Search in sources :

Example 1 with InputIterable

use of org.neo4j.unsafe.impl.batchimport.InputIterable in project neo4j by neo4j.

the class EncodingIdMapperTest method shouldHandleGreatAmountsOfStuff.

@Test
public void shouldHandleGreatAmountsOfStuff() throws Exception {
    // GIVEN
    IdMapper idMapper = mapper(new StringEncoder(), Radix.STRING, NO_MONITOR);
    InputIterable<Object> ids = new InputIterable<Object>() {

        @Override
        public InputIterator<Object> iterator() {
            return new InputIterator.Adapter<Object>() {

                private int i;

                @Override
                protected Object fetchNextOrNull() {
                    return i < 300_000 ? "" + (i++) : null;
                }
            };
        }

        @Override
        public boolean supportsMultiplePasses() {
            return false;
        }
    };
    // WHEN
    long index = 0;
    for (Object id : ids) {
        idMapper.put(id, index++, GLOBAL);
    }
    idMapper.prepare(ids, mock(Collector.class), NONE);
    // THEN
    for (Object id : ids) {
        // the UUIDs here will be generated in the same sequence as above because we reset the random
        if (idMapper.get(id, GLOBAL) == ID_NOT_FOUND) {
            fail("Couldn't find " + id + " even though I added it just previously");
        }
    }
}
Also used : InputIterable(org.neo4j.unsafe.impl.batchimport.InputIterable) Collector(org.neo4j.unsafe.impl.batchimport.input.Collector) Collectors.badCollector(org.neo4j.unsafe.impl.batchimport.input.Collectors.badCollector) IdMapper(org.neo4j.unsafe.impl.batchimport.cache.idmapping.IdMapper) Test(org.junit.Test)

Example 2 with InputIterable

use of org.neo4j.unsafe.impl.batchimport.InputIterable in project neo4j by neo4j.

the class StoreMigrator method migrateWithBatchImporter.

