Skip to content

[HUDI-8842] Support skipping compaction and cluster for spark increme… #12639

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
danny0405 merged 1 commit into from
Jan 16, 2025
Merged

[HUDI-8842] Support skipping compaction and cluster for spark increme… #12639

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 14 additions & 0 deletions ...spark-datasource/hudi-spark-common/src/main/scala/org/apache/hudi/DataSourceOptions.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -151,6 +151,20 @@ object DataSourceReadOptions {
.withDocumentation("For the use-cases like users only want to incremental pull from certain partitions "
+ "instead of the full table. This option allows using glob pattern to directly filter on path.")

val INCREMENTAL_READ_SKIP_COMPACT: ConfigProperty[Boolean] = ConfigProperty
.key("hoodie.datasource.read.incr.skip_compact")
.defaultValue(false)
.markAdvanced()
.withDocumentation("Whether to skip compaction instants and avoid reading compacted base files for streaming "
+ "read to improve read performance.")

val INCREMENTAL_READ_SKIP_CLUSTER: ConfigProperty[Boolean] = ConfigProperty
.key("hoodie.datasource.read.incr.skip_cluster")
.defaultValue(false)
.markAdvanced()
.withDocumentation("Whether to skip clustering instants to avoid reading base files of clustering operations "
+ "for streaming read to improve read performance.")

val TIME_TRAVEL_AS_OF_INSTANT: ConfigProperty[String] = HoodieCommonConfig.TIMESTAMP_AS_OF

val ENABLE_DATA_SKIPPING: ConfigProperty[Boolean] = ConfigProperty
Expand Down
4 changes: 4 additions & 0 deletions ...e/hudi-spark-common/src/main/scala/org/apache/hudi/MergeOnReadIncrementalRelationV2.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -200,6 +200,10 @@ trait HoodieIncrementalRelationV2Trait extends HoodieBaseRelation {
.metaClient(metaClient)
.startCompletionTime(optParams(DataSourceReadOptions.START_COMMIT.key))
.endCompletionTime(optParams.getOrElse(DataSourceReadOptions.END_COMMIT.key, null))
.skipClustering(optParams.getOrElse(DataSourceReadOptions.INCREMENTAL_READ_SKIP_CLUSTER.key(),
Copy link
Contributor

@danny0405 danny0405 Jan 15, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does this change also fix the DeltaStreamer streaming read?

Copy link
Collaborator Author

@cshuo cshuo Jan 15, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does this change also fix the DeltaStreamer streaming read?

No, this change only covers spark struct streaming as the title describes.

String.valueOf(DataSourceReadOptions.INCREMENTAL_READ_SKIP_CLUSTER.defaultValue)).toBoolean)
.skipCompaction(optParams.getOrElse(DataSourceReadOptions.INCREMENTAL_READ_SKIP_COMPACT.key(),
String.valueOf(DataSourceReadOptions.INCREMENTAL_READ_SKIP_COMPACT.defaultValue)).toBoolean)
.rangeType(rangeType)
.build()
.analyze()
Expand Down
122 changes: 88 additions & 34 deletions ...datasource/hudi-spark/src/test/scala/org/apache/hudi/functional/TestStreamingSource.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,12 +17,13 @@

package org.apache.hudi.functional

import org.apache.hudi.DataSourceReadOptions
import org.apache.hudi.DataSourceReadOptions.{START_OFFSET, STREAMING_READ_TABLE_VERSION}
import org.apache.hudi.DataSourceWriteOptions.{PRECOMBINE_FIELD, RECORDKEY_FIELD}
import org.apache.hudi.common.model.HoodieTableType.{COPY_ON_WRITE, MERGE_ON_READ}
import org.apache.hudi.common.table.{HoodieTableMetaClient, HoodieTableVersion}
import org.apache.hudi.common.table.timeline.HoodieTimeline
import org.apache.hudi.config.HoodieCompactionConfig
import org.apache.hudi.config.{HoodieClusteringConfig, HoodieCompactionConfig}
import org.apache.hudi.config.HoodieWriteConfig.{DELETE_PARALLELISM_VALUE, INSERT_PARALLELISM_VALUE, TBL_NAME, UPSERT_PARALLELISM_VALUE, WRITE_TABLE_VERSION}
import org.apache.hudi.hadoop.fs.HadoopFSUtils
import org.apache.hudi.util.JavaConversions
Expand Down Expand Up @@ -210,41 +211,89 @@ class TestStreamingSource extends StreamTest {
}
}

test("test mor stream source with compaction") {
withTempDir { inputDir =>
val tablePath = s"${inputDir.getCanonicalPath}/test_mor_stream"
val metaClient = HoodieTableMetaClient.newTableBuilder()
.setTableType(MERGE_ON_READ)
.setTableName(getTableName(tablePath))
.setRecordKeyFields("id")
.setPreCombineField("ts")
.initTable(HadoopFSUtils.getStorageConf(spark.sessionState.newHadoopConf()), tablePath)
test("Test mor streaming source with clustering") {
Array("true", "false").foreach(skipCluster => {
withTempDir { inputDir =>
val tablePath = s"${inputDir.getCanonicalPath}/test_mor_stream_cluster"
val metaClient = HoodieTableMetaClient.newTableBuilder()
.setTableType(MERGE_ON_READ)
.setTableName(getTableName(tablePath))
.setRecordKeyFields("id")
.setPreCombineField("ts")
.initTable(HadoopFSUtils.getStorageConf(spark.sessionState.newHadoopConf()), tablePath)

