You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hudi.apache.org by GitBox <gi...@apache.org> on 2022/07/25 11:37:43 UTC
[GitHub] [hudi] vinothchandar commented on a diff in pull request #5470: [HUDI-3993] Replacing UDF in Bulk Insert w/ RDD transformation
vinothchandar commented on code in PR #5470:
URL: https://github.com/apache/hudi/pull/5470#discussion_r928336189
##########
hudi-client/hudi-spark-client/src/main/java/org/apache/hudi/keygen/BuiltinKeyGenerator.java:
##########
@@ -18,26 +18,24 @@
package org.apache.hudi.keygen;
+import org.apache.avro.generic.GenericRecord;
import org.apache.hudi.ApiMaturityLevel;
import org.apache.hudi.AvroConversionUtils;
import org.apache.hudi.PublicAPIMethod;
import org.apache.hudi.common.config.TypedProperties;
import org.apache.hudi.common.util.collection.Pair;
-import org.apache.hudi.exception.HoodieIOException;
-
-import org.apache.avro.generic.GenericRecord;
+import org.apache.hudi.exception.HoodieException;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.catalyst.InternalRow;
import org.apache.spark.sql.types.DataType;
import org.apache.spark.sql.types.StructType;
+import scala.Function1;
Review Comment:
don't really like scala imports in Java (becomes an issue - for us one day when we want to shrink scala spread in code). Any way we can avoid this
##########
hudi-spark-datasource/hudi-spark-common/src/main/scala/org/apache/hudi/HoodieDatasetBulkInsertHelper.scala:
##########
@@ -0,0 +1,163 @@
+/*
+ * 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
+
+import org.apache.hudi.common.config.TypedProperties
+import org.apache.hudi.common.model.HoodieRecord
+import org.apache.hudi.common.util.ReflectionUtils
+import org.apache.hudi.config.HoodieWriteConfig
+import org.apache.hudi.keygen.BuiltinKeyGenerator
+import org.apache.hudi.table.BulkInsertPartitioner
+import org.apache.spark.internal.Logging
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.HoodieUnsafeRDDUtils.createDataFrame
+import org.apache.spark.sql.HoodieUnsafeRowUtils.{composeNestedFieldPath, getNestedInternalRowValue}
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.GenericInternalRow
+import org.apache.spark.sql.types.{StringType, StructField, StructType}
+import org.apache.spark.sql.{DataFrame, Dataset, HoodieUnsafeRDDUtils, Row}
+import org.apache.spark.unsafe.types.UTF8String
+
+import scala.collection.JavaConverters.asScalaBufferConverter
+
+object HoodieDatasetBulkInsertHelper extends Logging {
+
+ /**
+ * Prepares [[DataFrame]] for bulk-insert into Hudi table, taking following steps:
+ *
+ * <ol>
+ * <li>Invoking configured [[KeyGenerator]] to produce record key, alas partition-path value</li>
+ * <li>Prepends Hudi meta-fields to every row in the dataset</li>
+ * <li>Dedupes rows (if necessary)</li>
+ * <li>Partitions dataset using provided [[partitioner]]</li>
+ * </ol>
+ */
+ def prepareForBulkInsert(df: DataFrame,
+ config: HoodieWriteConfig,
+ partitioner: BulkInsertPartitioner[Dataset[Row]],
+ isGlobalIndex: Boolean,
+ dropPartitionColumns: Boolean): Dataset[Row] = {
+ val populateMetaFields = config.populateMetaFields()
+ val schema = df.schema
+
+ val keyGeneratorClassName = config.getStringOrThrow(DataSourceWriteOptions.KEYGENERATOR_CLASS_NAME,
+ "Key-generator class name is required")
+
+ val prependedRdd: RDD[InternalRow] =
+ df.queryExecution.toRdd.mapPartitions { iter =>
Review Comment:
Can you confirm this issue has been resolved in Master branch
##########
hudi-spark-datasource/hudi-spark-common/src/main/scala/org/apache/hudi/HoodieDatasetBulkInsertHelper.scala:
##########
@@ -0,0 +1,154 @@
+/*
+ * 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
+
+import org.apache.hudi.client.model.HoodieInternalRow
+import org.apache.hudi.common.config.TypedProperties
+import org.apache.hudi.common.model.HoodieRecord
+import org.apache.hudi.common.util.ReflectionUtils
+import org.apache.hudi.config.HoodieWriteConfig
+import org.apache.hudi.keygen.BuiltinKeyGenerator
+import org.apache.hudi.table.BulkInsertPartitioner
+import org.apache.spark.internal.Logging
