Github user squito commented on a diff in the pull request:
https://github.com/apache/spark/pull/17088#discussion_r106684005
--- Diff: core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala
---
@@ -1390,7 +1401,34 @@ class DAGScheduler(
}
} else {
logDebug("Additional executor lost message for " + execId +
- "(epoch " + currentEpoch + ")")
+ "(epoch " + currentEpoch + ")")
+ }
+ }
+
+ private[scheduler] def removeExecutorAndUnregisterOutputOnHost(
+ execId: String,
+ host: String,
+ maybeEpoch: Option[Long] = None) {
+ val currentEpoch = maybeEpoch.getOrElse(mapOutputTracker.getEpoch)
+ if (!failedEpoch.contains(execId) || failedEpoch(execId) <
currentEpoch) {
+ failedEpoch(execId) = currentEpoch
+ logInfo("Executor lost: %s (epoch %d)".format(execId, currentEpoch))
+ blockManagerMaster.removeExecutor(execId)
+ for ((shuffleId, stage) <- shuffleIdToMapStage) {
+ logInfo("Shuffle files lost for host: %s (epoch %d)".format(host,
currentEpoch))
+ stage.removeOutputsOnHost(host)
+ mapOutputTracker.registerMapOutputs(
+ shuffleId,
+ stage.outputLocInMapOutputTrackerFormat(),
+ changeEpoch = true)
+ }
+ if (shuffleIdToMapStage.isEmpty) {
+ mapOutputTracker.incrementEpoch()
+ }
+ clearCacheLocs()
+ } else {
+ logDebug("Additional executor lost message for " + execId +
+ "(epoch " + currentEpoch + ")")
--- End diff --
thanks for trying this refactoring, but I don't like the amount of
repitition between these two helper methods now. Sorry this code was very
confusing even before, and I haven't given constructive suggestions so far ...
what do you think of this
version?https://github.com/squito/spark/commit/3e33d5ea0780e558346e69883d3b4de458d8cdd1
---
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.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
