Github user zsxwing commented on a diff in the pull request:
https://github.com/apache/spark/pull/11140#discussion_r52812799
--- Diff: core/src/main/scala/org/apache/spark/api/python/PythonRDD.scala
---
@@ -892,27 +911,30 @@ private[spark] class PythonBroadcast(@transient var
path: String) extends Serial
private def readObject(in: ObjectInputStream): Unit =
Utils.tryOrIOException {
val dir = new File(Utils.getLocalDir(SparkEnv.get.conf))
val file = File.createTempFile("broadcast", "", dir)
+ phantomReferences += new FilePhantomReference(file, queue)
path = file.getAbsolutePath
val out = new FileOutputStream(file)
Utils.tryWithSafeFinally {
Utils.copyStream(in, out)
} {
- out.close()
+ in.close();
}
}
- /**
- * Delete the file once the object is GCed.
+ /** Create a seperate daemon thread
+ * to remove phantomreferences from queue and invoke cleanup
*/
- override def finalize() {
- if (!path.isEmpty) {
- val file = new File(path)
- if (file.exists()) {
- if (!file.delete()) {
- logWarning(s"Error deleting ${file.getPath}")
- }
+ val referenceThread = new Thread(threadName){
--- End diff --
This will create a thread for each `PythonBroadcast`. We should avoid that.
Maybe use a global thread to do that? However, this will leak a thread even
if it's not used any more (e.g., after stopping SparkContext).
Maybe `finalize` is the best option.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
