HeartSaVioR commented on code in PR #44884:
URL: https://github.com/apache/spark/pull/44884#discussion_r1480841542
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/TransformWithStateExec.scala:
##########
@@ -155,23 +161,112 @@ case class TransformWithStateExec(
override protected def doExecute(): RDD[InternalRow] = {
metrics // force lazy init at driver
- child.execute().mapPartitionsWithStateStore[InternalRow](
- getStateInfo,
- schemaForKeyRow,
- schemaForValueRow,
- numColsPrefixKey = 0,
- session.sqlContext.sessionState,
- Some(session.sqlContext.streams.stateStoreCoordinator),
- useColumnFamilies = true
- ) {
- case (store: StateStore, singleIterator: Iterator[InternalRow]) =>
- val processorHandle = new StatefulProcessorHandleImpl(store,
getStateInfo.queryRunId,
- keyEncoder)
- assert(processorHandle.getHandleState ==
StatefulProcessorHandleState.CREATED)
- statefulProcessor.init(processorHandle, outputMode)
-
processorHandle.setHandleState(StatefulProcessorHandleState.INITIALIZED)
- val result = processDataWithPartition(singleIterator, store,
processorHandle)
- result
+ if (isStreaming) {
+ child.execute().mapPartitionsWithStateStore[InternalRow](
+ getStateInfo,
+ schemaForKeyRow,
+ schemaForValueRow,
+ numColsPrefixKey = 0,
+ session.sqlContext.sessionState,
+ Some(session.sqlContext.streams.stateStoreCoordinator),
+ useColumnFamilies = true
+ ) {
+ case (store: StateStore, singleIterator: Iterator[InternalRow]) =>
+ processData(store, singleIterator)
+ }
+ } else {
+ // If the query is running in batch mode, we need to create a new
StateStore and instantiate
+ // a temp directory on the executors in mapPartitionsWithIndex.
+ child.execute().mapPartitionsWithIndex[InternalRow](
+ (i, iter) => {
+ val providerId = {
+ // lazy creation to initialize tempDirPath once
+ lazy val tempDirPath = Utils.createTempDir().getAbsolutePath
Review Comment:
Please see the question again - my question was "will this be evaluate once
per executor **after serde**?". If then this is a good optimization, otherwise
lazy val does nothing and it just confuses people.
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/TransformWithStateExec.scala:
##########
@@ -155,23 +161,112 @@ case class TransformWithStateExec(
override protected def doExecute(): RDD[InternalRow] = {
metrics // force lazy init at driver
- child.execute().mapPartitionsWithStateStore[InternalRow](
- getStateInfo,
- schemaForKeyRow,
- schemaForValueRow,
- numColsPrefixKey = 0,
- session.sqlContext.sessionState,
- Some(session.sqlContext.streams.stateStoreCoordinator),
- useColumnFamilies = true
- ) {
- case (store: StateStore, singleIterator: Iterator[InternalRow]) =>
- val processorHandle = new StatefulProcessorHandleImpl(store,
getStateInfo.queryRunId,
- keyEncoder)
- assert(processorHandle.getHandleState ==
StatefulProcessorHandleState.CREATED)
- statefulProcessor.init(processorHandle, outputMode)
-
processorHandle.setHandleState(StatefulProcessorHandleState.INITIALIZED)
- val result = processDataWithPartition(singleIterator, store,
processorHandle)
- result
+ if (isStreaming) {
+ child.execute().mapPartitionsWithStateStore[InternalRow](
+ getStateInfo,
+ schemaForKeyRow,
+ schemaForValueRow,
+ numColsPrefixKey = 0,
+ session.sqlContext.sessionState,
+ Some(session.sqlContext.streams.stateStoreCoordinator),
+ useColumnFamilies = true
+ ) {
+ case (store: StateStore, singleIterator: Iterator[InternalRow]) =>
+ processData(store, singleIterator)
+ }
+ } else {
+ // If the query is running in batch mode, we need to create a new
StateStore and instantiate
+ // a temp directory on the executors in mapPartitionsWithIndex.
+ child.execute().mapPartitionsWithIndex[InternalRow](
+ (i, iter) => {
+ val providerId = {
+ // lazy creation to initialize tempDirPath once
+ lazy val tempDirPath = Utils.createTempDir().getAbsolutePath
Review Comment:
Please see the question again - my question was "will this be evaluate once
per executor **_after serde_**?". If then this is a good optimization,
otherwise lazy val does nothing and it just confuses people.
--
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]
