Ngone51 commented on a change in pull request #23871:
[SPARK-23433][SPARK-25250] [CORE] Later created TaskSet should learn about the
finished partitions
URL: https://github.com/apache/spark/pull/23871#discussion_r261486872
##########
File path: core/src/main/scala/org/apache/spark/scheduler/TaskSetManager.scala
##########
@@ -530,7 +530,7 @@ private[spark] class TaskSetManager(
private def maybeFinishTaskSet() {
if (isZombie && runningTasks == 0) {
- sched.taskSetFinished(this)
+ sched.taskSetFinished(this, tasksSuccessful == numTasks)
Review comment:
Clean up `stageIdToFinishedPartitions` in `DAGSchedule.markStageAsFinished`
would be definitely work, though it would add an new place to race for
`TaskSchedulerImpl` lock.
Another way is to distinguish TSM active or zombie in
`markPartitionCompleted()` where an active TSM won't be zombie at the time and
clean up `stageIdToFinishedPartitions` right there. Which one do you prefer ?
@cloud-fan
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on 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]
