Examples with ConnectorPolicyAssignmentPolicy org.apache.hyracks.algebricks.core.jobgen.impl.ConnectorPolicyAssignmentPolicy used on opensource projects

Example 6 with ConnectorPolicyAssignmentPolicy

use of org.apache.hyracks.algebricks.core.jobgen.impl.ConnectorPolicyAssignmentPolicy in project asterixdb by apache.

the class SecondaryInvertedIndexOperationsHelper method buildLoadingJobSpec.

@Override
public JobSpecification buildLoadingJobSpec() throws AlgebricksException {
    JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
    JobId jobId = IndexUtil.bindJobEventListener(spec, metadataProvider);
    // Create dummy key provider for feeding the primary index scan.
    IOperatorDescriptor keyProviderOp = DatasetUtil.createDummyKeyProviderOp(spec, dataset, metadataProvider);
    // Create primary index scan op.
    IOperatorDescriptor primaryScanOp = DatasetUtil.createPrimaryIndexScanOp(spec, metadataProvider, dataset, jobId);
    IOperatorDescriptor sourceOp = primaryScanOp;
    boolean isEnforcingKeyTypes = index.isEnforcingKeyFileds();
    int numSecondaryKeys = index.getKeyFieldNames().size();
    if (isEnforcingKeyTypes && !enforcedItemType.equals(itemType)) {
        sourceOp = createCastOp(spec, dataset.getDatasetType());
        spec.connect(new OneToOneConnectorDescriptor(spec), primaryScanOp, 0, sourceOp, 0);
    }
    AlgebricksMetaOperatorDescriptor asterixAssignOp = createAssignOp(spec, numSecondaryKeys, secondaryRecDesc);
    // If any of the secondary fields are nullable, then add a select op
    // that filters nulls.
    AlgebricksMetaOperatorDescriptor selectOp = null;
    if (anySecondaryKeyIsNullable || isEnforcingKeyTypes) {
        selectOp = createFilterNullsSelectOp(spec, numSecondaryKeys, secondaryRecDesc);
    }
    // Create a tokenizer op.
    AbstractOperatorDescriptor tokenizerOp = createTokenizerOp(spec);
    // Sort by token + primary keys.
    ExternalSortOperatorDescriptor sortOp = createSortOp(spec, tokenKeyPairComparatorFactories, tokenKeyPairRecDesc);
    // Create secondary inverted index bulk load op.
    AbstractSingleActivityOperatorDescriptor invIndexBulkLoadOp = createInvertedIndexBulkLoadOp(spec);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, invIndexBulkLoadOp, secondaryPartitionConstraint);
    AlgebricksMetaOperatorDescriptor metaOp = new AlgebricksMetaOperatorDescriptor(spec, 1, 0, new IPushRuntimeFactory[] { new SinkRuntimeFactory() }, new RecordDescriptor[] {});
    // Connect the operators.
    spec.connect(new OneToOneConnectorDescriptor(spec), keyProviderOp, 0, primaryScanOp, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), sourceOp, 0, asterixAssignOp, 0);
    if (anySecondaryKeyIsNullable || isEnforcingKeyTypes) {
        spec.connect(new OneToOneConnectorDescriptor(spec), asterixAssignOp, 0, selectOp, 0);
        spec.connect(new OneToOneConnectorDescriptor(spec), selectOp, 0, tokenizerOp, 0);
    } else {
        spec.connect(new OneToOneConnectorDescriptor(spec), asterixAssignOp, 0, tokenizerOp, 0);
    }
    spec.connect(new OneToOneConnectorDescriptor(spec), tokenizerOp, 0, sortOp, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), sortOp, 0, invIndexBulkLoadOp, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), invIndexBulkLoadOp, 0, metaOp, 0);
    spec.addRoot(metaOp);
    spec.setConnectorPolicyAssignmentPolicy(new ConnectorPolicyAssignmentPolicy());
    return spec;
}

Example 7 with ConnectorPolicyAssignmentPolicy

use of org.apache.hyracks.algebricks.core.jobgen.impl.ConnectorPolicyAssignmentPolicy in project asterixdb by apache.