private void migrateWithBatchImporter(File storeDir, File migrationDir, long lastTxId, long lastTxChecksum, long lastTxLogVersion, long lastTxLogByteOffset, MigrationProgressMonitor.Section progressMonitor, RecordFormats oldFormat, RecordFormats newFormat) throws IOException {
    prepareBatchImportMigration(storeDir, migrationDir, oldFormat, newFormat);
    boolean requiresDynamicStoreMigration = !newFormat.dynamic().equals(oldFormat.dynamic());
    boolean requiresPropertyMigration = !newFormat.property().equals(oldFormat.property()) || requiresDynamicStoreMigration;
    File badFile = new File(storeDir, Configuration.BAD_FILE_NAME);
    try (NeoStores legacyStore = instantiateLegacyStore(oldFormat, storeDir);
        RecordCursors nodeInputCursors = new RecordCursors(legacyStore);
        RecordCursors relationshipInputCursors = new RecordCursors(legacyStore);
        OutputStream badOutput = new BufferedOutputStream(new FileOutputStream(badFile, false))) {
        Configuration importConfig = new Configuration.Overridden(config);
        AdditionalInitialIds additionalInitialIds = readAdditionalIds(lastTxId, lastTxChecksum, lastTxLogVersion, lastTxLogByteOffset);
        // We have to make sure to keep the token ids if we're migrating properties/labels
        BatchImporter importer = new ParallelBatchImporter(migrationDir.getAbsoluteFile(), fileSystem, pageCache, importConfig, logService, withDynamicProcessorAssignment(migrationBatchImporterMonitor(legacyStore, progressMonitor, importConfig), importConfig), additionalInitialIds, config, newFormat);
        InputIterable<InputNode> nodes = legacyNodesAsInput(legacyStore, requiresPropertyMigration, nodeInputCursors);
        InputIterable<InputRelationship> relationships = legacyRelationshipsAsInput(legacyStore, requiresPropertyMigration, relationshipInputCursors);
        importer.doImport(Inputs.input(nodes, relationships, IdMappers.actual(), IdGenerators.fromInput(), Collectors.badCollector(badOutput, 0)));
        // During migration the batch importer doesn't necessarily writes all entities, depending on
        // which stores needs migration. Node, relationship, relationship group stores are always written
        // anyways and cannot be avoided with the importer, but delete the store files that weren't written
        // (left empty) so that we don't overwrite those in the real store directory later.
        Collection<StoreFile> storesToDeleteFromMigratedDirectory = new ArrayList<>();
        storesToDeleteFromMigratedDirectory.add(StoreFile.NEO_STORE);
        if (!requiresPropertyMigration) {
            // We didn't migrate properties, so the property stores in the migrated store are just empty/bogus
            storesToDeleteFromMigratedDirectory.addAll(asList(StoreFile.PROPERTY_STORE, StoreFile.PROPERTY_STRING_STORE, StoreFile.PROPERTY_ARRAY_STORE));
        }
        if (!requiresDynamicStoreMigration) {
            // We didn't migrate labels (dynamic node labels) or any other dynamic store
            storesToDeleteFromMigratedDirectory.addAll(asList(StoreFile.NODE_LABEL_STORE, StoreFile.LABEL_TOKEN_STORE, StoreFile.LABEL_TOKEN_NAMES_STORE, StoreFile.RELATIONSHIP_TYPE_TOKEN_STORE, StoreFile.RELATIONSHIP_TYPE_TOKEN_NAMES_STORE, StoreFile.PROPERTY_KEY_TOKEN_STORE, StoreFile.PROPERTY_KEY_TOKEN_NAMES_STORE, StoreFile.SCHEMA_STORE));
        }
        StoreFile.fileOperation(DELETE, fileSystem, migrationDir, null, storesToDeleteFromMigratedDirectory, true, null, StoreFileType.values());
        // When migrating on a block device there might be some files only accessible via the page cache.
        try {
            Predicate<FileHandle> fileHandlePredicate = fileHandle -> storesToDeleteFromMigratedDirectory.stream().anyMatch(storeFile -> storeFile.fileName(StoreFileType.STORE).equals(fileHandle.getFile().getName()));
            pageCache.streamFilesRecursive(migrationDir).filter(fileHandlePredicate).forEach(FileHandle.HANDLE_DELETE);
        } catch (NoSuchFileException e) {
        // This means that we had no files only present in the page cache, this is fine.
        }
    }
}
