Examples with DatasetDecl org.apache.asterix.lang.common.statement.DatasetDecl used on opensource projects

Example 1 with DatasetDecl

use of org.apache.asterix.lang.common.statement.DatasetDecl in project asterixdb by apache.

the class AbstractLangTranslator method validateOperation.

public void validateOperation(ICcApplicationContext appCtx, Dataverse defaultDataverse, Statement stmt) throws AsterixException {
    if (!(ClusterStateManager.INSTANCE.getState().equals(ClusterState.ACTIVE) && ClusterStateManager.INSTANCE.isGlobalRecoveryCompleted())) {
        int maxWaitCycles = appCtx.getExternalProperties().getMaxWaitClusterActive();
        try {
            ClusterStateManager.INSTANCE.waitForState(ClusterState.ACTIVE, maxWaitCycles, TimeUnit.SECONDS);
        } catch (HyracksDataException e) {
            throw new AsterixException(e);
        } catch (InterruptedException e) {
            if (LOGGER.isLoggable(Level.WARNING)) {
                LOGGER.warning("Thread interrupted while waiting for cluster to be " + ClusterState.ACTIVE);
            }
            Thread.currentThread().interrupt();
        }
        if (!ClusterStateManager.INSTANCE.getState().equals(ClusterState.ACTIVE)) {
            throw new AsterixException("Cluster is in " + ClusterState.UNUSABLE + " state." + "\n One or more Node Controllers have left or haven't joined yet.\n");
        } else {
            if (LOGGER.isLoggable(Level.INFO)) {
                LOGGER.info("Cluster is now " + ClusterState.ACTIVE);
            }
        }
    }
    if (ClusterStateManager.INSTANCE.getState().equals(ClusterState.UNUSABLE)) {
        throw new AsterixException("Cluster is in " + ClusterState.UNUSABLE + " state." + "\n One or more Node Controllers have left.\n");
    }
    if (!ClusterStateManager.INSTANCE.isGlobalRecoveryCompleted()) {
        int maxWaitCycles = appCtx.getExternalProperties().getMaxWaitClusterActive();
        int waitCycleCount = 0;
        try {
            while (!ClusterStateManager.INSTANCE.isGlobalRecoveryCompleted() && waitCycleCount < maxWaitCycles) {
                Thread.sleep(1000);
                waitCycleCount++;
            }
        } catch (InterruptedException e) {
            if (LOGGER.isLoggable(Level.WARNING)) {
                LOGGER.warning("Thread interrupted while waiting for cluster to complete global recovery ");
            }
            Thread.currentThread().interrupt();
        }
        if (!ClusterStateManager.INSTANCE.isGlobalRecoveryCompleted()) {
            throw new AsterixException("Cluster Global recovery is not yet complete and the system is in " + ClusterState.ACTIVE + " state");
        }
    }
    boolean invalidOperation = false;
    String message = null;
    String dataverse = defaultDataverse != null ? defaultDataverse.getDataverseName() : null;
    switch(stmt.getKind()) {
        case Statement.Kind.INSERT:
            InsertStatement insertStmt = (InsertStatement) stmt;
            if (insertStmt.getDataverseName() != null) {
                dataverse = insertStmt.getDataverseName().getValue();
            }
            invalidOperation = MetadataConstants.METADATA_DATAVERSE_NAME.equals(dataverse);
            if (invalidOperation) {
                message = "Insert operation is not permitted in dataverse " + MetadataConstants.METADATA_DATAVERSE_NAME;
            }
            break;
        case Statement.Kind.DELETE:
            DeleteStatement deleteStmt = (DeleteStatement) stmt;
            if (deleteStmt.getDataverseName() != null) {
                dataverse = deleteStmt.getDataverseName().getValue();
            }
            invalidOperation = MetadataConstants.METADATA_DATAVERSE_NAME.equals(dataverse);
            if (invalidOperation) {
                message = "Delete operation is not permitted in dataverse " + MetadataConstants.METADATA_DATAVERSE_NAME;
            }
            break;
        case Statement.Kind.DATAVERSE_DROP:
            DataverseDropStatement dvDropStmt = (DataverseDropStatement) stmt;
            invalidOperation = MetadataConstants.METADATA_DATAVERSE_NAME.equals(dvDropStmt.getDataverseName().getValue());
            if (invalidOperation) {
                message = "Cannot drop dataverse:" + dvDropStmt.getDataverseName().getValue();
            }
            break;
        case Statement.Kind.DATASET_DROP:
            DropDatasetStatement dropStmt = (DropDatasetStatement) stmt;
            if (dropStmt.getDataverseName() != null) {
                dataverse = dropStmt.getDataverseName().getValue();
            }
            invalidOperation = MetadataConstants.METADATA_DATAVERSE_NAME.equals(dataverse);
            if (invalidOperation) {
                message = "Cannot drop a dataset belonging to the dataverse:" + MetadataConstants.METADATA_DATAVERSE_NAME;
            }
            break;
        case Statement.Kind.DATASET_DECL:
            DatasetDecl datasetStmt = (DatasetDecl) stmt;
            Map<String, String> hints = datasetStmt.getHints();
            if (hints != null && !hints.isEmpty()) {
                Pair<Boolean, String> validationResult = null;
                StringBuffer errorMsgBuffer = new StringBuffer();
                for (Entry<String, String> hint : hints.entrySet()) {
                    validationResult = DatasetHints.validate(appCtx, hint.getKey(), hint.getValue());
                    if (!validationResult.first) {
                        errorMsgBuffer.append("Dataset: " + datasetStmt.getName().getValue() + " error in processing hint: " + hint.getKey() + " " + validationResult.second);
                        errorMsgBuffer.append(" \n");
                    }
                }
                invalidOperation = errorMsgBuffer.length() > 0;
                if (invalidOperation) {
                    message = errorMsgBuffer.toString();
                }
            }
            break;
        default:
            break;
    }
    if (invalidOperation) {
        throw new AsterixException("Invalid operation - " + message);
    }
}

