dongjoon-hyun commented on a change in pull request #31373:
URL: https://github.com/apache/spark/pull/31373#discussion_r566277434
##########
File path: core/src/main/scala/org/apache/spark/HeartbeatReceiver.scala
##########
@@ -128,14 +128,16 @@ private[spark] class HeartbeatReceiver(sc: SparkContext,
clock: Clock)
// Messages received from executors
case heartbeat @ Heartbeat(executorId, accumUpdates, blockManagerId,
executorUpdates) =>
+ var reregisterBlockManager = !sc.isStopped
Review comment:
Thank. you for review, @Ngone51 . That's true. In the production
environment, I hit those intermediate status. And, this will help us simplify
the situation.
1. The case you mentioned, `Send HeartbeatResponse and sc.stop invoked`, is
a normal situation. The users don't complain about this.
2. The case in this PR, `sc.stop invoked and Spark works inefficiently by
sending HeartbeatResponse(true)` is a problem. The users complain about this.
For the following, yes. We are After the apps `sc.stop` takes a longer time
than we expect.
> Does the current behavior cause any real issue?
----------------------------------------------------------------
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