Also used : PropertyDeduplicator(org.neo4j.kernel.impl.storemigration.legacystore.v21.propertydeduplication.PropertyDeduplicator) Arrays(java.util.Arrays) PageCursor(org.neo4j.io.pagecache.PageCursor) NullLogProvider(org.neo4j.logging.NullLogProvider) StringUtils(org.apache.commons.lang3.StringUtils) UNKNOWN_TX_COMMIT_TIMESTAMP(org.neo4j.kernel.impl.transaction.log.TransactionIdStore.UNKNOWN_TX_COMMIT_TIMESTAMP) COPY(org.neo4j.kernel.impl.storemigration.FileOperation.COPY) SchemaIndexProvider(org.neo4j.kernel.api.index.SchemaIndexProvider) CountsComputer(org.neo4j.kernel.impl.store.CountsComputer) Arrays.asList(java.util.Arrays.asList) BASE_TX_COMMIT_TIMESTAMP(org.neo4j.kernel.impl.transaction.log.TransactionIdStore.BASE_TX_COMMIT_TIMESTAMP) Position(org.neo4j.kernel.impl.store.MetaDataStore.Position) Lifespan(org.neo4j.kernel.lifecycle.Lifespan) RecordCursors(org.neo4j.kernel.impl.store.RecordCursors) SilentMigrationProgressMonitor(org.neo4j.kernel.impl.storemigration.monitoring.SilentMigrationProgressMonitor) StandardCharsets(java.nio.charset.StandardCharsets) MetaDataRecordFormat(org.neo4j.kernel.impl.store.format.standard.MetaDataRecordFormat) DELETE(org.neo4j.kernel.impl.storemigration.FileOperation.DELETE) Stream(java.util.stream.Stream) ExecutionSupervisors.withDynamicProcessorAssignment(org.neo4j.unsafe.impl.batchimport.staging.ExecutionSupervisors.withDynamicProcessorAssignment) StoreType(org.neo4j.kernel.impl.store.StoreType) BatchImporter(org.neo4j.unsafe.impl.batchimport.BatchImporter) NodeStore(org.neo4j.kernel.impl.store.NodeStore) StoreFactory(org.neo4j.kernel.impl.store.StoreFactory) Collectors(org.neo4j.unsafe.impl.batchimport.input.Collectors) PagedFile(org.neo4j.io.pagecache.PagedFile) VERSION_TRAILERS(org.neo4j.kernel.impl.store.format.Capability.VERSION_TRAILERS) FormatFamily(org.neo4j.kernel.impl.store.format.FormatFamily) Supplier(java.util.function.Supplier) UNKNOWN_TX_CHECKSUM(org.neo4j.kernel.impl.transaction.log.TransactionIdStore.UNKNOWN_TX_CHECKSUM) BufferedOutputStream(java.io.BufferedOutputStream) StandardCopyOption(java.nio.file.StandardCopyOption) ArrayList(java.util.ArrayList) TransactionId(org.neo4j.kernel.impl.store.TransactionId) StandardV2_2(org.neo4j.kernel.impl.store.format.standard.StandardV2_2) LogPosition(org.neo4j.kernel.impl.transaction.log.LogPosition) StandardV2_1(org.neo4j.kernel.impl.store.format.standard.StandardV2_1) RelationshipStore(org.neo4j.kernel.impl.store.RelationshipStore) StandardV2_0(org.neo4j.kernel.impl.store.format.standard.StandardV2_0) IdMappers(org.neo4j.unsafe.impl.batchimport.cache.idmapping.IdMappers) BiConsumer(java.util.function.BiConsumer) IdGenerators(org.neo4j.unsafe.impl.batchimport.cache.idmapping.IdGenerators) StreamSupport(java.util.stream.StreamSupport) DirectRecordStoreMigrator(org.neo4j.kernel.impl.storemigration.DirectRecordStoreMigrator) CoarseBoundedProgressExecutionMonitor(org.neo4j.unsafe.impl.batchimport.staging.CoarseBoundedProgressExecutionMonitor) StoreFile(org.neo4j.kernel.impl.storemigration.StoreFile) FileOutputStream(java.io.FileOutputStream) RecordFormatSelector.selectForVersion(org.neo4j.kernel.impl.store.format.RecordFormatSelector.selectForVersion) IOException(java.io.IOException) NodeRecordFormat(org.neo4j.kernel.impl.store.format.standard.NodeRecordFormat) LegacyLogs(org.neo4j.kernel.impl.storemigration.legacylogs.LegacyLogs) File(java.io.File) Iterables(org.neo4j.helpers.collection.Iterables) ParallelBatchImporter(org.neo4j.unsafe.impl.batchimport.ParallelBatchImporter) NodeRecord(org.neo4j.kernel.impl.store.record.NodeRecord) Configuration(org.neo4j.unsafe.impl.batchimport.Configuration) BufferedReader(java.io.BufferedReader) MigrationProgressMonitor(org.neo4j.kernel.impl.storemigration.monitoring.MigrationProgressMonitor) RelationshipRecord(org.neo4j.kernel.impl.store.record.RelationshipRecord) StoreUpgrader(org.neo4j.kernel.impl.storemigration.StoreUpgrader) NoSuchFileException(java.nio.file.NoSuchFileException) BASE_TX_LOG_VERSION(org.neo4j.kernel.impl.transaction.log.TransactionIdStore.BASE_TX_LOG_VERSION) BASE_TX_CHECKSUM(org.neo4j.kernel.impl.transaction.log.TransactionIdStore.BASE_TX_CHECKSUM) PageCache(org.neo4j.io.pagecache.PageCache) InputNode(org.neo4j.unsafe.impl.batchimport.input.InputNode) Predicate(java.util.function.Predicate) Collection(java