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.HoodieUnsafeRDDUtils.createDataFrame
+import org.apache.spark.sql.HoodieUnsafeRowUtils.{composeNestedFieldPath, getNestedInternalRowValue}
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.types.{StringType, StructField, StructType}
+import org.apache.spark.sql.{DataFrame, Dataset, HoodieUnsafeRDDUtils, Row}
+import org.apache.spark.unsafe.types.UTF8String
+
+import scala.collection.JavaConverters.asScalaBufferConverter
+
+object HoodieDatasetBulkInsertHelper extends Logging {
+
+ /**
+ * Prepares [[DataFrame]] for bulk-insert into Hudi table, taking following steps:
+ *
+ * <ol>
+ * <li>Invoking configured [[KeyGenerator]] to produce record key, alas partition-path value</li>
+ * <li>Prepends Hudi meta-fields to every row in the dataset</li>
+ * <li>Dedupes rows (if necessary)</li>
+ * <li>Partitions dataset using provided [[partitioner]]</li>
+ * </ol>
+ */
+ def prepareForBulkInsert(df: DataFrame,
+ config: HoodieWriteConfig,
+ partitioner: BulkInsertPartitioner[Dataset[Row]],
+ isGlobalIndex: Boolean,
+ dropPartitionColumns: Boolean): Dataset[Row] = {
+ val populateMetaFields = config.populateMetaFields()
+ val schema = df.schema
+
+ val keyGeneratorClassName = config.getStringOrThrow(DataSourceWriteOptions.KEYGENERATOR_CLASS_NAME,
+ "Key-generator class name is required")
+
+ val prependedRdd: RDD[InternalRow] =
+ df.queryExecution.toRdd.mapPartitions { iter =>
+ val keyGenerator =
+ ReflectionUtils.loadClass(keyGeneratorClassName, new TypedProperties(config.getProps))
+ .asInstanceOf[BuiltinKeyGenerator]
+
+ iter.map { row =>
+ val (recordKey, partitionPath) =
+ if (populateMetaFields) {
+ (UTF8String.fromString(keyGenerator.getRecordKey(row, schema)),
+ UTF8String.fromString(keyGenerator.getPartitionPath(row, schema)))
+ } else {
+ (UTF8String.EMPTY_UTF8, UTF8String.EMPTY_UTF8)
+ }
+ val commitTimestamp = UTF8String.EMPTY_UTF8
+ val commitSeqNo = UTF8String.EMPTY_UTF8
+ val filename = UTF8String.EMPTY_UTF8
+
+ // TODO use mutable row, avoid re-allocating
+ new HoodieInternalRow(commitTimestamp, commitSeqNo, recordKey, partitionPath, filename, row, false)
+ }
+ }
+
+ val metaFields = Seq(
Review Comment:
We should follow up and consolidate into one list in HoodieRecord. +1. Unless the other usages break with or . different ordering, I don't see any reason why we won't
##########
hudi-client/hudi-spark-client/src/main/java/org/apache/hudi/keygen/BuiltinKeyGenerator.java:
##########
@@ -66,12 +66,26 @@ protected BuiltinKeyGenerator(TypedProperties config) {
@Override
@PublicAPIMethod(maturity = ApiMaturityLevel.EVOLVING)
public String getRecordKey(Row row) {
+ // TODO avoid conversion to avro
+ // since converterFn is transient this will be repeatedly initialized over and over again
if (null == converterFn) {
converterFn = AvroConversionUtils.createConverterToAvro(row.schema(), STRUCT_NAME, NAMESPACE);
}
return getKey(converterFn.apply(row)).getRecordKey();
}
+ @Override
+ @PublicAPIMethod(maturity = ApiMaturityLevel.EVOLVING)
+ public String getRecordKey(InternalRow internalRow, StructType schema) {
+ try {
Review Comment:
are resolution on this? Did you end up backing out the temporary changes
##########
hudi-client/hudi-spark-client/src/main/java/org/apache/hudi/io/storage/row/HoodieRowParquetWriteSupport.java:
##########
@@ -73,18 +75,15 @@ public WriteSupport.FinalizedWriteContext finalizeWrite() {
return new WriteSupport.FinalizedWriteContext(extraMetaData);
}
- public void add(String recordKey) {
- this.bloomFilter.add(recordKey);
- if (minRecordKey != null) {
- minRecordKey = minRecordKey.compareTo(recordKey) <= 0 ? minRecordKey : recordKey;
- } else {
- minRecordKey = recordKey;
+ public void add(UTF8String recordKey) {
+ this.bloomFilter.add(recordKey.getBytes());
Review Comment:
BloomFilter add does . So we seem to be fine. it's good to trust-but-verify once though that `recordKey.getBytes()` is equal to `string.getBytes(StandardCharsets.UTF_8)`. @alexeykudinkin you probably checked that during development?
