Skip to content

[SPARK-21444] Be more defensive when removing broadcasts in MapOutputTracker #18662

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
JoshRosen wants to merge 1 commit into from
Closed

[SPARK-21444] Be more defensive when removing broadcasts in MapOutputTracker #18662

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 6 additions & 1 deletion core/src/main/scala/org/apache/spark/MapOutputTracker.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -194,7 +194,12 @@ private class ShuffleStatus(numPartitions: Int) {
*/
def invalidateSerializedMapOutputStatusCache(): Unit = synchronized {
if (cachedSerializedBroadcast != null) {
cachedSerializedBroadcast.destroy()
// Prevent errors during broadcast cleanup from crashing the DAGScheduler (see SPARK-21444)
Utils.tryLogNonFatalError {
// Use `blocking = false` so that this operation doesn't hang while trying to send cleanup
// RPCs to dead executors.
cachedSerializedBroadcast.destroy(blocking = false)
}
cachedSerializedBroadcast = null
}
cachedSerializedMapStatus = null
Expand Down