hvanhovell commented on code in PR #38468:
URL: https://github.com/apache/spark/pull/38468#discussion_r1014279677
##########
connector/connect/src/main/scala/org/apache/spark/sql/connect/service/SparkConnectStreamHandler.scala:
##########
@@ -117,7 +126,70 @@ 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
= {
+ val spark = dataframe.sparkSession
+ val schema = dataframe.schema
+ // TODO: control the batch size instead of max records
+ val maxRecordsPerBatch = spark.sessionState.conf.arrowMaxRecordsPerBatch
+ val timeZoneId = spark.sessionState.conf.sessionLocalTimeZone
+
+ val rows = dataframe.queryExecution.executedPlan.execute()
+ var numBatches = 0L
+
+ if (rows.getNumPartitions > 0) {
+ val batches = rows.mapPartitionsInternal { iter =>
+ ArrowConverters
+ .toArrowBatchIterator(iter, schema, maxRecordsPerBatch, timeZoneId)
+ }
+
+ val obj = new Object
+
+ val processPartition = (iter: Iterator[(Array[Byte], Long, Long)]) =>
iter.toArray
+
+ val resultHandler = (partitionId: Int, taskResult: Array[(Array[Byte],
Long, Long)]) =>
+ obj.synchronized {
+ var batchId = partitionId.toLong << 33
+ taskResult.foreach { case (bytes, count, size) =>
+ val response = proto.Response.newBuilder().setClientId(clientId)
+ val batch = proto.Response.ArrowBatch
+ .newBuilder()
+ .setBatchId(batchId)
+ .setRowCount(count)
+ .setUncompressedBytes(size)
+ .setCompressedBytes(bytes.length)
+ .setData(ByteString.copyFrom(bytes))
+ .build()
+ response.setArrowBatch(batch)
+ responseObserver.onNext(response.build())
Review Comment:
This callback is currently executed by the DAGScheduler thread, and if this
expensive no job/stage can be scheduler during this call. We really should move
this to a separate thread.
--
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]