##########
hudi-spark-datasource/hudi-spark-common/src/main/scala/org/apache/hudi/HoodieDatasetBulkInsertHelper.scala:
##########
@@ -0,0 +1,154 @@
+/*
+ * 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
+
+import org.apache.hudi.client.model.HoodieInternalRow
+import org.apache.hudi.common.config.TypedProperties
+import org.apache.hudi.common.model.HoodieRecord
+import org.apache.hudi.common.util.ReflectionUtils
+import org.apache.hudi.config.HoodieWriteConfig
+import org.apache.hudi.keygen.BuiltinKeyGenerator
+import org.apache.hudi.table.BulkInsertPartitioner
+import org.apache.spark.internal.Logging
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.HoodieUnsafeRDDUtils.createDataFrame
+import org.apache.spark.sql.HoodieUnsafeRowUtils.{composeNestedFieldPath, getNestedInternalRowValue}
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.types.{StringType, StructField, StructType}
+import org.apache.spark.sql.{DataFrame, Dataset, HoodieUnsafeRDDUtils, Row}
+import org.apache.spark.unsafe.types.UTF8String
+
+import scala.collection.JavaConverters.asScalaBufferConverter
+
+object HoodieDatasetBulkInsertHelper extends Logging {
+
+ /**
+ * Prepares [[DataFrame]] for bulk-insert into Hudi table, taking following steps:
+ *
+ * <ol>
+ * <li>Invoking configured [[KeyGenerator]] to produce record key, alas partition-path value</li>
+ * <li>Prepends Hudi meta-fields to every row in the dataset</li>
+ * <li>Dedupes rows (if necessary)</li>
+ * <li>Partitions dataset using provided [[partitioner]]</li>
+ * </ol>
+ */
+ def prepareForBulkInsert(df: DataFrame,
+ config: HoodieWriteConfig,
+ partitioner: BulkInsertPartitioner[Dataset[Row]],
+ isGlobalIndex: Boolean,
+ dropPartitionColumns: Boolean): Dataset[Row] = {
+ val populateMetaFields = config.populateMetaFields()
+ val schema = df.schema
+
+ val keyGeneratorClassName = config.getStringOrThrow(DataSourceWriteOptions.KEYGENERATOR_CLASS_NAME,
+ "Key-generator class name is required")
+
+ val prependedRdd: RDD[InternalRow] =
+ df.queryExecution.toRdd.mapPartitions { iter =>
+ val keyGenerator =
+ ReflectionUtils.loadClass(keyGeneratorClassName, new TypedProperties(config.getProps))
+ .asInstanceOf[BuiltinKeyGenerator]
+
+ iter.map { row =>
+ val (recordKey, partitionPath) =
+ if (populateMetaFields) {
+ (UTF8String.fromString(keyGenerator.getRecordKey(row, schema)),
+ UTF8String.fromString(keyGenerator.getPartitionPath(row, schema)))
+ } else {
+ (UTF8String.EMPTY_UTF8, UTF8String.EMPTY_UTF8)
+ }
+ val commitTimestamp = UTF8String.EMPTY_UTF8
+ val commitSeqNo = UTF8String.EMPTY_UTF8
+ val filename = UTF8String.EMPTY_UTF8
+
+ // TODO use mutable row, avoid re-allocating
+ new HoodieInternalRow(commitTimestamp, commitSeqNo, recordKey, partitionPath, filename, row, false)
+ }
+ }
+
+ val metaFields = Seq(
+ StructField(HoodieRecord.COMMIT_TIME_METADATA_FIELD, StringType),
+ StructField(HoodieRecord.COMMIT_SEQNO_METADATA_FIELD, StringType),
+ StructField(HoodieRecord.RECORD_KEY_METADATA_FIELD, StringType),
+ StructField(HoodieRecord.PARTITION_PATH_METADATA_FIELD, StringType),
+ StructField(HoodieRecord.FILENAME_METADATA_FIELD, StringType))
+
+ val updatedSchema = StructType(metaFields ++ schema.fields)
+ val updatedDF = HoodieUnsafeRDDUtils.createDataFrame(df.sparkSession, prependedRdd, updatedSchema)
+
+ if (!populateMetaFields) {
+ updatedDF
+ } else {
+ val trimmedDF = if (dropPartitionColumns) {
+ val keyGenerator = ReflectionUtils.loadClass(keyGeneratorClassName, new TypedProperties(config.getProps)).asInstanceOf[BuiltinKeyGenerator]
+ val partitionPathFields = keyGenerator.getPartitionPathFields.asScala
+ val nestedPartitionPathFields = partitionPathFields.filter(f => f.contains('.'))
