[GitHub] [spark] tgravescs commented on a change in pull request #28848: [SPARK-32003][CORE] When external shuffle service is used, unregister outputs for executor on fetch failure after executor is lost

[GitBox <gi...@apache.org>]

tgravescs commented on a change in pull request #28848:
URL: https://github.com/apache/spark/pull/28848#discussion_r444282018



##########
File path: core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala
##########
@@ -177,6 +177,8 @@ private[spark] class DAGScheduler(
   // TODO: Garbage collect information about failure epochs when we know there are no more
   //       stray messages to detect.
   private val failedEpoch = new HashMap[String, Long]
+  // In addition, track epoch for failed executors that result in lost file output

Review comment:
       I think a better description here would be nice, because without the specific context in removeExecutorAndUnregisterOutputs lost file isn't a usual term we use.  Perhaps at least clarify saying shuffle output file(s) lost.
   it might also be nice to say why we need ti separate from the failedEpoch




----------------------------------------------------------------
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:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org