bogao007 commented on code in PR #47133: URL: https://github.com/apache/spark/pull/47133#discussion_r1715706057
########## sql/core/src/main/scala/org/apache/spark/sql/execution/python/TransformWithStateInPandasPythonRunner.scala: ########## @@ -0,0 +1,167 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.execution.python + +import java.io.DataOutputStream +import java.net.ServerSocket + +import scala.concurrent.ExecutionContext + +import org.apache.arrow.vector.VectorSchemaRoot +import org.apache.arrow.vector.ipc.ArrowStreamWriter + +import org.apache.spark.TaskContext +import org.apache.spark.api.python.{BasePythonRunner, ChainedPythonFunctions, PythonRDD} +import org.apache.spark.internal.Logging +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.execution.arrow +import org.apache.spark.sql.execution.arrow.ArrowWriter +import org.apache.spark.sql.execution.metric.SQLMetric +import org.apache.spark.sql.execution.python.TransformWithStateInPandasPythonRunner.{InType, OutType} +import org.apache.spark.sql.execution.streaming.StatefulProcessorHandleImpl +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.types.StructType +import org.apache.spark.sql.vectorized.ColumnarBatch +import org.apache.spark.util.ThreadUtils + +/** + * Python runner implementation for TransformWithStateInPandas. + */ +class TransformWithStateInPandasPythonRunner( + funcs: Seq[(ChainedPythonFunctions, Long)], + evalType: Int, + argOffsets: Array[Array[Int]], + _schema: StructType, + processorHandle: StatefulProcessorHandleImpl, + _timeZoneId: String, + initialWorkerConf: Map[String, String], + override val pythonMetrics: Map[String, SQLMetric], + jobArtifactUUID: Option[String], + groupingKeySchema: StructType) + extends BasePythonRunner[InType, OutType](funcs.map(_._1), evalType, argOffsets, jobArtifactUUID) + with PythonArrowInput[InType] + with BasicPythonArrowOutput + with Logging { + + private val sqlConf = SQLConf.get + private val arrowMaxRecordsPerBatch = sqlConf.arrowMaxRecordsPerBatch + + private var stateSocketSocketPort: Int = 0 Review Comment: good catch! ########## sql/core/src/main/scala/org/apache/spark/sql/execution/python/TransformWithStateInPandasPythonRunner.scala: ########## @@ -0,0 +1,169 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.execution.python + +import java.io.DataOutputStream +import java.net.ServerSocket +import java.util.concurrent.TimeUnit + +import scala.concurrent.ExecutionContext + +import org.apache.arrow.vector.VectorSchemaRoot +import org.apache.arrow.vector.ipc.ArrowStreamWriter + +import org.apache.spark.TaskContext +import org.apache.spark.api.python.{BasePythonRunner, ChainedPythonFunctions, PythonRDD} +import org.apache.spark.internal.Logging +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.execution.arrow +import org.apache.spark.sql.execution.arrow.ArrowWriter +import org.apache.spark.sql.execution.metric.SQLMetric +import org.apache.spark.sql.execution.python.TransformWithStateInPandasPythonRunner.{InType, OutType} +import org.apache.spark.sql.execution.streaming.StatefulProcessorHandleImpl +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.types.StructType +import org.apache.spark.sql.vectorized.ColumnarBatch +import org.apache.spark.util.ThreadUtils + +/** + * Python runner implementation for TransformWithStateInPandas. + */ +class TransformWithStateInPandasPythonRunner( + funcs: Seq[(ChainedPythonFunctions, Long)], + evalType: Int, + argOffsets: Array[Array[Int]], + _schema: StructType, + processorHandle: StatefulProcessorHandleImpl, + _timeZoneId: String, + initialWorkerConf: Map[String, String], + override val pythonMetrics: Map[String, SQLMetric], + jobArtifactUUID: Option[String], + groupingKeySchema: StructType) + extends BasePythonRunner[InType, OutType](funcs.map(_._1), evalType, argOffsets, jobArtifactUUID) + with PythonArrowInput[InType] + with BasicPythonArrowOutput + with Logging { + + private val sqlConf = SQLConf.get + private val arrowMaxRecordsPerBatch = sqlConf.arrowMaxRecordsPerBatch + + private var stateSocketSocketPort: Int = 0 + + override protected val workerConf: Map[String, String] = initialWorkerConf + + (SQLConf.ARROW_EXECUTION_MAX_RECORDS_PER_BATCH.key -> arrowMaxRecordsPerBatch.toString) + + private val arrowWriter: arrow.ArrowWriter = ArrowWriter.create(root) + + // Use lazy val to initialize the fields before these are accessed in [[PythonArrowInput]]'s + // constructor. + override protected lazy val schema: StructType = _schema + override protected lazy val timeZoneId: String = _timeZoneId + override protected val errorOnDuplicatedFieldNames: Boolean = true + override protected val largeVarTypes: Boolean = sqlConf.arrowUseLargeVarTypes + + override protected def handleMetadataBeforeExec(stream: DataOutputStream): Unit = { + super.handleMetadataBeforeExec(stream) + // Also write the port number for state server + stream.writeInt(stateSocketSocketPort) + PythonRDD.writeUTF(groupingKeySchema.json, stream) + } + + override def compute( + inputIterator: Iterator[InType], + partitionIndex: Int, + context: TaskContext): Iterator[OutType] = { + var stateServerSocket: ServerSocket = null + var failed = false + try { + stateServerSocket = new ServerSocket( /* port = */0, + /* backlog = */1) + stateSocketSocketPort = stateServerSocket.getLocalPort + } catch { + case e: Exception => + failed = true + throw e + } finally { + if (failed) { + closeServerSocketChannelSilently(stateServerSocket) + } + } + + val executor = ThreadUtils.newDaemonSingleThreadExecutor("stateConnectionListenerThread") + val executionContext = ExecutionContext.fromExecutor(executor) + + executionContext.execute( + new TransformWithStateInPandasStateServer(stateServerSocket, processorHandle, Review Comment: As discussed offline, we see the performance of the existing implementation on par with ApplyInPandasWithState with some benchmarks. We will move on with the existing approach. -- 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]
