dongjoon-hyun commented on a change in pull request #25385: optimization spark
event log that resulting in history server is too …
URL: https://github.com/apache/spark/pull/25385#discussion_r311885665
##########
File path:
core/src/main/scala/org/apache/spark/scheduler/EventLoggingListener.scala
##########
@@ -148,6 +152,59 @@ private[spark] class EventLoggingListener(
}
}
+ /** cut the event log file **/
+ private def logReload(): Unit = {
+ val tlogPath = getLogPath(logBaseDir, appId, appAttemptId,
compressionCodecName, cutType)
+ // ����ȡ��logPathû�䣬�Ͳ������������־�ļ�
+ if (tlogPath == logPath) {
+ return
+ }
+
+ if (!fileSystem.getFileStatus(new Path(logBaseDir)).isDirectory) {
+ throw new IllegalArgumentException(s"Log directory $logBaseDir is not a
directory.")
+ }
+
+ val workingPath = tlogPath + IN_PROGRESS
+ val path = new Path(workingPath)
+ val uri = path.toUri
+ val defaultFs = FileSystem.getDefaultUri(hadoopConf).getScheme
+ val isDefaultLocal = defaultFs == null || defaultFs == "file"
+
+ if (shouldOverwrite && fileSystem.delete(path, true)) {
+ logWarning(s"Event log $path already exists. Overwriting...")
+ }
+
+ /* The Hadoop LocalFileSystem (r1.0.4) has known issues with syncing
(HADOOP-7844).
+ * Therefore, for local files, use FileOutputStream instead. */
+ val dstream =
+ if ((isDefaultLocal && uri.getScheme == null) || uri.getScheme ==
"file") {
+ new FileOutputStream(uri.getPath)
+ } else {
+ hadoopDataStream = Some(fileSystem.create(path))
+ hadoopDataStream.get
+ }
+
+ try {
+ val cstream =
compressionCodec.map(_.compressedOutputStream(dstream)).getOrElse(dstream)
+ val bstream = new BufferedOutputStream(cstream, outputBufferSize)
+
+ EventLoggingListener.initEventLog(bstream, testing, loggedEvents)
+ fileSystem.setPermission(path, LOG_FILE_PERMISSIONS)
+ writer = Some(new PrintWriter(bstream))
+ logInfo("Logging events to %s".format(tlogPath))
+ } catch {
+ case e: Exception =>
+ dstream.close()
+ throw e
+ }
+
+ // �ر��ϵ���
+ stop()
Review comment:
`stop()`?
----------------------------------------------------------------
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]
