ChenjunZou commented on a change in pull request #23083: [SPARK-26114][CORE]
ExternalSorter's readingIterator field leak
URL: https://github.com/apache/spark/pull/23083#discussion_r312722564
##########
File path: core/src/main/scala/org/apache/spark/util/CompletionIterator.scala
##########
@@ -25,11 +25,14 @@ private[spark]
abstract class CompletionIterator[ +A, +I <: Iterator[A]](sub: I) extends
Iterator[A] {
private[this] var completed = false
- def next(): A = sub.next()
+ private[this] var iter = sub
+ def next(): A = iter.next()
def hasNext: Boolean = {
- val r = sub.hasNext
+ val r = iter.hasNext
if (!r && !completed) {
completed = true
+ // reassign to release resources of highly resource consuming iterators
early
+ iter = Iterator.empty.asInstanceOf[I]
Review comment:
@szhem
hi,
In your comment:
" reassign the reference of its sub-iterator to the empty iterator ..."
I don't understand why reassigning iter would take effect on original
sub-iterator
can you explain it more?
----------------------------------------------------------------
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]
