anishshri-db commented on code in PR #47445:
URL: https://github.com/apache/spark/pull/47445#discussion_r1687023328
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/v2/state/metadata/StateMetadataSource.scala:
##########
@@ -188,29 +191,56 @@ class StateMetadataPartitionReader(
} else Array.empty
}
- private def allOperatorStateMetadata: Array[OperatorStateMetadata] = {
+ private[sql] def allOperatorStateMetadata: Array[(OperatorStateMetadata,
Long)] = {
val stateDir = new Path(checkpointLocation, "state")
val opIds = fileManager
.list(stateDir, pathNameCanBeParsedAsLongFilter).map(f =>
pathToLong(f.getPath)).sorted
- opIds.map { opId =>
- new OperatorStateMetadataReader(new Path(stateDir, opId.toString),
hadoopConf).read()
+ opIds.flatMap { opId =>
+ val operatorIdPath = new Path(stateDir, opId.toString)
+ // check if OperatorStateMetadataV2 path exists, if it does, read it
+ // otherwise, fall back to OperatorStateMetadataV1
+ val operatorStateMetadataV2Path =
OperatorStateMetadataV2.metadataFilePath(operatorIdPath)
+ if (fileManager.exists(operatorStateMetadataV2Path)) {
+ val operatorStateMetadataLog = new OperatorStateMetadataLog(
+ hadoopConf, operatorStateMetadataV2Path.toString)
+ operatorStateMetadataLog.listBatchesOnDisk.flatMap { batchId =>
+ operatorStateMetadataLog.get(batchId).map((_, batchId))
+ }
+ } else {
+ Array((new OperatorStateMetadataReader(operatorIdPath,
hadoopConf).read(), -1L))
+ }
}
}
private[sql] lazy val stateMetadata: Iterator[StateMetadataTableEntry] = {
- allOperatorStateMetadata.flatMap { operatorStateMetadata =>
- require(operatorStateMetadata.version == 1)
- val operatorStateMetadataV1 =
operatorStateMetadata.asInstanceOf[OperatorStateMetadataV1]
- operatorStateMetadataV1.stateStoreInfo.map { stateStoreMetadata =>
-
StateMetadataTableEntry(operatorStateMetadataV1.operatorInfo.operatorId,
- operatorStateMetadataV1.operatorInfo.operatorName,
- stateStoreMetadata.storeName,
- stateStoreMetadata.numPartitions,
- if (batchIds.nonEmpty) batchIds.head else -1,
- if (batchIds.nonEmpty) batchIds.last else -1,
- stateStoreMetadata.numColsPrefixKey
- )
+ allOperatorStateMetadata.flatMap { case (operatorStateMetadata, batchId) =>
+ require(operatorStateMetadata.version == 1 ||
operatorStateMetadata.version == 2)
+ operatorStateMetadata match {
+ case v1: OperatorStateMetadataV1 =>
+ v1.stateStoreInfo.map { stateStoreMetadata =>
+ StateMetadataTableEntry(v1.operatorInfo.operatorId,
+ v1.operatorInfo.operatorName,
+ stateStoreMetadata.storeName,
+ stateStoreMetadata.numPartitions,
+ if (batchIds.nonEmpty) batchIds.head else -1,
+ if (batchIds.nonEmpty) batchIds.last else -1,
+ "",
+ stateStoreMetadata.numColsPrefixKey
+ )
+ }
+ case v2: OperatorStateMetadataV2 =>
+ v2.stateStoreInfo.map { stateStoreMetadata =>
+ StateMetadataTableEntry(v2.operatorInfo.operatorId,
+ v2.operatorInfo.operatorName,
+ stateStoreMetadata.storeName,
+ stateStoreMetadata.numPartitions,
+ batchId,
+ batchId,
+ v2.operatorPropertiesJson,
+ stateStoreMetadata.numColsPrefixKey
Review Comment:
For v2 - this is kind of misleading. Could we just remove this ?
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
