vamsikarnika commented on code in PR #17460:
URL: https://github.com/apache/hudi/pull/17460#discussion_r2601992535
##########
hudi-spark-datasource/hudi-spark/src/main/scala/org/apache/spark/sql/hudi/command/procedures/ShowCompactionProcedure.scala:
##########
@@ -57,24 +130,185 @@ class ShowCompactionProcedure extends BaseProcedure with
ProcedureBuilder with S
val tableName = getArgValueOrDefault(args, PARAMETERS(0))
val tablePath = getArgValueOrDefault(args, PARAMETERS(1))
val limit = getArgValueOrDefault(args, PARAMETERS(2)).get.asInstanceOf[Int]
+ val showArchived = getArgValueOrDefault(args,
PARAMETERS(3)).get.asInstanceOf[Boolean]
+ val filter = getArgValueOrDefault(args,
PARAMETERS(4)).get.asInstanceOf[String]
+ val startTime = getArgValueOrDefault(args,
PARAMETERS(5)).get.asInstanceOf[String]
+ val endTime = getArgValueOrDefault(args,
PARAMETERS(6)).get.asInstanceOf[String]
+ validateFilter(filter, outputType)
val basePath: String = getBasePath(tableName, tablePath)
val metaClient = createMetaClient(jsc, basePath)
- assert(metaClient.getTableType == HoodieTableType.MERGE_ON_READ,
- s"Cannot show compaction on a Non Merge On Read table.")
- val compactionInstants =
metaClient.getActiveTimeline.getInstants.iterator().asScala
+ if (metaClient.getTableType != HoodieTableType.MERGE_ON_READ) {
+ throw new IllegalArgumentException("Cannot show compaction on a Non
Merge On Read table.")
+ }
+
+ val activeResults =
getCombinedCompactionsWithPartitionMetadata(metaClient.getActiveTimeline,
limit, metaClient, startTime, endTime, "ACTIVE")
+ val finalResults = if (showArchived) {
+ val archivedResults =
getCombinedCompactionsWithPartitionMetadata(metaClient.getArchivedTimeline,
limit, metaClient, startTime, endTime, "ARCHIVED")
+ val combinedResults = (activeResults ++ archivedResults)
+ .sortWith((a, b) => a.getString(0) > b.getString(0))
+ if (startTime.trim.nonEmpty && endTime.trim.nonEmpty) {
+ combinedResults
+ } else {
+ combinedResults.take(limit)
+ }
+ } else {
+ if (startTime.trim.nonEmpty && endTime.trim.nonEmpty) {
+ activeResults
+ } else {
+ activeResults.take(limit)
+ }
+ }
+ applyFilter(finalResults, filter, outputType)
+ }
+
+ private def getCombinedCompactionsWithPartitionMetadata(timeline:
HoodieTimeline,
+ limit: Int,
+ metaClient:
HoodieTableMetaClient,
+ startTime: String,
+ endTime: String,
+ timelineType:
String): Seq[Row] = {
+ import scala.collection.mutable.ListBuffer
+
+ val allRows = ListBuffer[Row]()
+
+ val filteredCompactionInstants = timeline.getInstants.iterator().asScala
.filter(p => p.getAction == HoodieTimeline.COMPACTION_ACTION ||
p.getAction == HoodieTimeline.COMMIT_ACTION)
+ .filter { instant =>
+ val instantTime = instant.requestedTime()
+ val withinStartTime = if (startTime.nonEmpty) instantTime >= startTime
else true
+ val withinEndTime = if (endTime.nonEmpty) instantTime <= endTime else
true
+ withinStartTime && withinEndTime
+ }
.toSeq
- .sortBy(f => f.requestedTime)
- .reverse
- .take(limit)
-
- compactionInstants.map(instant =>
- (instant, CompactionUtils.getCompactionPlan(metaClient,
instant.requestedTime))
- ).map { case (instant, plan) =>
- Row(instant.requestedTime, plan.getOperations.size(),
instant.getState.name())
+ val allCompactionInstants = if (startTime.nonEmpty && endTime.nonEmpty) {
+ filteredCompactionInstants.sortBy(_.requestedTime).reverse
+ } else {
+ filteredCompactionInstants.sortBy(_.requestedTime).reverse.take(limit)
+ }
+
+ allCompactionInstants.foreach { compactionInstant =>
+ if (compactionInstant.getState == HoodieInstant.State.COMPLETED) {
+ Try {
+ val compactionMetadata =
timeline.readCommitMetadata(compactionInstant)
+ compactionMetadata.getPartitionToWriteStats.entrySet.asScala.foreach
{ partitionEntry =>
+ val partitionPath = partitionEntry.getKey
+ val writeStats = partitionEntry.getValue.asScala
+ val operationSize = Try {
+ val compactionPlan =
CompactionUtils.getCompactionPlan(metaClient, compactionInstant.requestedTime)
+ compactionPlan.getOperations.size()
+ }.getOrElse(null)
+
Review Comment:
can we read compaction plan only once for each instant instead of reading it
for every partition?
--
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]
