Examples with DatasetSpecification io.cdap.cdap.api.dataset.DatasetSpecification used on opensource projects

Example 26 with DatasetSpecification

use of io.cdap.cdap.api.dataset.DatasetSpecification in project cdap by caskdata.

the class ConversionHelpers method spec2Summary.

static Collection<DatasetSpecificationSummary> spec2Summary(Collection<DatasetSpecification> specs) {
    List<DatasetSpecificationSummary> datasetSummaries = Lists.newArrayList();
    for (DatasetSpecification spec : specs) {
        // TODO: (CDAP-3097) handle system datasets specially within a namespace instead of filtering them out
        // by the handler. This filter is only in the list endpoint because the other endpoints are used by
        // HBaseQueueAdmin through DatasetFramework.
        spec = DatasetsUtil.fixOriginalProperties(spec);
        datasetSummaries.add(new DatasetSpecificationSummary(spec.getName(), spec.getType(), spec.getDescription(), spec.getOriginalProperties()));
    }
    return datasetSummaries;
}

Example 27 with DatasetSpecification

use of io.cdap.cdap.api.dataset.DatasetSpecification in project cdap by caskdata.

the class DatasetTypeManager method addModule.

/**
 * Add datasets module in a namespace
 *
 * @param datasetModuleId the {@link DatasetModuleId} to add
 * @param className module class
 * @param jarLocation location of the module jar
 * @param force if true, an update will be allowed even if there are conflicts with other modules, or if
 *                     removal of a type would break other modules' dependencies.
 */
public void addModule(final DatasetModuleId datasetModuleId, final String className, final Location jarLocation, final boolean force) throws DatasetModuleConflictException {
    LOG.debug("adding module: {}, className: {}, jarLocation: {}", datasetModuleId, className, jarLocation == null ? "[local]" : jarLocation);
    try {
        TransactionRunners.run(transactionRunner, context -> {
            final DatasetTypeTable datasetTypeTable = DatasetTypeTable.create(context);
            final DatasetInstanceTable datasetInstanceTable = new DatasetInstanceTable(context);
            // 1. get existing module with all its types
            DatasetModuleMeta existing = datasetTypeTable.getModule(datasetModuleId);
            DependencyTrackingRegistry reg;
            // 2. unpack jar and create class loader
            ClassLoaderFolder classLoaderFolder = null;
            DirectoryClassLoader cl = null;
            try {
                // NOTE: if jarLocation is null, we assume that this is a system module, ie. always present in classpath
                if (jarLocation != null) {
                    classLoaderFolder = BundleJarUtil.prepareClassLoaderFolder(jarLocation, () -> Files.createTempDirectory(Files.createDirectories(systemTempPath), datasetModuleId.getEntityName()).toFile());
                    cl = new DirectoryClassLoader(classLoaderFolder.getDir(), cConf.get(Constants.AppFabric.PROGRAM_EXTRA_CLASSPATH), FilterClassLoader.create(getClass().getClassLoader()), "lib");
                }
                reg = new DependencyTrackingRegistry(datasetModuleId, datasetTypeTable, cl, force);
                // 3. register the new module while tracking dependencies.
                // this will fail if a type exists in a different module
                DatasetDefinitionRegistries.register(className, cl, reg);
            } catch (TypeConflictException e) {
                // type conflict from the registry, we want to throw that as is
                throw e;
            } catch (Exception e) {
                LOG.error("Could not instantiate instance of dataset module class {} for module {} using jarLocation {}", className, datasetModuleId, jarLocation);
                throw Throwables.propagate(e);
            } finally {
                // Close the ProgramClassLoader
                Closeables.closeQuietly(cl);
                Closeables.closeQuietly(classLoaderFolder);
            }
            // 4. determine whether any type were removed from the module, and whether any other modules depend on them
            if (existing != null) {
                Set<String> removedTypes = new HashSet<>(existing.getTypes());
                removedTypes.removeAll(reg.getTypes());
                // TODO (CDAP-6294): track dependencies at the type level
                if (!force && !removedTypes.isEmpty() && !existing.getUsedByModules().isEmpty()) {
                    throw new DatasetModuleConflictException(String.format("Cannot update module '%s' to remove types %s: Modules %s may depend on it. Delete them first", datasetModuleId, removedTypes, existing.getUsedByModules()));
                }
                Collection<DatasetSpecification> instances = datasetInstanceTable.getByTypes(datasetModuleId.getParent(), removedTypes);
                if (!instances.isEmpty()) {
                    throw new DatasetModuleConflictException(String.format("Attempt to remove dataset types %s from module '%s' that have existing instances: %s. " + "Delete them first.", removedTypes, datasetModuleId, instances.stream().map(input -> input.getName() + ":" + input.getType()).collect(Collectors.joining(", "))));
                }
            }
            // NOTE: we use set to avoid duplicated dependencies
            // NOTE: we use LinkedHashSet to preserve order in which dependencies must be loaded
            Set<String> moduleDependencies = new LinkedHashSet<String>();
            for (DatasetTypeId usedType : reg.getUsedTypes()) {
                DatasetModuleMeta usedModule = datasetTypeTable.getModuleByType(usedType);
                if (usedModule == null) {
                    throw new IllegalStateException(String.format("Found a null used module for type %s for while adding module %s", usedType, datasetModuleId));
                }
                // adding all used types and the module itself, in this very order to keep the order of loading modules
                // for instantiating a type
                moduleDependencies.addAll(usedModule.getUsesModules());
                boolean added = moduleDependencies.add(usedModule.getName());
                if (added) {
                    // also adding this module as a dependent for all modules it uses
                    usedModule.addUsedByModule(datasetModuleId.getEntityName());
                    datasetTypeTable.writeModule(usedType.getParent(), usedModule);
                }
            }
            URI jarURI = jarLocation == null ? null : jarLocation.toURI();
            DatasetModuleMeta moduleMeta = existing == null ? new DatasetModuleMeta(datasetModuleId.getEntityName(), className, jarURI, reg.getTypes(), Lists.newArrayList(moduleDependencies)) : new DatasetModuleMeta(datasetModuleId.getEntityName(), className, jarURI, reg.getTypes(), Lists.newArrayList(moduleDependencies), Lists.newArrayList(existing.getUsedByModules()));
            datasetTypeTable.writeModule(datasetModuleId.getParent(), moduleMeta);
        });
    } catch (RuntimeException e) {
        for (Throwable cause : Throwables.getCausalChain(e)) {
            if (cause instanceof DatasetModuleConflictException) {
                throw (DatasetModuleConflictException) cause;
            } else if (cause instanceof TypeConflictException) {
                throw new DatasetModuleConflictException(cause.getMessage(), cause);
            }
        }
        throw Throwables.propagate(e);
    } catch (Exception e) {
        LOG.error("Operation failed", e);
        throw Throwables.propagate(e);
    }
}

