nsivabalan commented on code in PR #8390: URL: https://github.com/apache/hudi/pull/8390#discussion_r1184620783
########## hudi-client/hudi-spark-client/src/main/java/org/apache/hudi/client/utils/SparkSampleWritesUtils.java: ########## @@ -0,0 +1,143 @@ +/* + * 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.client.utils; + +import org.apache.hudi.client.SparkRDDWriteClient; +import org.apache.hudi.client.WriteStatus; +import org.apache.hudi.client.common.HoodieSparkEngineContext; +import org.apache.hudi.common.config.HoodieMetadataConfig; +import org.apache.hudi.common.fs.FSUtils; +import org.apache.hudi.common.model.HoodieCommitMetadata; +import org.apache.hudi.common.model.HoodieRecord; +import org.apache.hudi.common.model.HoodieTableType; +import org.apache.hudi.common.table.HoodieTableMetaClient; +import org.apache.hudi.common.table.timeline.HoodieInstant; +import org.apache.hudi.common.util.Option; +import org.apache.hudi.common.util.collection.Pair; +import org.apache.hudi.config.HoodieWriteConfig; +import org.apache.hudi.hadoop.CachingPath; + +import org.apache.hadoop.fs.FileSystem; +import org.apache.hadoop.fs.Path; +import org.apache.spark.api.java.JavaRDD; +import org.apache.spark.api.java.JavaSparkContext; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import java.io.IOException; +import java.time.Instant; +import java.util.List; + +import static org.apache.hudi.common.table.HoodieTableMetaClient.SAMPLE_WRITES_FOLDER_PATH; +import static org.apache.hudi.common.util.ValidationUtils.checkState; +import static org.apache.hudi.config.HoodieCompactionConfig.COPY_ON_WRITE_RECORD_SIZE_ESTIMATE; +import static org.apache.hudi.config.HoodieWriteConfig.SAMPLE_WRITES_ENABLED; +import static org.apache.hudi.config.HoodieWriteConfig.SAMPLE_WRITES_SIZE; + +/** + * The utilities class is dedicated to estimating average record size by writing sample incoming records + * to `.hoodie/.aux/.sample_writes/<instant time>/<epoch millis>` and reading the commit metadata. + * + * TODO handle sample_writes sub-path clean-up w.r.t. rollback and insert overwrite. (HUDI-6044) + */ +public class SparkSampleWritesUtils { + + private static final Logger LOG = LoggerFactory.getLogger(SparkSampleWritesUtils.class); + + public static void overwriteRecordSizeEstimateIfNeeded(JavaSparkContext jsc, JavaRDD<HoodieRecord> records, HoodieWriteConfig writeConfig, String instantTime) { + if (!writeConfig.getBoolean(SAMPLE_WRITES_ENABLED)) { + LOG.debug("Skip overwriting record size estimate as it's disabled."); + return; + } + HoodieTableMetaClient metaClient = getMetaClient(jsc, writeConfig.getBasePath()); + if (metaClient.isTimelineNonEmpty()) { + LOG.info("Skip overwriting record size estimate due to timeline is non-empty."); + return; + } + try { + Pair<Boolean, String> result = doSampleWrites(jsc, records, writeConfig, instantTime); + if (result.getLeft()) { + long avgSize = getAvgSizeFromSampleWrites(jsc, result.getRight()); + LOG.info("Overwriting record size estimate to " + avgSize); + writeConfig.setValue(COPY_ON_WRITE_RECORD_SIZE_ESTIMATE, String.valueOf(avgSize)); + } + } catch (IOException e) { + LOG.error(String.format("Not overwriting record size estimate for table %s due to error when doing sample writes.", writeConfig.getTableName()), e); + } + } + + private static Pair<Boolean, String> doSampleWrites(JavaSparkContext jsc, JavaRDD<HoodieRecord> records, HoodieWriteConfig writeConfig, String instantTime) throws IOException { + long now = Instant.now().toEpochMilli(); + Path basePath = new CachingPath(writeConfig.getBasePath(), SAMPLE_WRITES_FOLDER_PATH + Path.SEPARATOR + instantTime + Path.SEPARATOR + now); + final String sampleWritesBasePath = basePath.toString(); + HoodieTableMetaClient.withPropertyBuilder() + .setTableType(HoodieTableType.COPY_ON_WRITE) + .setTableName(String.format("%s_samples_%s_%s", writeConfig.getTableName(), instantTime, now)) + .setCDCEnabled(false) + .initTable(jsc.hadoopConfiguration(), sampleWritesBasePath); + HoodieWriteConfig sampleWriteConfig = HoodieWriteConfig.newBuilder() + .withProps(writeConfig.getProps()) + .withPath(sampleWritesBasePath) + .withMetadataConfig(HoodieMetadataConfig.newBuilder().enable(false).build()) + .withSampleWritesEnabled(false) + .withTableServicesEnabled(false) + .withSchemaEvolutionEnable(false) + .withBulkInsertParallelism(1) + .withAutoCommit(true) + .build(); + try (SparkRDDWriteClient sampleWriteClient = new SparkRDDWriteClient(new HoodieSparkEngineContext(jsc), sampleWriteConfig, Option.empty())) { + int size = writeConfig.getIntOrDefault(SAMPLE_WRITES_SIZE); + List<HoodieRecord> samples = records.coalesce(1).take(size); Review Comment: so, wouldn't this OOM if we try to bring all records to just 1 spark partition? or since we do take(5k), spark is efficient and will not bring all records to 1 spark partition, but just 5k records. -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