Example 2 with DatasetDecl

use of org.apache.asterix.lang.common.statement.DatasetDecl in project asterixdb by apache.

the class QueryTranslator method handleCreateDatasetStatement.

public void handleCreateDatasetStatement(MetadataProvider metadataProvider, Statement stmt, IHyracksClientConnection hcc) throws CompilationException, Exception {
    MutableObject<ProgressState> progress = new MutableObject<>(ProgressState.NO_PROGRESS);
    DatasetDecl dd = (DatasetDecl) stmt;
    String dataverseName = getActiveDataverse(dd.getDataverse());
    String datasetName = dd.getName().getValue();
    DatasetType dsType = dd.getDatasetType();
    String itemTypeDataverseName = getActiveDataverse(dd.getItemTypeDataverse());
    String itemTypeName = dd.getItemTypeName().getValue();
    String metaItemTypeDataverseName = getActiveDataverse(dd.getMetaItemTypeDataverse());
    String metaItemTypeName = dd.getMetaItemTypeName().getValue();
    Identifier ngNameId = dd.getNodegroupName();
    String nodegroupName = ngNameId == null ? null : ngNameId.getValue();
    String compactionPolicy = dd.getCompactionPolicy();
    Map<String, String> compactionPolicyProperties = dd.getCompactionPolicyProperties();
    boolean defaultCompactionPolicy = compactionPolicy == null;
    boolean temp = dd.getDatasetDetailsDecl().isTemp();
    MetadataTransactionContext mdTxnCtx = MetadataManager.INSTANCE.beginTransaction();
    boolean bActiveTxn = true;
    metadataProvider.setMetadataTxnContext(mdTxnCtx);
    MetadataLockManager.INSTANCE.createDatasetBegin(metadataProvider.getLocks(), dataverseName, itemTypeDataverseName, itemTypeDataverseName + "." + itemTypeName, metaItemTypeDataverseName, metaItemTypeDataverseName + "." + metaItemTypeName, nodegroupName, compactionPolicy, dataverseName + "." + datasetName, defaultCompactionPolicy);
    Dataset dataset = null;
    try {
        IDatasetDetails datasetDetails = null;
        Dataset ds = metadataProvider.findDataset(dataverseName, datasetName);
        if (ds != null) {
            if (dd.getIfNotExists()) {
                MetadataManager.INSTANCE.commitTransaction(mdTxnCtx);
                return;
            } else {
                throw new AlgebricksException("A dataset with this name " + datasetName + " already exists.");
            }
        }
        Datatype dt = MetadataManager.INSTANCE.getDatatype(metadataProvider.getMetadataTxnContext(), itemTypeDataverseName, itemTypeName);
        if (dt == null) {
            throw new AlgebricksException(": type " + itemTypeName + " could not be found.");
        }
        String ngName = ngNameId != null ? ngNameId.getValue() : configureNodegroupForDataset(appCtx, dd.getHints(), dataverseName, datasetName, metadataProvider);
        if (compactionPolicy == null) {
            compactionPolicy = GlobalConfig.DEFAULT_COMPACTION_POLICY_NAME;
            compactionPolicyProperties = GlobalConfig.DEFAULT_COMPACTION_POLICY_PROPERTIES;
        } else {
            validateCompactionPolicy(compactionPolicy, compactionPolicyProperties, mdTxnCtx, false);
        }
        switch(dd.getDatasetType()) {
            case INTERNAL:
                IAType itemType = dt.getDatatype();
                if (itemType.getTypeTag() != ATypeTag.OBJECT) {
                    throw new AlgebricksException("Dataset type has to be a record type.");
                }
                IAType metaItemType = null;
                if (metaItemTypeDataverseName != null && metaItemTypeName != null) {
                    metaItemType = metadataProvider.findType(metaItemTypeDataverseName, metaItemTypeName);
                }
                if (metaItemType != null && metaItemType.getTypeTag() != ATypeTag.OBJECT) {
                    throw new AlgebricksException("Dataset meta type has to be a record type.");
                }
                ARecordType metaRecType = (ARecordType) metaItemType;
                List<List<String>> partitioningExprs = ((InternalDetailsDecl) dd.getDatasetDetailsDecl()).getPartitioningExprs();
                List<Integer> keySourceIndicators = ((InternalDetailsDecl) dd.getDatasetDetailsDecl()).getKeySourceIndicators();
                boolean autogenerated = ((InternalDetailsDecl) dd.getDatasetDetailsDecl()).isAutogenerated();
                ARecordType aRecordType = (ARecordType) itemType;
                List<IAType> partitioningTypes = ValidateUtil.validatePartitioningExpressions(aRecordType, metaRecType, partitioningExprs, keySourceIndicators, autogenerated);
                List<String> filterField = ((InternalDetailsDecl) dd.getDatasetDetailsDecl()).getFilterField();
                if (filterField != null) {
                    ValidateUtil.validateFilterField(aRecordType, filterField);
                }
                if (compactionPolicy == null && filterField != null) {
