Search in sources :

Example 1 with DataGeneratorInput

use of org.neo4j.internal.batchimport.input.DataGeneratorInput in project neo4j by neo4j.

the class HumanUnderstandableExecutionMonitorIT method shouldReportProgressOfNodeImport.

@Test
void shouldReportProgressOfNodeImport() throws Exception {
    // given
    CapturingMonitor progress = new CapturingMonitor();
    HumanUnderstandableExecutionMonitor monitor = new HumanUnderstandableExecutionMonitor(progress);
    IdType idType = IdType.INTEGER;
    Input input = new DataGeneratorInput(NODE_COUNT, RELATIONSHIP_COUNT, idType, random.seed(), 0, bareboneNodeHeader(idType, new Extractors(';')), bareboneRelationshipHeader(idType, new Extractors(';')), 1, 1, 0, 0);
    Configuration configuration = new Configuration.Overridden(Configuration.DEFAULT) {

        @Override
        public long pageCacheMemory() {
            return mebiBytes(8);
        }
    };
    // when
    try (JobScheduler jobScheduler = new ThreadPoolJobScheduler()) {
        new ParallelBatchImporter(databaseLayout, fileSystem, NULL, configuration, NullLogService.getInstance(), monitor, EMPTY, defaults(), LATEST_RECORD_FORMATS, ImportLogic.NO_MONITOR, jobScheduler, Collector.EMPTY, LogFilesInitializer.NULL, IndexImporterFactory.EMPTY, EmptyMemoryTracker.INSTANCE).doImport(input);
        // then
        progress.assertAllProgressReachedEnd();
    }
}
Also used : JobScheduler(org.neo4j.scheduler.JobScheduler) ThreadPoolJobScheduler(org.neo4j.test.scheduler.ThreadPoolJobScheduler) ParallelBatchImporter(org.neo4j.internal.batchimport.ParallelBatchImporter) DataGeneratorInput(org.neo4j.internal.batchimport.input.DataGeneratorInput) Input(org.neo4j.internal.batchimport.input.Input) Extractors(org.neo4j.csv.reader.Extractors) Configuration(org.neo4j.internal.batchimport.Configuration) ThreadPoolJobScheduler(org.neo4j.test.scheduler.ThreadPoolJobScheduler) DataGeneratorInput(org.neo4j.internal.batchimport.input.DataGeneratorInput) IdType(org.neo4j.internal.batchimport.input.IdType) Test(org.junit.jupiter.api.Test)

Example 2 with DataGeneratorInput

use of org.neo4j.internal.batchimport.input.DataGeneratorInput in project neo4j by neo4j.

the class QuickImport method main.

public static void main(String[] arguments) throws IOException {
    Args args = Args.parse(arguments);
    long nodeCount = parseLongWithUnit(args.get("nodes", null));
    long relationshipCount = parseLongWithUnit(args.get("relationships", null));
    int labelCount = args.getNumber("labels", 4).intValue();
    int relationshipTypeCount = args.getNumber("relationship-types", 4).intValue();
    Path dir = Path.of(args.get("into"));
    long randomSeed = args.getNumber("random-seed", currentTimeMillis()).longValue();
    Configuration config = Configuration.COMMAS;
    Extractors extractors = new Extractors(config.arrayDelimiter());
    IdType idType = IdType.valueOf(args.get("id-type", IdType.INTEGER.name()));
    Groups groups = new Groups();
    Header nodeHeader = parseNodeHeader(args, idType, extractors, groups);
    Header relationshipHeader = parseRelationshipHeader(args, idType, extractors, groups);
    Config dbConfig;
    String dbConfigFileName = args.get("db-config", null);
    if (dbConfigFileName != null) {
        dbConfig = Config.newBuilder().fromFile(Path.of(dbConfigFileName)).build();
    } else {
        dbConfig = Config.defaults();
    }
    Boolean highIo = args.has("high-io") ? args.getBoolean("high-io") : null;
    LogProvider logging = NullLogProvider.getInstance();
    long pageCacheMemory = args.getNumber("pagecache-memory", org.neo4j.internal.batchimport.Configuration.MAX_PAGE_CACHE_MEMORY).longValue();
    org.neo4j.internal.batchimport.Configuration importConfig = new org.neo4j.internal.batchimport.Configuration.Overridden(defaultConfiguration(dir)) {

        @Override
        public int maxNumberOfProcessors() {
            return args.getNumber("processors", super.maxNumberOfProcessors()).intValue();
        }

        @Override
        public boolean highIO() {
            return highIo != null ? highIo : super.highIO();
        }

        @Override
        public long pageCacheMemory() {
            return pageCacheMemory;
        }

        @Override
        public long maxMemoryUsage() {
            String custom = args.get("max-memory", null);
            return custom != null ? parseMaxMemory(custom) : super.maxMemoryUsage();
        }

