vanzin commented on a change in pull request #27085: [SPARK-29779][CORE] Compact old event log files and cleanup - part 1 URL: https://github.com/apache/spark/pull/27085#discussion_r362922114
########## File path: core/src/main/scala/org/apache/spark/deploy/history/EventLogFileCompactor.scala ########## @@ -0,0 +1,212 @@ +/* + * 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.deploy.history + +import java.io.IOException +import java.net.URI +import java.util.ServiceLoader + +import scala.collection.JavaConverters._ + +import org.apache.hadoop.conf.Configuration +import org.apache.hadoop.fs.{FileStatus, FileSystem, Path} + +import org.apache.spark.SparkConf +import org.apache.spark.deploy.history.EventFilter.FilterStatistics +import org.apache.spark.internal.Logging +import org.apache.spark.internal.config.{EVENT_LOG_COMPACTION_SCORE_THRESHOLD, EVENT_LOG_ROLLING_MAX_FILES_TO_RETAIN} +import org.apache.spark.scheduler.ReplayListenerBus +import org.apache.spark.util.Utils + +/** + * This class compacts the old event log files into one compact file, via two phases reading: + * + * 1) Initialize available [[EventFilterBuilder]] instances, and replay the old event log files with + * builders, so that these builders can gather the information to create [[EventFilter]] instances. + * 2) Initialize [[EventFilter]] instances from [[EventFilterBuilder]] instances, and replay the + * old event log files with filters. Rewrite the events to the compact file which the filters decide + * to accept. + * + * This class will calculate the score based on statistic from [[EventFilter]] instances, which + * represents approximate rate of filtered-out events. Score is being calculated via applying + * heuristic; task events tend to take most size in event log. + * + * This class assumes caller will provide the sorted list of files which are sorted by the index of + * event log file, with "at most" one compact file placed first if it exists. Caller should keep in + * mind that this class doesn't care about the semantic of ordering. + * + * When compacting the files, the range of compaction for given file list is determined as: + * (first ~ the file where there're `maxFilesToRetain` files on the right side) + * + * If there're not enough files on the range of compaction, compaction will be skipped. + */ +class EventLogFileCompactor( + sparkConf: SparkConf, + hadoopConf: Configuration, + fs: FileSystem) extends Logging { + private val maxFilesToRetain: Int = sparkConf.get(EVENT_LOG_ROLLING_MAX_FILES_TO_RETAIN) + private val compactionThresholdScore: Double = sparkConf.get(EVENT_LOG_COMPACTION_SCORE_THRESHOLD) + + def compact(eventLogFiles: Seq[FileStatus]): (CompactionResult.Value, Option[Long]) = { Review comment: Should probably explain the return value, at least. The comments about file ordering and the compact file being the first actually make more sense here, too. ---------------------------------------------------------------- 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]
