Ngone51 commented on a change in pull request #27223:
[SPARK-30511][SPARK-28403][CORE] Don't treat failed/killed speculative tasks as
pending in ExecutorAllocationManager
URL: https://github.com/apache/spark/pull/27223#discussion_r373926875
##########
File path: core/src/main/scala/org/apache/spark/ExecutorAllocationManager.scala
##########
@@ -614,18 +615,30 @@ private[spark] class ExecutorAllocationManager(
stageAttemptToNumRunningTask -= stageAttempt
}
}
- // If the task failed, we expect it to be resubmitted later. To ensure
we have
- // enough resources to run the resubmitted task, we need to mark the
scheduler
- // as backlogged again if it's not already marked as such (SPARK-8366)
- if (taskEnd.reason != Success) {
- if (totalPendingTasks() == 0) {
- allocationManager.onSchedulerBacklogged()
- }
- if (taskEnd.taskInfo.speculative) {
- stageAttemptToSpeculativeTaskIndices.get(stageAttempt).foreach
{_.remove(taskIndex)}
- } else {
- stageAttemptToTaskIndices.get(stageAttempt).foreach
{_.remove(taskIndex)}
- }
+
+ if (taskEnd.taskInfo.speculative) {
+ stageAttemptToSpeculativeTaskIndices.get(stageAttempt).foreach
{_.remove{taskIndex}}
+ stageAttemptToNumSpeculativeTasks(stageAttempt) -= 1
+ }
+
+ taskEnd.reason match {
+ case Success | _: TaskKilled =>
+ case _ =>
+ if (totalPendingTasks() == 0) {
+ // If the task failed (not intentionally killed), we expect it
to be resubmitted
+ // later. To ensure we have enough resources to run the
resubmitted task, we need to
+ // mark the scheduler as backlogged again if it's not already
marked as such
+ // (SPARK-8366)
+ allocationManager.onSchedulerBacklogged()
Review comment:
> If a speculative task fails, while it will not be directly resubmitted, a
new speculative task will be launched in next speculation cycle. So it's OK for
us to mark the scheduler as backlogged in this case.
Hmm...but the new speculative task may not be launched if the normal task
finish. And even if it launched, `ExecutorAllocationManager` could still handle
it by receiving `SparkListenerSpeculativeTaskSubmitted`. Though, calling
`onSchedulerBacklogged` may could reserve executor resource more early to
reduce the delay. Fine!
----------------------------------------------------------------
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:
us...@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