addData(tablePath, Seq(("1", "a1", "10", "000")))
addData(tablePath, Seq(("2", "a1", "11", "001")))
addData(tablePath, Seq(("3", "a1", "12", "002")))
addData(tablePath, Seq(("4", "a1", "13", "003")), enableInlineCluster = true)
addData(tablePath, Seq(("5", "a1", "14", "004")))

val timestamp =
metaClient.getActiveTimeline.getCommitsTimeline.filterCompletedInstants()
.firstInstant().get().getCompletionTime

addData(tablePath, Seq(("1", "a1", "10", "000")))
val df = spark.readStream
.format("org.apache.hudi")
.load(tablePath)
.select("id", "name", "price", "ts")
val df = spark.readStream
.format("org.apache.hudi")
.option(START_OFFSET.key(), timestamp)
.option(DataSourceReadOptions.INCREMENTAL_READ_SKIP_CLUSTER.key(), skipCluster)
.load(tablePath)
.select("id", "name", "price", "ts")

addData(tablePath,
Seq(("1", "a2", "12", "000"),
("2", "a3", "12", "000")))
addData(tablePath, Seq(("2", "a5", "12", "000"), ("1", "a6", "12", "001")))
// trigger compaction
addData(tablePath, Seq(("3", "a6", "12", "000")), true)
testStream(df)(
AssertOnQuery { q => q.processAllAvailable(); true },
// Start after the first commit
CheckAnswerRows(Seq(
Row("2", "a1", "11", "001"),
Row("3", "a1", "12", "002"),
Row("4", "a1", "13", "003"),
Row("5", "a1", "14", "004")), lastOnly = true, isSorted = false)
)
assertTrue(metaClient.reloadActiveTimeline
.filter(JavaConversions.getPredicate(
e => e.isCompleted && HoodieTimeline.REPLACE_COMMIT_ACTION.equals(e.getAction)))
.countInstants() > 0)
}
})
}

testStream(df)(
AssertOnQuery {q => q.processAllAvailable(); true },
CheckAnswerRows(Seq(Row("1", "a6", "12", "001"),
Row("2", "a5", "12", "000"),
Row("3", "a6", "12", "000")), lastOnly = true, isSorted = false),
StopStream
)
assertTrue(metaClient.reloadActiveTimeline
.filter(JavaConversions.getPredicate(
e => e.isCompleted && HoodieTimeline.COMMIT_ACTION.equals(e.getAction)))
.countInstants() > 0)
}
test("test mor stream source with compaction") {
Array("true", "false").foreach(skipCompact => {
withTempDir { inputDir =>
val tablePath = s"${inputDir.getCanonicalPath}/test_mor_stream"
val metaClient = HoodieTableMetaClient.newTableBuilder()
.setTableType(MERGE_ON_READ)
.setTableName(getTableName(tablePath))
.setRecordKeyFields("id")
.setPreCombineField("ts")
.initTable(HadoopFSUtils.getStorageConf(spark.sessionState.newHadoopConf()), tablePath)

addData(tablePath, Seq(("1", "a1", "10", "000")))
val df = spark.readStream
.format("org.apache.hudi")
.option(DataSourceReadOptions.INCREMENTAL_READ_SKIP_COMPACT.key(), skipCompact)
.load(tablePath)
.select("id", "name", "price", "ts")

addData(tablePath,
Seq(("1", "a2", "12", "000"),
("2", "a3", "12", "000")))
addData(tablePath, Seq(("2", "a5", "12", "000"), ("1", "a6", "12", "001")))
// trigger compaction
addData(tablePath, Seq(("3", "a6", "12", "000")), enableInlineCompaction = true)

testStream(df)(
AssertOnQuery {q => q.processAllAvailable(); true },
CheckAnswerRows(Seq(Row("1", "a6", "12", "001"),
Row("2", "a5", "12", "000"),
Row("3", "a6", "12", "000")), lastOnly = true, isSorted = false),
StopStream
)
assertTrue(metaClient.reloadActiveTimeline
.filter(JavaConversions.getPredicate(
e => e.isCompleted && HoodieTimeline.COMMIT_ACTION.equals(e.getAction)))
.countInstants() > 0)
}
})
}

test("Test checkpoint translation") {
Expand Down Expand Up @@ -289,14 +338,19 @@ class TestStreamingSource extends StreamTest {
}
}

private def addData(inputPath: String, rows: Seq[(String, String, String, String)], enableInlineCompaction: Boolean = false) : Unit = {
private def addData(inputPath: String,
rows: Seq[(String, String, String, String)],
enableInlineCompaction: Boolean = false,
enableInlineCluster: Boolean = false) : Unit = {
rows.toDF(columns: _*)
.write
.format("org.apache.hudi")
.options(commonOptions)
.option(TBL_NAME.key, getTableName(inputPath))
.option(HoodieCompactionConfig.INLINE_COMPACT.key(), enableInlineCompaction.toString)
.option(HoodieCompactionConfig.INLINE_COMPACT_NUM_DELTA_COMMITS.key(), "2")
.option(HoodieClusteringConfig.INLINE_CLUSTERING.key(), enableInlineCluster.toString)
.option(HoodieClusteringConfig.INLINE_CLUSTERING_MAX_COMMITS.key(), "2")
.mode(SaveMode.Append)
.save(inputPath)
}
Expand Down
Loading