Search in sources :

Example 6 with BootstrapFileMapping

use of org.apache.hudi.common.model.BootstrapFileMapping in project hudi by apache.

the class SparkBootstrapCommitActionExecutor method commit.

@Override
protected void commit(Option<Map<String, String>> extraMetadata, HoodieWriteMetadata<HoodieData<WriteStatus>> result) {
    // Perform bootstrap index write and then commit. Make sure both record-key and bootstrap-index
    // is all done in a single job DAG.
    Map<String, List<Pair<BootstrapFileMapping, HoodieWriteStat>>> bootstrapSourceAndStats = result.getWriteStatuses().collectAsList().stream().map(w -> {
        BootstrapWriteStatus ws = (BootstrapWriteStatus) w;
        return Pair.of(ws.getBootstrapSourceFileMapping(), ws.getStat());
    }).collect(Collectors.groupingBy(w -> w.getKey().getPartitionPath()));
    HoodieTableMetaClient metaClient = table.getMetaClient();
    try (BootstrapIndex.IndexWriter indexWriter = BootstrapIndex.getBootstrapIndex(metaClient).createWriter(metaClient.getTableConfig().getBootstrapBasePath().get())) {
        LOG.info("Starting to write bootstrap index for source " + config.getBootstrapSourceBasePath() + " in table " + config.getBasePath());
        indexWriter.begin();
        bootstrapSourceAndStats.forEach((key, value) -> indexWriter.appendNextPartition(key, value.stream().map(Pair::getKey).collect(Collectors.toList())));
        indexWriter.finish();
        LOG.info("Finished writing bootstrap index for source " + config.getBootstrapSourceBasePath() + " in table " + config.getBasePath());
    }
    commit(extraMetadata, result, bootstrapSourceAndStats.values().stream().flatMap(f -> f.stream().map(Pair::getValue)).collect(Collectors.toList()));
    LOG.info("Committing metadata bootstrap !!");
}
Also used : HoodieTable(org.apache.hudi.table.HoodieTable) BootstrapMode(org.apache.hudi.client.bootstrap.BootstrapMode) FileSystem(org.apache.hadoop.fs.FileSystem) HoodieInstant(org.apache.hudi.common.table.timeline.HoodieInstant) BaseCommitActionExecutor(org.apache.hudi.table.action.commit.BaseCommitActionExecutor) HoodieJavaRDD(org.apache.hudi.data.HoodieJavaRDD) KeyGeneratorInterface(org.apache.hudi.keygen.KeyGeneratorInterface) Logger(org.apache.log4j.Logger) HoodieFileStatus(org.apache.hudi.avro.model.HoodieFileStatus) Duration(java.time.Duration) Map(java.util.Map) HoodieSparkEngineContext(org.apache.hudi.client.common.HoodieSparkEngineContext) HoodieWriteMetadata(org.apache.hudi.table.action.HoodieWriteMetadata) HoodieActiveTimeline(org.apache.hudi.common.table.timeline.HoodieActiveTimeline) HoodieSparkKeyGeneratorFactory(org.apache.hudi.keygen.factory.HoodieSparkKeyGeneratorFactory) ValidationUtils(org.apache.hudi.common.util.ValidationUtils) BootstrapPartitionPathTranslator(org.apache.hudi.client.bootstrap.translator.BootstrapPartitionPathTranslator) Collection(java.util.Collection) Instant(java.time.Instant) Collectors(java.util.stream.Collectors) StandardCharsets(java.nio.charset.StandardCharsets) List(java.util.List) WRITE_STATUS_STORAGE_LEVEL_VALUE(org.apache.hudi.config.HoodieWriteConfig.WRITE_STATUS_STORAGE_LEVEL_VALUE) HoodieWriteStat(org.apache.hudi.common.model.HoodieWriteStat) BootstrapWriteStatus(org.apache.hudi.client.bootstrap.BootstrapWriteStatus) SparkBulkInsertCommitActionExecutor(org.apache.hudi.table.action.commit.SparkBulkInsertCommitActionExecutor) WriteOperationType(org.apache.hudi.common.model.WriteOperationType) ReflectionUtils(org.apache.hudi.common