the class ExternalIndexingOperations method buildFilesIndexUpdateJobSpec.

public static JobSpecification buildFilesIndexUpdateJobSpec(Dataset dataset, List<ExternalFile> externalFilesSnapshot, MetadataProvider metadataProvider) throws AlgebricksException {
    IStorageComponentProvider storageComponentProvider = metadataProvider.getStorageComponentProvider();
    JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
    Pair<IFileSplitProvider, AlgebricksPartitionConstraint> secondarySplitsAndConstraint = metadataProvider.getSplitProviderAndConstraints(dataset, IndexingConstants.getFilesIndexName(dataset.getDatasetName()));
    IFileSplitProvider secondaryFileSplitProvider = secondarySplitsAndConstraint.first;
    IIndexDataflowHelperFactory dataflowHelperFactory = new IndexDataflowHelperFactory(storageComponentProvider.getStorageManager(), secondaryFileSplitProvider);
    ExternalFilesIndexModificationOperatorDescriptor externalFilesOp = new ExternalFilesIndexModificationOperatorDescriptor(spec, dataflowHelperFactory, externalFilesSnapshot);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, externalFilesOp, secondarySplitsAndConstraint.second);
    spec.addRoot(externalFilesOp);
    spec.setConnectorPolicyAssignmentPolicy(new ConnectorPolicyAssignmentPolicy());
    return spec;
}

Example 8 with ConnectorPolicyAssignmentPolicy

use of org.apache.hyracks.algebricks.core.jobgen.impl.ConnectorPolicyAssignmentPolicy in project asterixdb by apache.

the class ExternalIndexingOperations method buildRecoverOp.

public static JobSpecification buildRecoverOp(Dataset ds, List<Index> indexes, MetadataProvider metadataProvider) throws AlgebricksException {
    JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
    IStorageManager storageMgr = metadataProvider.getStorageComponentProvider().getStorageManager();
    ArrayList<IIndexDataflowHelperFactory> treeDataflowHelperFactories = new ArrayList<>();
    AlgebricksPartitionConstraint constraints = null;
    for (Index index : indexes) {
        IFileSplitProvider indexSplitProvider;
        if (isValidIndexName(index.getDatasetName(), index.getIndexName())) {
            Pair<IFileSplitProvider, AlgebricksPartitionConstraint> sAndConstraints = metadataProvider.getSplitProviderAndConstraints(ds, index.getIndexName());
            indexSplitProvider = sAndConstraints.first;
            constraints = sAndConstraints.second;
        } else {
            indexSplitProvider = metadataProvider.getSplitProviderAndConstraints(ds, IndexingConstants.getFilesIndexName(ds.getDatasetName())).first;
        }
        IIndexDataflowHelperFactory indexDataflowHelperFactory = new IndexDataflowHelperFactory(storageMgr, indexSplitProvider);
        treeDataflowHelperFactories.add(indexDataflowHelperFactory);
    }
    ExternalDatasetIndexesRecoverOperatorDescriptor op = new ExternalDatasetIndexesRecoverOperatorDescriptor(spec, treeDataflowHelperFactories);
    spec.addRoot(op);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, op, constraints);
    spec.setConnectorPolicyAssignmentPolicy(new ConnectorPolicyAssignmentPolicy());
    return spec;
}

Example 9 with ConnectorPolicyAssignmentPolicy

use of org.apache.hyracks.algebricks.core.jobgen.impl.ConnectorPolicyAssignmentPolicy in project asterixdb by apache.

the class ExternalIndexingOperations method buildCommitJob.

