hvanhovell commented on code in PR #38468:
URL: https://github.com/apache/spark/pull/38468#discussion_r1012428719
##########
connector/connect/src/main/scala/org/apache/spark/sql/connect/service/SparkConnectStreamHandler.scala:
##########
@@ -117,7 +121,38 @@ class SparkConnectStreamHandler(responseObserver:
StreamObserver[Response]) exte
responseObserver.onNext(response.build())
}
- responseObserver.onNext(sendMetricsToResponse(clientId, rows))
+ responseObserver.onNext(sendMetricsToResponse(clientId, dataframe))
+ responseObserver.onCompleted()
+ }
+
+ def processRowsAsArrowBatches(clientId: String, dataframe: DataFrame): Unit
= {
+ import org.apache.arrow.vector.ipc.message.{IpcOption, MessageSerializer}
+
+ val schema = dataframe.schema
+ val maxRecordsPerBatch =
dataframe.sparkSession.sessionState.conf.arrowMaxRecordsPerBatch
+ val timeZoneId =
dataframe.sparkSession.sessionState.conf.sessionLocalTimeZone
+ val arrowSchema = ArrowUtils.toArrowSchema(schema, timeZoneId)
+ val schemaBuffer = MessageSerializer.serializeMetadata(arrowSchema,
IpcOption.DEFAULT)
+
+ val rows =
dataframe.queryExecution.executedPlan.execute().map(_.copy()).collect()
Review Comment:
- Wouldn't it be better to do the heavy lifting on the executors? IMO it is
better to convert to arrow directly. `Dataset.toArrowBatchRdd` seems to be a
good start.
- It would also be nice if we can avoid materializing the entire result on
the driver. We should be able to forward the batches for a partition
immediately when we receive them.
--
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: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
