weixiuli commented on code in PR #36162:
URL: https://github.com/apache/spark/pull/36162#discussion_r862442218
##########
core/src/main/scala/org/apache/spark/scheduler/TaskSetManager.scala:
##########
@@ -109,6 +121,10 @@ private[spark] class TaskSetManager(
private val executorDecommissionKillInterval =
conf.get(EXECUTOR_DECOMMISSION_KILL_INTERVAL).map(TimeUnit.SECONDS.toMillis)
+ private[scheduler] val inefficientTask = new InefficientTask(
+ recomputeIntervalMs = speculationTaskStatsCacheInterval,
+ maxTasksToCheck = max(4096, minFinishedForSpeculation))
Review Comment:
It helps to get as many task metrics as possible for a small numTasks, to
make the evaluation more accurate. But now I think using the `maxTasksToCheck
= minFinishedForSpeculation `should be ok.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
