bozhang2820 commented on a change in pull request #33763:
URL: https://github.com/apache/spark/pull/33763#discussion_r698016404
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/FileStreamSource.scala
##########
@@ -105,6 +109,8 @@ class FileStreamSource(
// Visible for testing and debugging in production.
val seenFiles = new SeenFilesMap(maxFileAgeMs, fileNameOnly)
+ var allFilesForTriggerAvailableNow: Seq[(String, Long)] = null
Review comment:
Thanks for catching this! Will change to private.
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/MicroBatchExecution.scala
##########
@@ -382,6 +404,22 @@ class MicroBatchExecution(
// Generate a map from each unique source to the next available offset.
val (nextOffsets, recentOffsets) = uniqueSources.toSeq.map {
+ case (s: AvailableNowDataStreamWrapper, limit) =>
+ val originalSource = s.delegate
+ updateStatusMessage(s"Getting offsets from $s")
+ reportTimeTaken("latestOffset") {
+ val startOffsetOpt = availableOffsets.get(originalSource)
+ val startOffset = s match {
Review comment:
It should not matter since the result will always be the same. However
since we get `startOffsetOpt` by `originalSource`, I think it makes sense to
change to match against that.
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/MicroBatchExecution.scala
##########
@@ -121,18 +122,36 @@ class MicroBatchExecution(
// v2 source
case r: StreamingDataSourceV2Relation => r.stream
}
- uniqueSources = sources.distinct.map {
- case source: SupportsAdmissionControl =>
- val limit = source.getDefaultReadLimit
- if (trigger == OneTimeTrigger && limit != ReadLimit.allAvailable()) {
- logWarning(s"The read limit $limit for $source is ignored when
Trigger.Once() is used.")
- source -> ReadLimit.allAvailable()
- } else {
- source -> limit
- }
- case other =>
- other -> ReadLimit.allAvailable()
- }.toMap
+ uniqueSources = triggerExecutor match {
+ case _: SingleBatchExecutor =>
+ sources.distinct.map {
+ case s: SupportsAdmissionControl =>
+ val limit = s.getDefaultReadLimit
+ if (limit != ReadLimit.allAvailable()) {
+ logWarning(
+ s"The read limit $limit for $s is ignored when Trigger.Once is
used.")
+ }
+ s -> ReadLimit.allAvailable()
+ case s =>
+ s -> ReadLimit.allAvailable()
+ }.toMap
+
+ case _: MultiBatchExecutor =>
+ sources.distinct.map {
+ case s: SupportsTriggerAvailableNow => s
+ case s: Source => new AvailableNowSourceWrapper(s)
+ case s: MicroBatchStream => new
AvailableNowMicroBatchStreamWrapper(s)
+ }.map { s =>
+ s.prepareForTriggerAvailableNow()
+ s -> s.getDefaultReadLimit
Review comment:
I think it's fine. Users might put read limit on some sources but not
the others when using `Trigger.AvailableNow` and we do not need to warn in
those cases.
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/FileStreamSource.scala
##########
@@ -126,7 +132,12 @@ class FileStreamSource(
unreadFiles
} else {
// All the new files found - ignore aged files and files that we have
seen.
Review comment:
Will update the comment here to describe the logic for
Trigger.AvailableNow instead.
--
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]
