vanzin commented on a change in pull request #27085: [SPARK-29779][CORE] 
Compact old event log files and cleanup
URL: https://github.com/apache/spark/pull/27085#discussion_r364477952
 
 

 ##########
 File path: 
core/src/main/scala/org/apache/spark/deploy/history/EventLogFileCompactor.scala
 ##########
 @@ -0,0 +1,224 @@
+/*
+ * 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.
+ */
+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)
+
+  /**
+   * Compacts the old event log files into one compact file, and clean old 
event log files being
+   * compacted away.
+   *
+   * This method 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.
+   *
+   * 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)
+   *
+   * This method skips compaction for some circumstances described below:
+   * - not enough files on the range of compaction
+   * - score is lower than the threshold of compaction (meaning compaction 
won't help much)
+   *
+   * If this method returns the compaction result as SUCCESS, caller needs to 
re-read the list
+   * of event log files, as new compact file is available as well as old event 
log files are
+   * removed.
+   */
+  def compact(eventLogFiles: Seq[FileStatus]): CompactionResult = {
+    assertPrecondition(eventLogFiles)
+
+    if (eventLogFiles.length < maxFilesToRetain) {
+      return CompactionResult(CompactionResultCode.NOT_ENOUGH_FILES, None)
+    }
+
+    val filesToCompact = findFilesToCompact(eventLogFiles)
+    if (filesToCompact.isEmpty) {
+      CompactionResult(CompactionResultCode.NOT_ENOUGH_FILES, None)
+    } else {
+      val builders = initializeBuilders(fs, filesToCompact.map(_.getPath))
+
+      val filters = builders.map(_.createFilter())
+      val minScore = filters.flatMap(_.statistics()).map(calculateScore).min
+
+      if (minScore < compactionThresholdScore) {
+        CompactionResult(CompactionResultCode.LOW_SCORE_FOR_COMPACTION, None)
+      } else {
+        rewrite(filters, filesToCompact)
+        cleanupCompactedFiles(filesToCompact)
+        CompactionResult(CompactionResultCode.SUCCESS, Some(
+          
RollingEventLogFilesWriter.getEventLogFileIndex(filesToCompact.last.getPath.getName)))
+      }
+    }
+  }
+
+  private def assertPrecondition(eventLogFiles: Seq[FileStatus]): Unit = {
+    val idxCompactedFiles = eventLogFiles.zipWithIndex.filter { case (file, _) 
=>
+      EventLogFileWriter.isCompacted(file.getPath)
+    }
+    require(idxCompactedFiles.size < 2 && 
idxCompactedFiles.headOption.forall(_._2 == 0),
+      "The number of compact files should be at most 1, and should be placed 
first if exists.")
+  }
+
+  /**
+   * Loads all available EventFilterBuilders in classloader via ServiceLoader, 
and initializes
+   * them via replaying events in given files.
+   */
+  private def initializeBuilders(fs: FileSystem, files: Seq[Path]): 
Seq[EventFilterBuilder] = {
+    val bus = new ReplayListenerBus()
+
+    val builders = ServiceLoader.load(classOf[EventFilterBuilder],
+      Utils.getContextOrSparkClassLoader).asScala.toSeq
+    builders.foreach(bus.addListener)
+
+    files.foreach { log =>
+      Utils.tryWithResource(EventLogFileReader.openEventLog(log, fs)) { in =>
+        bus.replay(in, log.getName)
+      }
+    }
+
+    builders
+  }
+
+  private def calculateScore(stats: FilterStatistics): Double = {
+    // For now it's simply measuring how many task events will be filtered out 
(rejected)
+    // but it can be sophisticated later once we get more heuristic 
information and found
+    // the case where this simple calculation doesn't work.
+    (stats.totalTasks - stats.liveTasks) * 1.0 / stats.totalTasks
+  }
+
+  /**
+   * This class rewrites the event log files into one compact file: the 
compact file will only
 
 Review comment:
   s/class/method

----------------------------------------------------------------
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]

Reply via email to