gczsjdy commented on a change in pull request #24462: [SPARK-26268][CORE] Do
not resubmit tasks when executors are lost
URL: https://github.com/apache/spark/pull/24462#discussion_r278817477
##########
File path: core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala
##########
@@ -1791,7 +1791,8 @@ private[spark] class DAGScheduler(
// if the cluster manager explicitly tells us that the entire worker was
lost, then
// we know to unregister shuffle output. (Note that "worker" specifically
refers to the process
// from a Standalone cluster, where the shuffle service lives in the
Worker.)
- val fileLost = workerLost ||
!env.blockManager.externalShuffleServiceEnabled
+ val fileLost = workerLost ||
(!env.blockManager.externalShuffleServiceEnabled &&
+ !sc.conf.get(config.EXTERNAL_SHUFFLE_ENABLED))
Review comment:
With worker lost and EXTERNAL_SHUFFLE_ENABLED = true, fileLost should be
false? Due to the shuffle files metadata doesn't reside in Worker.
I think this should be `!sc.conf.get(config.EXTERNAL_SHUFFLE_ENABLED) &&
(workerLost || (!env.blockManager.externalShuffleServiceEnabled))`
----------------------------------------------------------------
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]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