public static JobSpecification buildCommitJob(Dataset ds, List<Index> indexes, MetadataProvider metadataProvider) throws AlgebricksException {
    JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
    IStorageManager storageMgr = metadataProvider.getStorageComponentProvider().getStorageManager();
    ArrayList<IIndexDataflowHelperFactory> treeDataflowHelperFactories = new ArrayList<>();
    AlgebricksPartitionConstraint constraints = null;
    for (Index index : indexes) {
        IFileSplitProvider indexSplitProvider;
        if (isValidIndexName(index.getDatasetName(), index.getIndexName())) {
            Pair<IFileSplitProvider, AlgebricksPartitionConstraint> sAndConstraints = metadataProvider.getSplitProviderAndConstraints(ds, index.getIndexName());
            indexSplitProvider = sAndConstraints.first;
            constraints = sAndConstraints.second;
        } else {
            indexSplitProvider = metadataProvider.getSplitProviderAndConstraints(ds, IndexingConstants.getFilesIndexName(ds.getDatasetName())).first;
        }
        IIndexDataflowHelperFactory indexDataflowHelperFactory = new IndexDataflowHelperFactory(storageMgr, indexSplitProvider);
        treeDataflowHelperFactories.add(indexDataflowHelperFactory);
    }
    ExternalDatasetIndexesCommitOperatorDescriptor op = new ExternalDatasetIndexesCommitOperatorDescriptor(spec, treeDataflowHelperFactories);
    spec.addRoot(op);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, op, constraints);
    spec.setConnectorPolicyAssignmentPolicy(new ConnectorPolicyAssignmentPolicy());
    return spec;
}

Example 10 with ConnectorPolicyAssignmentPolicy

use of org.apache.hyracks.algebricks.core.jobgen.impl.ConnectorPolicyAssignmentPolicy in project asterixdb by apache.

the class ExternalIndexingOperations method buildFilesIndexCreateJobSpec.

public static JobSpecification buildFilesIndexCreateJobSpec(Dataset dataset, List<ExternalFile> externalFilesSnapshot, MetadataProvider metadataProvider) throws AlgebricksException {
    IStorageComponentProvider storageComponentProvider = metadataProvider.getStorageComponentProvider();
    JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
    Pair<ILSMMergePolicyFactory, Map<String, String>> compactionInfo = DatasetUtil.getMergePolicyFactory(dataset, metadataProvider.getMetadataTxnContext());
    ILSMMergePolicyFactory mergePolicyFactory = compactionInfo.first;
    Map<String, String> mergePolicyProperties = compactionInfo.second;
    Pair<IFileSplitProvider, AlgebricksPartitionConstraint> secondarySplitsAndConstraint = metadataProvider.getSplitProviderAndConstraints(dataset, IndexingConstants.getFilesIndexName(dataset.getDatasetName()));
    IFileSplitProvider secondaryFileSplitProvider = secondarySplitsAndConstraint.first;
    String fileIndexName = IndexingConstants.getFilesIndexName(dataset.getDatasetName());
    Index fileIndex = MetadataManager.INSTANCE.getIndex(metadataProvider.getMetadataTxnContext(), dataset.getDataverseName(), dataset.getDatasetName(), fileIndexName);
    ARecordType recordType = (ARecordType) metadataProvider.findType(dataset.getItemTypeDataverseName(), dataset.getItemTypeName());
    IResourceFactory resourceFactory = dataset.getResourceFactory(metadataProvider, fileIndex, recordType, null, mergePolicyFactory, mergePolicyProperties);
    IIndexBuilderFactory indexBuilderFactory = new IndexBuilderFactory(storageComponentProvider.getStorageManager(), secondaryFileSplitProvider, resourceFactory, !dataset.isTemp());
    IIndexDataflowHelperFactory dataflowHelperFactory = new IndexDataflowHelperFactory(storageComponentProvider.getStorageManager(), secondaryFileSplitProvider);
    ExternalFilesIndexCreateOperatorDescriptor externalFilesOp = new ExternalFilesIndexCreateOperatorDescriptor(spec, indexBuilderFactory, dataflowHelperFactory, externalFilesSnapshot);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, externalFilesOp, secondarySplitsAndConstraint.second);
    spec.addRoot(externalFilesOp);
    spec.setConnectorPolicyAssignmentPolicy(new ConnectorPolicyAssignmentPolicy());
    return spec;
}