        @Override
        public IndexConfig indexConfig() {
            return IndexConfig.create().withLabelIndex().withRelationshipTypeIndex();
        }
    };
    float factorBadNodeData = args.getNumber("factor-bad-node-data", 0).floatValue();
    float factorBadRelationshipData = args.getNumber("factor-bad-relationship-data", 0).floatValue();
    Input input = new DataGeneratorInput(nodeCount, relationshipCount, idType, randomSeed, 0, nodeHeader, relationshipHeader, labelCount, relationshipTypeCount, factorBadNodeData, factorBadRelationshipData);
    try (FileSystemAbstraction fileSystem = new DefaultFileSystemAbstraction();
        Lifespan life = new Lifespan()) {
        BatchImporter consumer;
        if (args.getBoolean("to-csv")) {
            consumer = new CsvOutput(dir, nodeHeader, relationshipHeader, config);
        } else {
            System.out.println("Seed " + randomSeed);
            final JobScheduler jobScheduler = life.add(createScheduler());
            boolean verbose = args.getBoolean("v");
            ExecutionMonitor monitor = verbose ? new SpectrumExecutionMonitor(2, TimeUnit.SECONDS, System.out, 100) : defaultVisible();
            consumer = BatchImporterFactory.withHighestPriority().instantiate(DatabaseLayout.ofFlat(dir), fileSystem, PageCacheTracer.NULL, importConfig, new SimpleLogService(logging, logging), monitor, EMPTY, dbConfig, RecordFormatSelector.selectForConfig(dbConfig, logging), NO_MONITOR, jobScheduler, Collector.EMPTY, TransactionLogInitializer.getLogFilesInitializer(), new IndexImporterFactoryImpl(dbConfig), INSTANCE);
        }
        consumer.doImport(input);
    }
}
Also used : DefaultFileSystemAbstraction(org.neo4j.io.fs.DefaultFileSystemAbstraction) FileSystemAbstraction(org.neo4j.io.fs.FileSystemAbstraction) SpectrumExecutionMonitor(org.neo4j.internal.batchimport.staging.SpectrumExecutionMonitor) Configuration(org.neo4j.csv.reader.Configuration) Configuration.defaultConfiguration(org.neo4j.internal.batchimport.Configuration.defaultConfiguration) SimpleLogService(org.neo4j.logging.internal.SimpleLogService) Config(org.neo4j.configuration.Config) IndexConfig(org.neo4j.internal.batchimport.IndexConfig) DataGeneratorInput(org.neo4j.internal.batchimport.input.DataGeneratorInput) Input(org.neo4j.internal.batchimport.input.Input) BatchImporter(org.neo4j.internal.batchimport.BatchImporter) ParallelBatchImporter(org.neo4j.internal.batchimport.ParallelBatchImporter) Groups(org.neo4j.internal.batchimport.input.Groups) SpectrumExecutionMonitor(org.neo4j.internal.batchimport.staging.SpectrumExecutionMonitor) ExecutionMonitor(org.neo4j.internal.batchimport.staging.ExecutionMonitor) Path(java.nio.file.Path) JobScheduler(org.neo4j.scheduler.JobScheduler) Args(org.neo4j.internal.helpers.Args) DefaultFileSystemAbstraction(org.neo4j.io.fs.DefaultFileSystemAbstraction) IdType(org.neo4j.internal.batchimport.input.IdType) LogProvider(org.neo4j.logging.LogProvider) NullLogProvider(org.neo4j.logging.NullLogProvider) Extractors(org.neo4j.csv.reader.Extractors) Header(org.neo4j.internal.batchimport.input.csv.Header) IndexImporterFactoryImpl(org.neo4j.kernel.impl.index.schema.IndexImporterFactoryImpl) DataGeneratorInput(org.neo4j.internal.batchimport.input.DataGeneratorInput) Lifespan(org.neo4j.kernel.lifecycle.Lifespan)

Aggregations

Extractors (org.neo4j.csv.reader.Extractors)2 ParallelBatchImporter (org.neo4j.internal.batchimport.ParallelBatchImporter)2 DataGeneratorInput (org.neo4j.internal.batchimport.input.DataGeneratorInput)2 IdType (org.neo4j.internal.batchimport.input.IdType)2 Input (org.neo4j.internal.batchimport.input.Input)2 JobScheduler (org.neo4j.scheduler.JobScheduler)2 Path (java.nio.file.Path)1 Test (org.junit.jupiter.api.Test)1 Config (org.neo4j.configuration.Config)1 Configuration (org.neo4j.csv.reader.Configuration)1 BatchImporter (org.neo4j.internal.batchimport.BatchImporter)1 Configuration (org.neo4j.internal.batchimport.Configuration)1 Configuration.defaultConfiguration (org.neo4j.internal.batchimport.Configuration.defaultConfiguration)1 IndexConfig (org.neo4j.internal.batchimport.IndexConfig)1 Groups (org.neo4j.internal.batchimport.input.Groups)1 Header (org.neo4j.internal.batchimport.input.csv.Header)1 ExecutionMonitor (org.neo4j.internal.batchimport.staging.ExecutionMonitor)1 SpectrumExecutionMonitor (org.neo4j.internal.batchimport.staging.SpectrumExecutionMonitor)1 Args (org.neo4j.internal.helpers.Args)1 DefaultFileSystemAbstraction (org.neo4j.io.fs.DefaultFileSystemAbstraction)1