HyukjinKwon commented on a change in pull request #23977: [SPARK-26923][SQL][R] Refactor ArrowRRunner and RRunner to share one BaseRRunner URL: https://github.com/apache/spark/pull/23977#discussion_r263657283
########## File path: core/src/main/scala/org/apache/spark/api/r/BaseRRunner.scala ########## @@ -0,0 +1,345 @@ +/* + * 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.api.r + +import java.io._ +import java.net.{InetAddress, ServerSocket} +import java.util.Arrays + +import scala.io.Source +import scala.util.Try + +import org.apache.spark._ +import org.apache.spark.broadcast.Broadcast +import org.apache.spark.internal.Logging +import org.apache.spark.internal.config.BUFFER_SIZE +import org.apache.spark.internal.config.R._ +import org.apache.spark.util.Utils + +/** + * A helper class to run R UDFs in Spark. + */ +private[spark] abstract class BaseRRunner[IN, OUT]( + func: Array[Byte], + deserializer: String, + serializer: String, + packageNames: Array[Byte], + broadcastVars: Array[Broadcast[Object]], + numPartitions: Int, + isDataFrame: Boolean, + colNames: Array[String], + mode: Int) + extends Logging { + protected var bootTime: Double = _ + protected var dataStream: DataInputStream = _ + + def compute( + inputIterator: Iterator[IN], + partitionIndex: Int): Iterator[OUT] = { + // Timing start + bootTime = System.currentTimeMillis / 1000.0 + + // we expect two connections + val serverSocket = new ServerSocket(0, 2, InetAddress.getByName("localhost")) + val listenPort = serverSocket.getLocalPort() + + // The stdout/stderr is shared by multiple tasks, because we use one daemon + // to launch child process as worker. + val errThread = BaseRRunner.createRWorker(listenPort) + + // We use two sockets to separate input and output, then it's easy to manage + // the lifecycle of them to avoid deadlock. + // TODO: optimize it to use one socket + + // the socket used to send out the input of task + serverSocket.setSoTimeout(10000) Review comment: Actually, this one is something already existing before in RRunner side, and there are multiple instances in Python side as well. The default value of `spark.r.backendConnectionTimeout` is 6000. So, if we change this, this could cause potential behaviour change. Let me leave those out in this PR. ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
