AlgebricksException {
    // Get basic info
    List<List<String>> primaryKeys = dataset.getPrimaryKeys();
    List<List<String>> secondaryKeys = index.getKeyFieldNames();
    List<String> filterFieldName = DatasetUtil.getFilterField(dataset);
    int numPrimaryKeys = primaryKeys.size();
    int numSecondaryKeys = secondaryKeys.size();
    // Validate
    if (dataset.getDatasetType() != DatasetType.INTERNAL) {
        throw new CompilationException(ErrorCode.COMPILATION_INDEX_TYPE_NOT_SUPPORTED_FOR_DATASET_TYPE, index.getIndexType().name(), dataset.getDatasetType());
    }
    if (numPrimaryKeys > 1) {
        throw new AsterixException("Cannot create inverted index on dataset with composite primary key.");
    }
    if (numSecondaryKeys > 1) {
        throw new AsterixException("Cannot create composite inverted index on multiple fields.");
    }
    boolean isPartitioned = index.getIndexType() == IndexType.LENGTH_PARTITIONED_WORD_INVIX || index.getIndexType() == IndexType.LENGTH_PARTITIONED_NGRAM_INVIX;
    int numTokenKeyPairFields = (!isPartitioned) ? 1 + numPrimaryKeys : 2 + numPrimaryKeys;
    int[] invertedIndexFields = null;
    int[] secondaryFilterFieldsForNonBulkLoadOps = null;
    int[] invertedIndexFieldsForNonBulkLoadOps = null;
    int[] secondaryFilterFields = null;
    if (filterFieldName != null) {
        invertedIndexFields = new int[numTokenKeyPairFields];
        for (int i = 0; i < invertedIndexFields.length; i++) {
            invertedIndexFields[i] = i;
        }
        secondaryFilterFieldsForNonBulkLoadOps = new int[filterFieldName.size()];
        secondaryFilterFieldsForNonBulkLoadOps[0] = numSecondaryKeys + numPrimaryKeys;
        invertedIndexFieldsForNonBulkLoadOps = new int[numSecondaryKeys + numPrimaryKeys];
        for (int i = 0; i < invertedIndexFieldsForNonBulkLoadOps.length; i++) {
            invertedIndexFieldsForNonBulkLoadOps[i] = i;
        }
        secondaryFilterFields = new int[filterFieldName.size()];
        secondaryFilterFields[0] = numTokenKeyPairFields - numPrimaryKeys + numPrimaryKeys;
    }
    IStorageComponentProvider storageComponentProvider = mdProvider.getStorageComponentProvider();
    IStorageManager storageManager = storageComponentProvider.getStorageManager();
    ILSMOperationTrackerFactory opTrackerFactory = dataset.getIndexOperationTrackerFactory(index);
    ILSMIOOperationCallbackFactory ioOpCallbackFactory = dataset.getIoOperationCallbackFactory(index);
    IMetadataPageManagerFactory metadataPageManagerFactory = storageComponentProvider.getMetadataPageManagerFactory();
    AsterixVirtualBufferCacheProvider vbcProvider = new AsterixVirtualBufferCacheProvider(dataset.getDatasetId());
    ILSMIOOperationSchedulerProvider ioSchedulerProvider = storageComponentProvider.getIoOperationSchedulerProvider();
    boolean durable = !dataset.isTemp();
    double bloomFilterFalsePositiveRate = mdProvider.getStorageProperties().getBloomFilterFalsePositiveRate();
    ITypeTraits[] typeTraits = getInvListTypeTraits(mdProvider, dataset, recordType, metaType);
    IBinaryComparatorFactory[] cmpFactories = getInvListComparatorFactories(mdProvider, dataset, recordType, metaType);
    ITypeTraits[] tokenTypeTraits = getTokenTypeTraits(dataset, index, recordType, metaType);
