Github user kayousterhout commented on a diff in the pull request:
https://github.com/apache/spark/pull/1309#discussion_r14695572
--- Diff: core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala
---
@@ -809,12 +810,24 @@ class DAGScheduler(
listenerBus.post(SparkListenerStageCompleted(stageToInfos(stage)))
runningStages -= stage
}
+
event.reason match {
case Success =>
logInfo("Completed " + task)
if (event.accumUpdates != null) {
// TODO: fail the stage if the accumulator update fails...
Accumulators.add(event.accumUpdates) // TODO: do this only if
task wasn't resubmitted
+ event.accumUpdates.foreach { case (id, partialValue) =>
--- End diff --
Can you move this code to TaskSetManager? Where possible I think we should
keep task-specific code out of the DAGScheduler, which should be dealing
primarily with stages (I think @rxin may be on a mission to more clearly
define/document this...)
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
