cxzl25 commented on a change in pull request #24497: [SPARK-27630][CORE] Stage
retry causes totalRunningTasks calculation to be negative
URL: https://github.com/apache/spark/pull/24497#discussion_r288847965
##########
File path: core/src/main/scala/org/apache/spark/ExecutorAllocationManager.scala
##########
@@ -756,10 +757,10 @@ private[spark] class ExecutorAllocationManager(
val taskId = taskEnd.taskInfo.taskId
val taskIndex = taskEnd.taskInfo.index
val stageId = taskEnd.stageId
+ val stageAttemptId = taskEnd.stageAttemptId
+ val stageAttempt = (stageId, stageAttemptId)
Review comment:
In the stage retry process, `StageCompleted` will be sent first and then
`StageSubmitted` will be sent, so `StageCompleted` will be received multiple
times.
https://github.com/apache/spark/pull/24497#discussion_r280147031
Based on this statement, the corresponding stage attempt is cleaned up when
`StageCompleted` is processed, so the `TaskEnd` after `StageCompleted` should
have no relevant structure.
Unless `EventBus` is full, the `StageCompleted` is discarded.
----------------------------------------------------------------
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:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
