Github user andrewor14 commented on a diff in the pull request:
https://github.com/apache/spark/pull/3783#discussion_r22744863
--- Diff:
core/src/main/scala/org/apache/spark/ExecutorAllocationManager.scala ---
@@ -315,7 +314,7 @@ private[spark] class ExecutorAllocationManager(
private def onExecutorAdded(executorId: String): Unit = synchronized {
if (!executorIds.contains(executorId)) {
executorIds.add(executorId)
- executorIds.foreach(onExecutorIdle)
+ onExecutorIdle(executorId)
--- End diff --
Wait, but if we simply don't remove `executorId` from `removeExecutor` when
we've reached the minimum, we will try to remove it again as soon as the next
time `schedule` is called. This means we'll be constantly calling
`removeExecutor` on that executor and failing. I think we should mark it
specially as "should be removed as soon as new executors join" instead. (and we
can do this with the -1 thing I mentioned earlier)
---
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.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
