Github user ankuriitg commented on a diff in the pull request: https://github.com/apache/spark/pull/22504#discussion_r226418231 --- Diff: core/src/main/scala/org/apache/spark/util/logging/DriverLogger.scala --- @@ -0,0 +1,196 @@ +/* + * 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.util.logging + +import java.io._ +import java.util.concurrent.TimeUnit + +import scala.util.{Failure, Try} + +import org.apache.commons.io.FileUtils +import org.apache.hadoop.conf.Configuration +import org.apache.hadoop.fs.{FileSystem, FSDataOutputStream, Path} +import org.apache.hadoop.fs.permission.FsPermission +import org.apache.log4j.{FileAppender => Log4jFileAppender, _} + +import org.apache.spark.SparkConf +import org.apache.spark.internal.Logging +import org.apache.spark.internal.config._ +import org.apache.spark.network.util.JavaUtils +import org.apache.spark.util.{ThreadUtils, Utils} + +private[spark] class DriverLogger(conf: SparkConf) extends Logging { + + private val UPLOAD_CHUNK_SIZE = 1024 * 1024 + private val UPLOAD_INTERVAL_IN_SECS = 5 + private val DRIVER_LOG_FILE = "driver.log" + private val LOG_FILE_PERMISSIONS = new FsPermission(Integer.parseInt("770", 8).toShort) + + private var localLogFile: String = FileUtils.getFile( + Utils.getLocalDir(conf), "driver_logs", DRIVER_LOG_FILE).getAbsolutePath() + private var writer: Option[DfsAsyncWriter] = None + + addLogAppender() + + private def addLogAppender(): Unit = { + val appenders = LogManager.getRootLogger().getAllAppenders() + val layout = if (conf.contains(DRIVER_LOG_LAYOUT)) { + new PatternLayout(conf.get(DRIVER_LOG_LAYOUT)) + } else if (appenders.hasMoreElements()) { + appenders.nextElement().asInstanceOf[Appender].getLayout() + } else { + new PatternLayout(DRIVER_LOG_LAYOUT.defaultValueString) + } + val fa = new Log4jFileAppender(layout, localLogFile) + fa.setName(DriverLogger.APPENDER_NAME) + LogManager.getRootLogger().addAppender(fa) + logInfo(s"Added a local log appender at: ${localLogFile}") + } + + def startSync(hadoopConf: Configuration): Unit = { + try { + // Setup a writer which moves the local file to hdfs continuously + val appId = Utils.sanitizeDirName(conf.getAppId) + writer = Some(new DfsAsyncWriter(appId, hadoopConf)) + } catch { + case e: Exception => + logError(s"Could not sync driver logs to spark dfs", e) + } + } + + def stop(): Unit = { + try { + val fa = LogManager.getRootLogger.getAppender(DriverLogger.APPENDER_NAME) + LogManager.getRootLogger().removeAppender(DriverLogger.APPENDER_NAME) + Utils.tryLogNonFatalError(fa.close()) + writer.foreach(_.closeWriter()) + } catch { + case e: Exception => + logError(s"Error in persisting driver logs", e) + } finally { + Utils.tryLogNonFatalError(JavaUtils.deleteRecursively( + FileUtils.getFile(localLogFile).getParentFile())) + } + } + + // Visible for testing + private[spark] class DfsAsyncWriter(appId: String, hadoopConf: Configuration) extends Runnable + with Logging { + + private var streamClosed = false + private val fileSystem: FileSystem = FileSystem.get(hadoopConf) + private val dfsLogFile: String = { + val rootDir = conf.get(DRIVER_LOG_DFS_DIR).get + if (!fileSystem.exists(new Path(rootDir))) { + throw new RuntimeException(s"${rootDir} does not exist." + + s" Please create this dir in order to sync driver logs") + } + FileUtils.getFile(rootDir, appId, DRIVER_LOG_FILE).getAbsolutePath() + } + private var inStream: InputStream = null + private var outputStream: FSDataOutputStream = null + try { + inStream = new BufferedInputStream(new FileInputStream(localLogFile)) + outputStream = fileSystem.create(new Path(dfsLogFile), true) + fileSystem.setPermission(new Path(dfsLogFile), LOG_FILE_PERMISSIONS) + } catch { + case e: Exception => + if (inStream != null) { + Utils.tryLogNonFatalError(inStream.close()) + } + if (outputStream != null) { + Utils.tryLogNonFatalError(outputStream.close()) + } + throw e + } + private val tmpBuffer = new Array[Byte](UPLOAD_CHUNK_SIZE) + private val threadpool = ThreadUtils.newDaemonSingleThreadScheduledExecutor("dfsSyncThread") + threadpool.scheduleWithFixedDelay(this, UPLOAD_INTERVAL_IN_SECS, UPLOAD_INTERVAL_IN_SECS, + TimeUnit.SECONDS) + logInfo(s"The local driver log file is being synced to spark dfs at: ${dfsLogFile}") + + def run(): Unit = { + if (streamClosed) { + return + } + try { + var remaining = inStream.available() + while (remaining > 0) { + val read = inStream.read(tmpBuffer, 0, math.min(remaining, UPLOAD_CHUNK_SIZE)) + outputStream.write(tmpBuffer, 0, read) + remaining -= read + } + outputStream.hflush() + } catch { + case e: Exception => logError("Failed to write to spark dfs", e) --- End diff -- There can be cases when there are consistent error - such that someone deleted the file from filesystem, changed the permission, etc. But mostly these will be one off, so I am leaving the exception handling as it is for now. We can certainly change this later on if we discover otherwise.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org