.util.ReflectionUtils) FullRecordBootstrapDataProvider(org.apache.hudi.client.bootstrap.FullRecordBootstrapDataProvider) MetadataBootstrapHandlerFactory.getMetadataHandler(org.apache.hudi.table.action.bootstrap.MetadataBootstrapHandlerFactory.getMetadataHandler) HoodieBootstrapSchemaProvider(org.apache.hudi.client.bootstrap.HoodieBootstrapSchemaProvider) Option(org.apache.hudi.common.util.Option) HoodieCommitException(org.apache.hudi.exception.HoodieCommitException) HoodieKeyGeneratorException(org.apache.hudi.exception.HoodieKeyGeneratorException) State(org.apache.hudi.common.table.timeline.HoodieInstant.State) BaseSparkCommitActionExecutor(org.apache.hudi.table.action.commit.BaseSparkCommitActionExecutor) HoodieSparkTable(org.apache.hudi.table.HoodieSparkTable) HoodieTableMetaClient(org.apache.hudi.common.table.HoodieTableMetaClient) HoodieTimeline(org.apache.hudi.common.table.timeline.HoodieTimeline) JavaRDD(org.apache.spark.api.java.JavaRDD) BootstrapFileMapping(org.apache.hudi.common.model.BootstrapFileMapping) HoodieRecord(org.apache.hudi.common.model.HoodieRecord) BootstrapModeSelector(org.apache.hudi.client.bootstrap.selector.BootstrapModeSelector) HoodieData(org.apache.hudi.common.data.HoodieData) BootstrapIndex(org.apache.hudi.common.bootstrap.index.BootstrapIndex) TypedProperties(org.apache.hudi.common.config.TypedProperties) HoodieWriteConfig(org.apache.hudi.config.HoodieWriteConfig) WriteMarkersFactory(org.apache.hudi.table.marker.WriteMarkersFactory) Iterator(java.util.Iterator) HoodieCommitMetadata(org.apache.hudi.common.model.HoodieCommitMetadata) IOException(java.io.IOException) HoodieSparkBootstrapSchemaProvider(org.apache.hudi.client.bootstrap.HoodieSparkBootstrapSchemaProvider) WriteStatus(org.apache.hudi.client.WriteStatus) HoodieRecordPayload(org.apache.hudi.common.model.HoodieRecordPayload) SparkValidatorUtils(org.apache.hudi.client.utils.SparkValidatorUtils) HoodieKey(org.apache.hudi.common.model.HoodieKey) HoodieIOException(org.apache.hudi.exception.HoodieIOException) LogManager(org.apache.log4j.LogManager) FSUtils(org.apache.hudi.common.fs.FSUtils) Pair(org.apache.hudi.common.util.collection.Pair) HoodieTableMetaClient(org.apache.hudi.common.table.HoodieTableMetaClient) HoodieWriteStat(org.apache.hudi.common.model.HoodieWriteStat) BootstrapWriteStatus(org.apache.hudi.client.bootstrap.BootstrapWriteStatus) List(java.util.List) BootstrapIndex(org.apache.hudi.common.bootstrap.index.BootstrapIndex) BootstrapFileMapping(org.apache.hudi.common.model.BootstrapFileMapping)

Aggregations

BootstrapFileMapping (org.apache.hudi.common.model.BootstrapFileMapping)6 IOException (java.io.IOException)5 ArrayList (java.util.ArrayList)4 List (java.util.List)4 Collectors (java.util.stream.Collectors)4 BootstrapIndex (org.apache.hudi.common.bootstrap.index.BootstrapIndex)4 HoodieFileGroupId (org.apache.hudi.common.model.HoodieFileGroupId)4 Arrays (java.util.Arrays)3 Map (java.util.Map)3 Path (org.apache.hadoop.fs.Path)3 FSUtils (org.apache.hudi.common.fs.FSUtils)3 StandardCharsets (java.nio.charset.StandardCharsets)2 HashMap (java.util.HashMap)2 HashSet (java.util.HashSet)2 Set (java.util.Set)2 Stream (java.util.stream.Stream)2 HoodieFileStatus (org.apache.hudi.avro.model.HoodieFileStatus)2 BootstrapWriteStatus (org.apache.hudi.client.bootstrap.BootstrapWriteStatus)2 HFileBootstrapIndex (org.apache.hudi.common.bootstrap.index.HFileBootstrapIndex)2 HoodieCommitMetadata (org.apache.hudi.common.model.HoodieCommitMetadata)2