prashantwason commented on code in PR #8684: URL: https://github.com/apache/hudi/pull/8684#discussion_r1192137497
########## hudi-client/hudi-spark-client/src/main/java/org/apache/hudi/metadata/SparkHoodieMetadataBulkInsertPartitioner.java: ########## @@ -0,0 +1,109 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.hudi.metadata; + +import java.io.Serializable; +import java.util.ArrayList; +import java.util.Comparator; +import java.util.List; + +import org.apache.hudi.common.model.HoodieRecord; +import org.apache.hudi.common.util.ValidationUtils; +import org.apache.hudi.table.BulkInsertPartitioner; +import org.apache.spark.Partitioner; +import org.apache.spark.api.java.JavaRDD; + +import scala.Tuple2; + +/** + * A {@code BulkInsertPartitioner} implementation for Metadata Table to improve performance of initialization of metadata + * table partition when a very large number of records are inserted. + * + * This partitioner requires the records to be already tagged with location. + */ +public class SparkHoodieMetadataBulkInsertPartitioner implements BulkInsertPartitioner<JavaRDD<HoodieRecord>> { + final int numPartitions; + public SparkHoodieMetadataBulkInsertPartitioner(int numPartitions) { + this.numPartitions = numPartitions; + } + + private class FileGroupPartitioner extends Partitioner { + + @Override + public int getPartition(Object key) { + return ((Tuple2<Integer, String>)key)._1; + } + + @Override + public int numPartitions() { + return numPartitions; + } + } + + // FileIDs for the various partitions + private List<String> fileIDPfxs; + + /** + * Partition the records by their location. The number of partitions is determined by the number of MDT fileGroups being udpated rather than the + * specific value of outputSparkPartitions. + */ + @Override + public JavaRDD<HoodieRecord> repartitionRecords(JavaRDD<HoodieRecord> records, int outputSparkPartitions) { + Comparator<Tuple2<Integer, String>> keyComparator = + (Comparator<Tuple2<Integer, String>> & Serializable)(t1, t2) -> t1._2.compareTo(t2._2); + + // Partition the records by their file group + JavaRDD<HoodieRecord> partitionedRDD = records + // key by <file group index, record key>. The file group index is used to partition and the record key is used to sort within the partition. + .keyBy(r -> { + int fileGroupIndex = HoodieTableMetadataUtil.getFileGroupIndexFromFileId(r.getCurrentLocation().getFileId()); + return new Tuple2<>(fileGroupIndex, r.getRecordKey()); + }) + .repartitionAndSortWithinPartitions(new FileGroupPartitioner(), keyComparator) + .map(t -> t._2); + + fileIDPfxs = partitionedRDD.mapPartitions(recordItr -> { + // Due to partitioning, all record in the partition should have same fileID. So we only can get the fileID prefix from the first record. + List<String> fileIds = new ArrayList<>(1); + if (recordItr.hasNext()) { + HoodieRecord record = recordItr.next(); + final String fileID = HoodieTableMetadataUtil.getFileGroupPrefix(record.getCurrentLocation().getFileId()); + fileIds.add(fileID); + } else { + // Empty partition + fileIds.add(""); Review Comment: Added doc -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: commits-unsubscr...@hudi.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