    IBinaryComparatorFactory[] tokenCmpFactories = getTokenComparatorFactories(dataset, index, recordType, metaType);
    IBinaryTokenizerFactory tokenizerFactory = getTokenizerFactory(dataset, index, recordType, metaType);
    return new LSMInvertedIndexLocalResourceFactory(storageManager, typeTraits, cmpFactories, filterTypeTraits, filterCmpFactories, secondaryFilterFields, opTrackerFactory, ioOpCallbackFactory, metadataPageManagerFactory, vbcProvider, ioSchedulerProvider, mergePolicyFactory, mergePolicyProperties, durable, tokenTypeTraits, tokenCmpFactories, tokenizerFactory, isPartitioned, invertedIndexFields, secondaryFilterFieldsForNonBulkLoadOps, invertedIndexFieldsForNonBulkLoadOps, bloomFilterFalsePositiveRate);
}
Also used : CompilationException(org.apache.asterix.common.exceptions.CompilationException) ILSMIOOperationCallbackFactory(org.apache.hyracks.storage.am.lsm.common.api.ILSMIOOperationCallbackFactory) LSMInvertedIndexLocalResourceFactory(org.apache.hyracks.storage.am.lsm.invertedindex.dataflow.LSMInvertedIndexLocalResourceFactory) IStorageComponentProvider(org.apache.asterix.common.context.IStorageComponentProvider) ITypeTraits(org.apache.hyracks.api.dataflow.value.ITypeTraits) IMetadataPageManagerFactory(org.apache.hyracks.storage.am.common.api.IMetadataPageManagerFactory) AsterixVirtualBufferCacheProvider(org.apache.asterix.common.context.AsterixVirtualBufferCacheProvider) IBinaryComparatorFactory(org.apache.hyracks.api.dataflow.value.IBinaryComparatorFactory) ILSMOperationTrackerFactory(org.apache.hyracks.storage.am.lsm.common.api.ILSMOperationTrackerFactory) IStorageManager(org.apache.hyracks.storage.common.IStorageManager) AsterixException(org.apache.asterix.common.exceptions.AsterixException) ILSMIOOperationSchedulerProvider(org.apache.hyracks.storage.am.lsm.common.api.ILSMIOOperationSchedulerProvider) IBinaryTokenizerFactory(org.apache.hyracks.storage.am.lsm.invertedindex.tokenizers.IBinaryTokenizerFactory) List(java.util.List)

Aggregations

AsterixException (org.apache.asterix.common.exceptions.AsterixException)67 IOException (java.io.IOException)27 HyracksDataException (org.apache.hyracks.api.exceptions.HyracksDataException)26 DataOutput (java.io.DataOutput)15 IPointable (org.apache.hyracks.data.std.api.IPointable)15 IFrameTupleReference (org.apache.hyracks.dataflow.common.data.accessors.IFrameTupleReference)15 TypeMismatchException (org.apache.asterix.runtime.exceptions.TypeMismatchException)14 IScalarEvaluator (org.apache.hyracks.algebricks.runtime.base.IScalarEvaluator)14 VoidPointable (org.apache.hyracks.data.std.primitive.VoidPointable)14 ArrayBackedValueStorage (org.apache.hyracks.data.std.util.ArrayBackedValueStorage)14 ATypeTag (org.apache.asterix.om.types.ATypeTag)10 IAType (org.apache.asterix.om.types.IAType)10 ARecordType (org.apache.asterix.om.types.ARecordType)9 IHyracksTaskContext (org.apache.hyracks.api.context.IHyracksTaskContext)9 ISerializerDeserializer (org.apache.hyracks.api.dataflow.value.ISerializerDeserializer)9 AlgebricksException (org.apache.hyracks.algebricks.common.exceptions.AlgebricksException)8 IScalarEvaluatorFactory (org.apache.hyracks.algebricks.runtime.base.IScalarEvaluatorFactory)8 List (java.util.List)7 InputStream (java.io.InputStream)5 CompilationException (org.apache.asterix.common.exceptions.CompilationException)5