Search in sources :

Example 1 with TypeConflictException

use of io.cdap.cdap.data2.dataset2.TypeConflictException 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);
    }
}
Also used : TransactionRunners(io.cdap.cdap.spi.data.transaction.TransactionRunners) NamespaceId(io.cdap.cdap.proto.id.NamespaceId) Location(org.apache.twill.filesystem.Location) DatasetTypeMeta(io.cdap.cdap.proto.DatasetTypeMeta) Inject(com.google.inject.Inject) LoggerFactory(org.slf4j.LoggerFactory) Callable(java.util.concurrent.Callable) DatasetSpecification(io.cdap.cdap.api.dataset.DatasetSpecification) InMemoryDatasetDefinitionRegistry(io.cdap.cdap.data2.dataset2.InMemoryDatasetDefinitionRegistry) ArrayList(java.util.ArrayList) HashSet(java.util.HashSet) Lists(com.google.common.collect.Lists) Locations(io.cdap.cdap.common.io.Locations) Closeables(com.google.common.io.Closeables) DatasetTypeId(io.cdap.cdap.proto.id.DatasetTypeId) URI(java.net.URI) Path(java.nio.file.Path) LinkedHashSet(java.util.LinkedHashSet) Nullable(javax.annotation.Nullable) DatasetModuleId(io.cdap.cdap.proto.id.DatasetModuleId) ImmutableSet(com.google.common.collect.ImmutableSet) Logger(org.slf4j.Logger) Files(java.nio.file.Files) DatasetModuleMeta(io.cdap.cdap.proto.DatasetModuleMeta) Collection(java.util.Collection) Throwables(com.google.common.base.Throwables) DatasetDefinitionRegistry(io.cdap.cdap.api.dataset.module.DatasetDefinitionRegistry) Impersonator(io.cdap.cdap.security.impersonation.Impersonator) Set(java.util.Set) IOException(java.io.IOException) LocationFactory(org.apache.twill.filesystem.LocationFactory) TypeConflictException(io.cdap.cdap.data2.dataset2.TypeConflictException) Collectors(java.util.stream.Collectors) FilterClassLoader(io.cdap.cdap.common.lang.FilterClassLoader) List(java.util.List) CConfiguration(io.cdap.cdap.common.conf.CConfiguration) Paths(java.nio.file.Paths) DatasetInstanceTable(io.cdap.cdap.data2.datafabric.dataset.service.mds.DatasetInstanceTable) DirectoryClassLoader(io.cdap.cdap.common.lang.DirectoryClassLoader) BundleJarUtil(io.cdap.cdap.common.lang.jar.BundleJarUtil) DatasetDefinitionRegistries(io.cdap.cdap.data2.dataset2.DatasetDefinitionRegistries) TransactionRunner(io.cdap.cdap.spi.data.transaction.TransactionRunner) Preconditions(com.google.common.base.Preconditions) Constants(io.cdap.cdap.common.conf.Constants) ClassLoaderFolder(io.cdap.cdap.common.lang.jar.ClassLoaderFolder) VisibleForTesting(com.google.common.annotations.VisibleForTesting) DatasetDefinition(io.cdap.cdap.api.dataset.DatasetDefinition) DatasetTypeTable(io.cdap.cdap.data2.datafabric.dataset.service.mds.DatasetTypeTable) LinkedHashSet(java.util.LinkedHashSet) DirectoryClassLoader(io.cdap.cdap.common.lang.DirectoryClassLoader) TypeConflictException(io.cdap.cdap.data2.dataset2.TypeConflictException) DatasetTypeId(io.cdap.cdap.proto.id.DatasetTypeId) DatasetTypeTable(io.cdap.cdap.data2.datafabric.dataset.service.mds.DatasetTypeTable) DatasetSpecification(io.cdap.cdap.api.dataset.DatasetSpecification) ClassLoaderFolder(io.cdap.cdap.common.lang.jar.ClassLoaderFolder) URI(java.net.URI) IOException(java.io.IOException) TypeConflictException(io.cdap.cdap.data2.dataset2.TypeConflictException) DatasetInstanceTable(io.cdap.cdap.data2.datafabric.dataset.service.mds.DatasetInstanceTable) DatasetModuleMeta(io.cdap.cdap.proto.DatasetModuleMeta) HashSet(java.util.HashSet) LinkedHashSet(java.util.LinkedHashSet)

Example 2 with TypeConflictException

use of io.cdap.cdap.data2.dataset2.TypeConflictException 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 {
        final DatasetTypeMDS datasetTypeMDS = datasetCache.getDataset(DatasetMetaTableUtil.META_TABLE_NAME);
        final DatasetInstanceMDS datasetInstanceMDS = datasetCache.getDataset(DatasetMetaTableUtil.INSTANCE_TABLE_NAME);
        txExecutorFactory.createExecutor(datasetCache).execute(new TransactionExecutor.Subroutine() {