                    // If the dataset has a filter and the user didn't specify a merge
                    // policy, then we will pick the
                    // correlated-prefix as the default merge policy.
                    compactionPolicy = GlobalConfig.DEFAULT_FILTERED_DATASET_COMPACTION_POLICY_NAME;
                    compactionPolicyProperties = GlobalConfig.DEFAULT_COMPACTION_POLICY_PROPERTIES;
                }
                datasetDetails = new InternalDatasetDetails(InternalDatasetDetails.FileStructure.BTREE, InternalDatasetDetails.PartitioningStrategy.HASH, partitioningExprs, partitioningExprs, keySourceIndicators, partitioningTypes, autogenerated, filterField, temp);
                break;
            case EXTERNAL:
                String adapter = ((ExternalDetailsDecl) dd.getDatasetDetailsDecl()).getAdapter();
                Map<String, String> properties = ((ExternalDetailsDecl) dd.getDatasetDetailsDecl()).getProperties();
                datasetDetails = new ExternalDatasetDetails(adapter, properties, new Date(), TransactionState.COMMIT);
                break;
            default:
                throw new CompilationException("Unknown datatype " + dd.getDatasetType());
        }
        // #. initialize DatasetIdFactory if it is not initialized.
        if (!DatasetIdFactory.isInitialized()) {
            DatasetIdFactory.initialize(MetadataManager.INSTANCE.getMostRecentDatasetId());
        }
        // #. add a new dataset with PendingAddOp
        dataset = new Dataset(dataverseName, datasetName, itemTypeDataverseName, itemTypeName, metaItemTypeDataverseName, metaItemTypeName, ngName, compactionPolicy, compactionPolicyProperties, datasetDetails, dd.getHints(), dsType, DatasetIdFactory.generateDatasetId(), MetadataUtil.PENDING_ADD_OP);
        MetadataManager.INSTANCE.addDataset(metadataProvider.getMetadataTxnContext(), dataset);
        if (dd.getDatasetType() == DatasetType.INTERNAL) {
            JobSpecification jobSpec = DatasetUtil.createDatasetJobSpec(dataset, metadataProvider);
            // #. make metadataTxn commit before calling runJob.
            MetadataManager.INSTANCE.commitTransaction(mdTxnCtx);
            bActiveTxn = false;
            progress.setValue(ProgressState.ADDED_PENDINGOP_RECORD_TO_METADATA);
            // #. runJob
            JobUtils.runJob(hcc, jobSpec, true);
            // #. begin new metadataTxn
            mdTxnCtx = MetadataManager.INSTANCE.beginTransaction();
            bActiveTxn = true;
            metadataProvider.setMetadataTxnContext(mdTxnCtx);
        }
        // #. add a new dataset with PendingNoOp after deleting the dataset with PendingAddOp
        MetadataManager.INSTANCE.dropDataset(metadataProvider.getMetadataTxnContext(), dataverseName, datasetName);
        dataset.setPendingOp(MetadataUtil.PENDING_NO_OP);
        MetadataManager.INSTANCE.addDataset(metadataProvider.getMetadataTxnContext(), dataset);
        MetadataManager.INSTANCE.commitTransaction(mdTxnCtx);
    } catch (Exception e) {
        if (bActiveTxn) {
            abort(e, e, mdTxnCtx);
        }
        if (progress.getValue() == ProgressState.ADDED_PENDINGOP_RECORD_TO_METADATA) {
            // #. execute compensation operations
            // remove the index in NC
            // [Notice]
            // As long as we updated(and committed) metadata, we should remove any effect of the job
            // because an exception occurs during runJob.
            mdTxnCtx = MetadataManager.INSTANCE.beginTransaction();
            bActiveTxn = true;
            metadataProvider.setMetadataTxnContext(mdTxnCtx);
            try {
                JobSpecification jobSpec = DatasetUtil.dropDatasetJobSpec(dataset, metadataProvider);
                MetadataManager.INSTANCE.commitTransaction(mdTxnCtx);
                bActiveTxn = false;
                JobUtils.runJob(hcc, jobSpec, true);
            } catch (Exception e2) {
                e.addSuppressed(e2);
                if (bActiveTxn) {
                    abort(e, e2, mdTxnCtx);
                }
            }
            // remove the record from the metadata.
            mdTxnCtx = MetadataManager.INSTANCE.beginTransaction();
            metadataProvider.setMetadataTxnContext(mdTxnCtx);
            try {
                MetadataManager.INSTANCE.dropDataset(metadataProvider.getMetadataTxnContext(), dataverseName, datasetName);
                MetadataManager.INSTANCE.commitTransaction(mdTxnCtx);
            } catch (Exception e2) {
                e.addSuppressed(e2);
                abort(e, e2, mdTxnCtx);
                throw new IllegalStateException("System is inconsistent state: pending dataset(" + dataverseName + "." + datasetName + ") couldn't be removed from the metadata", e);
            }
        }
        throw e;
    } finally {
        metadataProvider.getLocks().unlock();
    }
}

