agrawaldevesh commented on a change in pull request #29014:
URL: https://github.com/apache/spark/pull/29014#discussion_r458465566
##########
File path:
core/src/main/scala/org/apache/spark/scheduler/TaskSchedulerImpl.scala
##########
@@ -912,13 +914,39 @@ private[spark] class TaskSchedulerImpl(
}
}
- override def executorDecommission(executorId: String): Unit = {
+ override def executorDecommission(
+ executorId: String, decommissionInfo: ExecutorDecommissionInfo): Unit = {
+ synchronized {
Review comment:
Ah, I see your point -- its because of the DecommissionExecutor messages
that the executor sends to the driver when work is assigned to it. And there,
it can reply back with a hostDecommissioned = false, because the executor
itself does not know if the host is going away or not (only the cluster manager
knows).
I like your suggestion of doing this "merging" semantics, it is
easier/less-work than plumbing host Lost to the executor :-)
----------------------------------------------------------------
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]
