Github user dhruve commented on a diff in the pull request:
https://github.com/apache/spark/pull/22288#discussion_r215036162
--- Diff:
core/src/main/scala/org/apache/spark/scheduler/TaskSchedulerImpl.scala ---
@@ -414,9 +425,54 @@ private[spark] class TaskSchedulerImpl(
launchedAnyTask |= launchedTaskAtCurrentMaxLocality
} while (launchedTaskAtCurrentMaxLocality)
}
+
if (!launchedAnyTask) {
- taskSet.abortIfCompletelyBlacklisted(hostToExecutors)
- }
+ taskSet.getCompletelyBlacklistedTaskIfAny(hostToExecutors) match
{
+ case taskIndex: Some[Int] => // Returns the taskIndex which
was unschedulable
+ if (conf.getBoolean("spark.dynamicAllocation.enabled",
false)) {
+ // If the taskSet is unschedulable we kill the existing
blacklisted executor/s and
+ // kick off an abortTimer which after waiting will abort
the taskSet if we were
+ // unable to get new executors and couldn't schedule a
task from the taskSet.
+ // Note: We keep a track of schedulability on a per
taskSet basis rather than on a
+ // per task basis.
+ if (!unschedulableTaskSetToExpiryTime.contains(taskSet)) {
+ hostToExecutors.valuesIterator.foreach(executors =>
executors.foreach({
+ executor =>
+ logDebug("Killing executor because of task
unschedulability: " + executor)
+ blacklistTrackerOpt.foreach(blt =>
blt.killBlacklistedExecutor(executor))
--- End diff --
- To refresh executors, you need to enable
`spark.blacklist.killBlacklistedExecutors`.
- I was thinking about it, killing all the executors is a little too harsh.
Killing only a single executor would help mitigate this, although this would
also lead to failing the running tasks on that executor.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