Example 3 with DatasetDecl

use of org.apache.asterix.lang.common.statement.DatasetDecl in project asterixdb by apache.

the class QueryTranslator method prepareRunExternalRuntime.

// Prepares to run a program on external runtime.
protected void prepareRunExternalRuntime(MetadataProvider metadataProvider, IHyracksClientConnection hcc, RunStatement pregelixStmt, String dataverseNameFrom, String dataverseNameTo, String datasetNameFrom, String datasetNameTo, MetadataTransactionContext mdTxnCtx) throws Exception {
    // Validates the source/sink dataverses and datasets.
    Dataset fromDataset = metadataProvider.findDataset(dataverseNameFrom, datasetNameFrom);
    if (fromDataset == null) {
        throw new CompilationException("The source dataset " + datasetNameFrom + " in dataverse " + dataverseNameFrom + " could not be found for the Run command");
    }
    Dataset toDataset = metadataProvider.findDataset(dataverseNameTo, datasetNameTo);
    if (toDataset == null) {
        throw new CompilationException("The sink dataset " + datasetNameTo + " in dataverse " + dataverseNameTo + " could not be found for the Run command");
    }
    try {
        // Find the primary index of the sink dataset.
        Index toIndex = null;
        List<Index> indexes = MetadataManager.INSTANCE.getDatasetIndexes(mdTxnCtx, dataverseNameTo, pregelixStmt.getDatasetNameTo().getValue());
        for (Index index : indexes) {
            if (index.isPrimaryIndex()) {
                toIndex = index;
                break;
            }
        }
        if (toIndex == null) {
            throw new AlgebricksException("Tried to access non-existing dataset: " + datasetNameTo);
        }
        // Cleans up the sink dataset -- Drop and then Create.
        DropDatasetStatement dropStmt = new DropDatasetStatement(new Identifier(dataverseNameTo), pregelixStmt.getDatasetNameTo(), true);
        this.handleDatasetDropStatement(metadataProvider, dropStmt, hcc);
        IDatasetDetailsDecl idd = new InternalDetailsDecl(toIndex.getKeyFieldNames(), toIndex.getKeyFieldSourceIndicators(), false, null, toDataset.getDatasetDetails().isTemp());
        DatasetDecl createToDataset = new DatasetDecl(new Identifier(dataverseNameTo), pregelixStmt.getDatasetNameTo(), new Identifier(toDataset.getItemTypeDataverseName()), new Identifier(toDataset.getItemTypeName()), new Identifier(toDataset.getMetaItemTypeDataverseName()), new Identifier(toDataset.getMetaItemTypeName()), new Identifier(toDataset.getNodeGroupName()), toDataset.getCompactionPolicy(), toDataset.getCompactionPolicyProperties(), toDataset.getHints(), toDataset.getDatasetType(), idd, false);
        this.handleCreateDatasetStatement(metadataProvider, createToDataset, hcc);
    } catch (Exception e) {
        LOGGER.log(Level.WARNING, e.getMessage(), e);
        throw new AlgebricksException("Error cleaning the result dataset. This should not happen.");
    }
    // Flushes source dataset.
    FlushDatasetUtil.flushDataset(hcc, metadataProvider, dataverseNameFrom, datasetNameFrom, datasetNameFrom);
}