[SPARK-2713] Executors of same application in same host should only download files & jars once

core/src/main/scala/org/apache/spark/SparkContext.scala

@@ -805,11 +805,12 @@ class SparkContext(config: SparkConf) extends Logging {
       case "local"       => "file:" + uri.getPath
       case _             => path
     }
-    addedFiles(key) = System.currentTimeMillis
+    val timestamp = System.currentTimeMillis
+    addedFiles(key) = timestamp
 
     // Fetch the file locally in case a job is executed using DAGScheduler.runLocally().
     Utils.fetchFile(path, new File(SparkFiles.getRootDirectory()), conf, env.securityManager,
-      hadoopConfiguration)
+      hadoopConfiguration, timestamp, useCache = false)
 
     logInfo("Added file " + path + " at " + key + " with timestamp " + addedFiles(key))
     postEnvironmentUpdate()

core/src/main/scala/org/apache/spark/executor/Executor.scala

@@ -322,14 +322,16 @@ private[spark] class Executor(
       // Fetch missing dependencies
       for ((name, timestamp) <- newFiles if currentFiles.getOrElse(name, -1L) < timestamp) {
         logInfo("Fetching " + name + " with timestamp " + timestamp)
-        Utils.fetchFile(name, new File(SparkFiles.getRootDirectory), conf, env.securityManager,
-          hadoopConf)
+        // Fetch file with useCache mode, close cache for local mode.
+        Utils.fetchFile(name, new File(SparkFiles.getRootDirectory), conf,
+          env.securityManager, hadoopConf, timestamp, useCache = !isLocal)
         currentFiles(name) = timestamp
       }
       for ((name, timestamp) <- newJars if currentJars.getOrElse(name, -1L) < timestamp) {
         logInfo("Fetching " + name + " with timestamp " + timestamp)
-        Utils.fetchFile(name, new File(SparkFiles.getRootDirectory), conf, env.securityManager,
-          hadoopConf)
+        // Fetch file with useCache mode, close cache for local mode.
+        Utils.fetchFile(name, new File(SparkFiles.getRootDirectory), conf,
+          env.securityManager, hadoopConf, timestamp, useCache = !isLocal)
         currentJars(name) = timestamp
         // Add it to our class loader
         val localName = name.split("/").last

core/src/main/scala/org/apache/spark/util/Utils.scala

@@ -313,15 +313,84 @@ private[spark] object Utils extends Logging {
   }
 
   /**
-   * Download a file requested by the executor. Supports fetching the file in a variety of ways,
+   * Download a file to target directory. Supports fetching the file in a variety of ways,
    * including HTTP, HDFS and files on a standard filesystem, based on the URL parameter.
    *
+   * If `useCache` is true, first attempts to fetch the file to a local cache that's shared 
+   * across executors running the same application. `useCache` is used mainly for 
+   * the executors, and not in local mode.
+   *
    * Throws SparkException if the target file already exists and has different contents than
    * the requested file.
    */
-  def fetchFile(url: String, targetDir: File, conf: SparkConf, securityMgr: SecurityManager,
-    hadoopConf: Configuration) {
-    val filename = url.split("/").last
+  def fetchFile(
+      url: String,
+      targetDir: File,
+      conf: SparkConf,
+      securityMgr: SecurityManager,
+      hadoopConf: Configuration,
+      timestamp: Long,
+      useCache: Boolean) {
+    val fileName = url.split("/").last
+    val targetFile = new File(targetDir, fileName)
+    if (useCache) {
+      val cachedFileName = s"${url.hashCode}${timestamp}_cache"
+      val lockFileName = s"${url.hashCode}${timestamp}_lock"
+      val localDir = new File(getLocalDir(conf))
+      val lockFile = new File(localDir, lockFileName)
+      val raf = new RandomAccessFile(lockFile, "rw")
+      // Only one executor entry.
+      // The FileLock is only used to control synchronization for executors download file,
+      // it's always safe regardless of lock type (mandatory or advisory).
+      val lock = raf.getChannel().lock()
+      val cachedFile = new File(localDir, cachedFileName)
+      try {
+        if (!cachedFile.exists()) {
+          doFetchFile(url, localDir, cachedFileName, conf, securityMgr, hadoopConf)
+        }
+      } finally {
+        lock.release()
+      }
+      if (targetFile.exists && !Files.equal(cachedFile, targetFile)) {
+        if (conf.getBoolean("spark.files.overwrite", false)) {
+          targetFile.delete()
+          logInfo((s"File $targetFile exists and does not match contents of $url, " +
+            s"replacing it with $url"))
+        } else {
+          throw new SparkException(s"File $targetFile exists and does not match contents of $url")
+        }
+      }
+      Files.copy(cachedFile, targetFile)
+    } else {
+      doFetchFile(url, targetDir, fileName, conf, securityMgr, hadoopConf)
+    }
+
+    // Decompress the file if it's a .tar or .tar.gz
+    if (fileName.endsWith(".tar.gz") || fileName.endsWith(".tgz")) {
+      logInfo("Untarring " + fileName)
+      Utils.execute(Seq("tar", "-xzf", fileName), targetDir)
+    } else if (fileName.endsWith(".tar")) {
+      logInfo("Untarring " + fileName)
+      Utils.execute(Seq("tar", "-xf", fileName), targetDir)
+    }
+    // Make the file executable - That's necessary for scripts
+    FileUtil.chmod(targetFile.getAbsolutePath, "a+x")
+  }
+
+  /**
+   * Download a file to target directory. Supports fetching the file in a variety of ways,
+   * including HTTP, HDFS and files on a standard filesystem, based on the URL parameter.
+   *
+   * Throws SparkException if the target file already exists and has different contents than
+   * the requested file.
+   */
+  private def doFetchFile(
+      url: String,
+      targetDir: File,
+      filename: String,
+      conf: SparkConf,
+      securityMgr: SecurityManager,
+      hadoopConf: Configuration) {
     val tempDir = getLocalDir(conf)
     val tempFile =  File.createTempFile("fetchFileTemp", null, new File(tempDir))
     val targetFile = new File(targetDir, filename)
@@ -409,16 +478,6 @@ private[spark] object Utils extends Logging {
         }
         Files.move(tempFile, targetFile)
     }
-    // Decompress the file if it's a .tar or .tar.gz
-    if (filename.endsWith(".tar.gz") || filename.endsWith(".tgz")) {
-      logInfo("Untarring " + filename)
-      Utils.execute(Seq("tar", "-xzf", filename), targetDir)
-    } else if (filename.endsWith(".tar")) {
-      logInfo("Untarring " + filename)
-      Utils.execute(Seq("tar", "-xf", filename), targetDir)
-    }
-    // Make the file executable - That's necessary for scripts
-    FileUtil.chmod(targetFile.getAbsolutePath, "a+x")
   }
 
   /**