Example 28 with DatasetSpecification

use of io.cdap.cdap.api.dataset.DatasetSpecification in project cdap by caskdata.

the class DatasetTypeManager method deleteModule.

/**
 * Deletes specified dataset module
 * @param datasetModuleId {@link DatasetModuleId} of the dataset module to delete
 * @return true if deleted successfully, false if module didn't exist: nothing to delete
 * @throws DatasetModuleConflictException when there are other modules depend on the specified one, in which case
 *         deletion does NOT happen
 */
public boolean deleteModule(final DatasetModuleId datasetModuleId) throws DatasetModuleConflictException {
    LOG.info("Deleting module {}", datasetModuleId);
    try {
        return TransactionRunners.run(transactionRunner, context -> {
            final DatasetTypeTable datasetTypeTable = DatasetTypeTable.create(context);
            final DatasetInstanceTable datasetInstanceTable = new DatasetInstanceTable(context);
            final DatasetModuleMeta module = datasetTypeTable.getModule(datasetModuleId);
            if (module == null) {
                return false;
            }
            // cannot delete when there's module that uses it
            if (module.getUsedByModules().size() > 0) {
                String msg = String.format("Cannot delete module %s: other modules depend on it. Delete them first", module);
                throw new DatasetModuleConflictException(msg);
            }
            Collection<DatasetSpecification> instances = datasetInstanceTable.getByTypes(datasetModuleId.getParent(), ImmutableSet.copyOf(module.getTypes()));
            // cannot delete when there's instance that uses it
            if (!instances.isEmpty()) {
                String msg = String.format("Cannot delete module %s: other instances depend on it. Delete them first", module);
                throw new DatasetModuleConflictException(msg);
            }
            // remove it from "usedBy" from other modules
            for (String usedModuleName : module.getUsesModules()) {
                DatasetModuleId usedModuleId = new DatasetModuleId(datasetModuleId.getNamespace(), usedModuleName);
                // not using getModuleWithFallback here because we want to know the namespace in which usedModule was found,
                // so we can overwrite it in the MDS in the appropriate namespace
                DatasetModuleMeta usedModule = datasetTypeTable.getModule(usedModuleId);
                // if the usedModule is not found in the current namespace, try finding it in the system namespace
                if (usedModule == null) {
                    usedModuleId = NamespaceId.SYSTEM.datasetModule(usedModuleName);
                    usedModule = datasetTypeTable.getModule(usedModuleId);
                    Preconditions.checkState(usedModule != null, "Could not find a module %s that the module %s uses.", usedModuleName, datasetModuleId.getEntityName());
                }
                usedModule.removeUsedByModule(datasetModuleId.getEntityName());
                datasetTypeTable.writeModule(usedModuleId.getParent(), usedModule);
            }
            datasetTypeTable.deleteModule(datasetModuleId);
            try {
                // Also delete module jar
                Location moduleJarLocation = impersonator.doAs(datasetModuleId, () -> Locations.getLocationFromAbsolutePath(locationFactory, module.getJarLocationPath()));
                if (!moduleJarLocation.delete()) {
                    LOG.debug("Could not delete dataset module archive");
                }
            } catch (Exception e) {
                // the only checked exception the try-catch throws is IOException
                Throwables.propagateIfInstanceOf(e, IOException.class);
                throw Throwables.propagate(e);
            }
            return true;
        });
    } catch (RuntimeException e) {
        for (Throwable cause : Throwables.getCausalChain(e)) {
            if (cause instanceof DatasetModuleConflictException) {
                throw (DatasetModuleConflictException) cause;
            }
        }
        throw Throwables.propagate(e);
    } catch (Exception e) {
        LOG.error("Operation failed", e);
        throw Throwables.propagate(e);
    }
}