            @Override
            public void apply() throws Exception {
                // 1. get existing module with all its types
                DatasetModuleMeta existing = datasetTypeMDS.getModule(datasetModuleId);
                DependencyTrackingRegistry reg;
                // 2. unpack jar and create class loader
                File unpackedLocation = Files.createTempDirectory(Files.createDirectories(systemTempPath), datasetModuleId.getEntityName()).toFile();
                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) {
                        BundleJarUtil.unJar(jarLocation, unpackedLocation);
                        cl = new DirectoryClassLoader(unpackedLocation, cConf.get(Constants.AppFabric.PROGRAM_EXTRA_CLASSPATH), FilterClassLoader.create(getClass().getClassLoader()), "lib");
                    }
                    reg = new DependencyTrackingRegistry(datasetModuleId, datasetTypeMDS, 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 {
                    if (cl != null) {
                        // Close the ProgramClassLoader
                        Closeables.closeQuietly(cl);
                    }
                    try {
                        DirUtils.deleteDirectoryContents(unpackedLocation);
                    } catch (IOException e) {
                        LOG.warn("Failed to delete directory {}", unpackedLocation, e);
                    }
                }
                // 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 = datasetInstanceMDS.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, Iterables.toString(Iterables.transform(instances, new Function<DatasetSpecification, String>() {

                            @Nullable
                            @Override
                            public String apply(@Nullable DatasetSpecification input) {
                                return input.getName() + ":" + input.getType();
                            }
                        }))));
                    }
                }
                // 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 = datasetTypeMDS.getModuleByType(usedType);
                    Preconditions.checkState(usedModule != null, 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());
                        datasetTypeMDS.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()));
                datasetTypeMDS.writeModule(datasetModuleId.getParent(), moduleMeta);
            }
        });
    } catch (TransactionFailureException e) {
        Throwable cause = e.getCause();
        if (cause != null) {
            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);
    }
}
Also used : ImmutableSet(com.google.common.collect.ImmutableSet) Set(java.util.Set) HashSet(java.util.HashSet) LinkedHashSet(java.util.LinkedHashSet) DatasetTypeId(co.cask.cdap.proto.id.DatasetTypeId) URI(java.net.URI) DatasetModuleMeta(co.cask.cdap.proto.DatasetModuleMeta) DirectoryClassLoader(co.cask.cdap.common.lang.DirectoryClassLoader) TypeConflictException(co.cask.cdap.data2.dataset2.TypeConflictException) DatasetInstanceMDS(co.cask.cdap.data2.datafabric.dataset.service.mds.DatasetInstanceMDS) DatasetSpecification(co.cask.cdap.api.dataset.DatasetSpecification) TransactionExecutor(org.apache.tephra.TransactionExecutor) IOException(java.io.IOException) TransactionFailureException(org.apache.tephra.TransactionFailureException) TypeConflictException(co.cask.cdap.data2.dataset2.TypeConflictException) IOException(java.io.IOException) DatasetTypeMDS(co.cask.cdap.data2.datafabric.dataset.service.mds.DatasetTypeMDS) TransactionFailureException(org.apache.tephra.TransactionFailureException) Collection(java.util.Collection) File(java.io.File) Nullable(javax.annotation.Nullable)

Aggregations

ImmutableSet (com.google.common.collect.ImmutableSet)2 IOException (java.io.IOException)2 URI (java.net.URI)2 Collection (java.util.Collection)2 HashSet (java.util.HashSet)2 LinkedHashSet (java.util.LinkedHashSet)2 DatasetSpecification (co.cask.cdap.api.dataset.DatasetSpecification)1 DirectoryClassLoader (co.cask.cdap.common.lang.DirectoryClassLoader)1 DatasetInstanceMDS (co.cask.cdap.data2.datafabric.dataset.service.mds.DatasetInstanceMDS)1 DatasetTypeMDS (co.cask.cdap.data2.datafabric.dataset.service.mds.DatasetTypeMDS)1 TypeConflictException (co.cask.cdap.data2.dataset2.TypeConflictException)1 DatasetModuleMeta (co.cask.cdap.proto.DatasetModuleMeta)1 DatasetTypeId (co.cask.cdap.proto.id.DatasetTypeId)1 VisibleForTesting (com.google.common.annotations.VisibleForTesting)1 Preconditions (com.google.common.base.Preconditions)1 Throwables (com.google.common.base.Throwables)1 Lists (com.google.common.collect.Lists)1 Closeables (com.google.common.io.Closeables)1 Inject (com.google.inject.Inject)1 DatasetDefinition (io.cdap.cdap.api.dataset.DatasetDefinition)1