.util.Collection) StandardOpenOption(java.nio.file.StandardOpenOption) MOVE(org.neo4j.kernel.impl.storemigration.FileOperation.MOVE) LogService(org.neo4j.kernel.impl.logging.LogService) List(java.util.List) StoreFileType(org.neo4j.kernel.impl.storemigration.StoreFileType) MetaDataStore(org.neo4j.kernel.impl.store.MetaDataStore) Writer(java.io.Writer) Optional(java.util.Optional) Inputs(org.neo4j.unsafe.impl.batchimport.input.Inputs) InputIterable(org.neo4j.unsafe.impl.batchimport.InputIterable) InputRelationship(org.neo4j.unsafe.impl.batchimport.input.InputRelationship) TransactionIdStore(org.neo4j.kernel.impl.transaction.log.TransactionIdStore) InputEntity(org.neo4j.unsafe.impl.batchimport.input.InputEntity) DEFAULT_NAME(org.neo4j.kernel.impl.store.MetaDataStore.DEFAULT_NAME) RecordFormats(org.neo4j.kernel.impl.store.format.RecordFormats) CustomIOConfigValidator(org.neo4j.kernel.impl.util.CustomIOConfigValidator) AdditionalInitialIds(org.neo4j.unsafe.impl.batchimport.AdditionalInitialIds) NeoStores(org.neo4j.kernel.impl.store.NeoStores) StorePropertyCursor(org.neo4j.kernel.impl.api.store.StorePropertyCursor) CountsTracker(org.neo4j.kernel.impl.store.counts.CountsTracker) ExistingTargetStrategy(org.neo4j.kernel.impl.storemigration.ExistingTargetStrategy) PhysicalLogFiles(org.neo4j.kernel.impl.transaction.log.PhysicalLogFiles) RelationshipRecordFormat(org.neo4j.kernel.impl.store.format.standard.RelationshipRecordFormat) OutputStream(java.io.OutputStream) Config(org.neo4j.kernel.configuration.Config) ReadOnlyIdGeneratorFactory(org.neo4j.kernel.impl.store.id.ReadOnlyIdGeneratorFactory) LockService(org.neo4j.kernel.impl.locking.LockService) FileHandle(org.neo4j.io.pagecache.FileHandle) CapabilityType(org.neo4j.kernel.impl.store.format.CapabilityType) PrimitiveRecord(org.neo4j.kernel.impl.store.record.PrimitiveRecord) FIELD_NOT_PRESENT(org.neo4j.kernel.impl.store.format.standard.MetaDataRecordFormat.FIELD_NOT_PRESENT) StoreMigratorCheckPointer(org.neo4j.kernel.impl.storemigration.StoreMigratorCheckPointer) ExecutionMonitor(org.neo4j.unsafe.impl.batchimport.staging.ExecutionMonitor) BASE_TX_LOG_BYTE_OFFSET(org.neo4j.kernel.impl.transaction.log.TransactionIdStore.BASE_TX_LOG_BYTE_OFFSET) FileSystemAbstraction(org.neo4j.io.fs.FileSystemAbstraction) InputNode(org.neo4j.unsafe.impl.batchimport.input.InputNode) Configuration(org.neo4j.unsafe.impl.batchimport.Configuration) FileHandle(org.neo4j.io.pagecache.FileHandle) RecordCursors(org.neo4j.kernel.impl.store.RecordCursors) BufferedOutputStream(java.io.BufferedOutputStream) FileOutputStream(java.io.FileOutputStream) OutputStream(java.io.OutputStream) ArrayList(java.util.ArrayList) NoSuchFileException(java.nio.file.NoSuchFileException) InputRelationship(org.neo4j.unsafe.impl.batchimport.input.InputRelationship) ParallelBatchImporter(org.neo4j.unsafe.impl.batchimport.ParallelBatchImporter) BatchImporter(org.neo4j.unsafe.impl.batchimport.BatchImporter) ParallelBatchImporter(org.neo4j.unsafe.impl.batchimport.ParallelBatchImporter) NeoStores(org.neo4j.kernel.impl.store.NeoStores) FileOutputStream(java.io.FileOutputStream) StoreFile(org.neo4j.kernel.impl.storemigration.StoreFile) PagedFile(org.neo4j.io.pagecache.PagedFile) StoreFile(org.neo4j.kernel.impl.storemigration.StoreFile) File(java.io.File) BufferedOutputStream(java.io.BufferedOutputStream) AdditionalInitialIds(org.neo4j.unsafe.impl.batchimport.AdditionalInitialIds)

Aggregations

BufferedOutputStream (java.io.BufferedOutputStream)1 BufferedReader (java.io.BufferedReader)1 File (java.io.File)1 FileOutputStream (java.io.FileOutputStream)1 IOException (java.io.IOException)1 OutputStream (java.io.OutputStream)1 Writer (java.io.Writer)1 StandardCharsets (java.nio.charset.StandardCharsets)1 NoSuchFileException (java.nio.file.NoSuchFileException)1 StandardCopyOption (java.nio.file.StandardCopyOption)1 StandardOpenOption (java.nio.file.StandardOpenOption)1 ArrayList (java.util.ArrayList)1 Arrays (java.util.Arrays)1 Arrays.asList (java.util.Arrays.asList)1 Collection (java.util.Collection)1 List (java.util.List)1 Optional (java.util.Optional)1 BiConsumer (java.util.function.BiConsumer)1 Predicate (java.util.function.Predicate)1 Supplier (java.util.function.Supplier)1