+ if (nestedPartitionPathFields.nonEmpty) {
+ logWarning(s"Can not drop nested partition path fields: $nestedPartitionPathFields")
+ }
+
+ val partitionPathCols = partitionPathFields -- nestedPartitionPathFields
+ updatedDF.drop(partitionPathCols: _*)
+ } else {
+ updatedDF
+ }
+
+ val dedupedDF = if (config.shouldCombineBeforeInsert) {
+ dedupeRows(trimmedDF, config.getPreCombineField, isGlobalIndex)
+ } else {
+ trimmedDF
+ }
+
+ partitioner.repartitionRecords(dedupedDF, config.getBulkInsertShuffleParallelism)
+ }
+ }
+
+ private def dedupeRows(df: DataFrame, preCombineFieldRef: String, isGlobalIndex: Boolean): DataFrame = {
+ val recordKeyMetaFieldOrd = df.schema.fieldIndex(HoodieRecord.RECORD_KEY_METADATA_FIELD)
+ val partitionPathMetaFieldOrd = df.schema.fieldIndex(HoodieRecord.PARTITION_PATH_METADATA_FIELD)
+ // NOTE: Pre-combine field could be a nested field
+ val preCombineFieldPath = composeNestedFieldPath(df.schema, preCombineFieldRef)
+
+ val dedupedRdd = df.queryExecution.toRdd
+ .map { row =>
+ val rowKey = if (isGlobalIndex) {
+ row.getString(recordKeyMetaFieldOrd)
+ } else {
+ val partitionPath = row.getString(partitionPathMetaFieldOrd)
+ val recordKey = row.getString(recordKeyMetaFieldOrd)
+ s"$partitionPath:$recordKey"
+ }
+ // NOTE: It's critical whenever we keep the reference to the row, to make a copy
+ // since Spark might be providing us with a mutable copy (updated during the iteration)
+ (rowKey, row.copy())
Review Comment:
What exact scenarios cause Spark to fail without copy. Could you please expand on that?
##########
hudi-spark-datasource/hudi-spark-common/src/main/scala/org/apache/hudi/HoodieDatasetBulkInsertHelper.scala:
##########
@@ -0,0 +1,154 @@
+/*
+ * 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
+
+import org.apache.hudi.client.model.HoodieInternalRow
+import org.apache.hudi.common.config.TypedProperties
+import org.apache.hudi.common.model.HoodieRecord
+import org.apache.hudi.common.util.ReflectionUtils
+import org.apache.hudi.config.HoodieWriteConfig
+import org.apache.hudi.keygen.BuiltinKeyGenerator
+import org.apache.hudi.table.BulkInsertPartitioner
+import org.apache.spark.internal.Logging
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.HoodieUnsafeRDDUtils.createDataFrame
+import org.apache.spark.sql.HoodieUnsafeRowUtils.{composeNestedFieldPath, getNestedInternalRowValue}
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.types.{StringType, StructField, StructType}
+import org.apache.spark.sql.{DataFrame, Dataset, HoodieUnsafeRDDUtils, Row}
+import org.apache.spark.unsafe.types.UTF8String
+
+import scala.collection.JavaConverters.asScalaBufferConverter
+
+object HoodieDatasetBulkInsertHelper extends Logging {
Review Comment:
Trying to understand this better. Why did this need to be in scala?
--
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-unsubscribe@hudi.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org