mridulm commented on a change in pull request #34500:
URL: https://github.com/apache/spark/pull/34500#discussion_r791961574
##########
File path: core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala
##########
@@ -164,6 +164,9 @@ private[spark] class DAGScheduler(
// Stages that must be resubmitted due to fetch failures
private[scheduler] val failedStages = new HashSet[Stage]
+ // Stages that have succeeded yet whose child stages aren't done
+ private[scheduler] val succeededStages = new HashSet[Stage]
Review comment:
Note:
We have to be careful with this and ensure things are cleaned up.
Else it will result in holding on to references and prevent cleanup via
gc/cleaner.
One thing I am looking at is what is the interaction between
`succeededStages ` and `submitMapStage` - and what is the possibility of leaks
here. If we end up accumulating stages, we can end up holding references to a
large DAG of state and prevent cleanup .
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
