Github user kayousterhout commented on a diff in the pull request:
https://github.com/apache/spark/pull/159#discussion_r10641328
--- Diff:
core/src/main/scala/org/apache/spark/scheduler/TaskSetManager.scala ---
@@ -533,8 +575,11 @@ private[spark] class TaskSetManager(
failureReason = "Lost result for TID %s on host %s".format(tid,
info.host)
logWarning(failureReason)
- case _ => {}
+ case _ =>
+ failureReason = "TID %s on host %s failed for unknown
reason".format(tid, info.host)
}
+ // Add to failed for everything else.
+ addToFailedExecutor()
--- End diff --
Yeah the idea here is that if Mesos kills the task (e.g., for fairness
reasons) we should try to re-run it. Job cancellation is handled separately --
and won't result in re-enqueing the task.
---
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.
---