Example 29 with DatasetSpecification

use of io.cdap.cdap.api.dataset.DatasetSpecification in project cdap by caskdata.

the class PartitionedFileSetDefinition method reconfigure.

@Override
public DatasetSpecification reconfigure(String instanceName, DatasetProperties properties, DatasetSpecification currentSpec) throws IncompatibleUpdateException {
    // validate that the partitioning is not changing
    Partitioning oldPartitioning = PartitionedFileSetProperties.getPartitioning(currentSpec.getProperties());
    Partitioning newPartitioning = PartitionedFileSetProperties.getPartitioning(properties.getProperties());
    Preconditions.checkNotNull(oldPartitioning, "Existing dataset has no partitioning");
    Preconditions.checkNotNull(newPartitioning, "New properties do not contain partitioning");
    if (!Iterators.elementsEqual(oldPartitioning.getFields().entrySet().iterator(), newPartitioning.getFields().entrySet().iterator())) {
        throw new IncompatibleUpdateException(String.format("Partitioning cannot be changed. Existing: %s, new: %s", oldPartitioning, newPartitioning));
    }
    Map<String, String> pfsProperties = new HashMap<>(properties.getProperties());
    // define the columns for indexing on the partitionsTable
    DatasetProperties indexedTableProperties = DatasetProperties.builder().addAll(properties.getProperties()).add(IndexedTable.INDEX_COLUMNS_CONF_KEY, INDEXED_COLS).build();
    // only set the default base path property if the default was set the last time it was configured,
    // and no base path is in the current properties.
    DatasetSpecification currentFileSpec = currentSpec.getSpecification(FILESET_NAME);
    DatasetProperties.Builder newFileProperties = DatasetProperties.builder().addAll(properties.getProperties());
    String useNameAsBasePathDefault = currentSpec.getProperty(NAME_AS_BASE_PATH_DEFAULT);
    if (Boolean.parseBoolean(useNameAsBasePathDefault) && !properties.getProperties().containsKey(FileSetProperties.BASE_PATH)) {
        newFileProperties.add(FileSetProperties.BASE_PATH, instanceName);
        pfsProperties.put(NAME_AS_BASE_PATH_DEFAULT, Boolean.TRUE.toString());
    }
    return DatasetSpecification.builder(instanceName, getName()).properties(pfsProperties).datasets(AbstractDatasetDefinition.reconfigure(filesetDef, FILESET_NAME, newFileProperties.build(), currentFileSpec), AbstractDatasetDefinition.reconfigure(indexedTableDef, PARTITION_TABLE_NAME, indexedTableProperties, currentSpec.getSpecification(PARTITION_TABLE_NAME))).build();
}

Example 30 with DatasetSpecification

use of io.cdap.cdap.api.dataset.DatasetSpecification in project cdap by caskdata.

the class InMemoryDatasetFramework method deleteAllModules.

@Override
public void deleteAllModules(NamespaceId namespaceId) throws ModuleConflictException {
    writeLock.lock();
    try {
        // check if there are any datasets that use types from the namespace from which we want to remove all modules
        Set<String> typesInNamespace = nonDefaultTypes.get(namespaceId);
        for (DatasetSpecification spec : instances.row(namespaceId).values()) {
            if (typesInNamespace.contains(spec.getType())) {
                throw new ModuleConflictException(String.format("Cannot delete all modules in namespace '%s', some datasets use them", namespaceId));
            }
        }
        moduleClasses.row(namespaceId).clear();
        nonDefaultTypes.removeAll(namespaceId);
        registries.put(namespaceId, registryFactory.create());
    } finally {
        writeLock.unlock();
    }
}