Skip to content

[SPARK-10081][Core ] Skip re-computing getMissingParentStages #8269

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
viirya wants to merge 1 commit into from
Closed

[SPARK-10081][Core ] Skip re-computing getMissingParentStages #8269

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Skip re-computing getMissingParentStages.
  • Loading branch information
@viirya
viirya committed Aug 18, 2015
commit 8416b7f83c86aa17ed2d7ffdf4a6aea40e48416a
9 changes: 5 additions & 4 deletions core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -732,7 +732,8 @@ class DAGScheduler(
job.jobId, callSite.shortForm, partitions.length))
logInfo("Final stage: " + finalStage + "(" + finalStage.name + ")")
logInfo("Parents of final stage: " + finalStage.parents)
logInfo("Missing parents: " + getMissingParentStages(finalStage))
val missingStages = getMissingParentStages(finalStage)
logInfo("Missing parents: " + missingStages)
Copy link
Member

@srowen srowen Aug 25, 2015

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Super minor but should this use string interpolation? as a general principle for speed and conciseness, though it won't matter here.

val jobSubmissionTime = clock.getTimeMillis()
jobIdToActiveJob(jobId) = job
activeJobs += job
Expand All @@ -741,18 +742,18 @@ class DAGScheduler(
val stageInfos = stageIds.flatMap(id => stageIdToStage.get(id).map(_.latestInfo))
listenerBus.post(
SparkListenerJobStart(job.jobId, jobSubmissionTime, stageInfos, properties))
submitStage(finalStage)
submitStage(finalStage, Some(missingStages))
}
submitWaitingStages()
}

/** Submits stage, but first recursively submits any missing parents. */
private def submitStage(stage: Stage) {
private def submitStage(stage: Stage, missingStages: Option[List[Stage]] = None) {
Copy link
Member

@srowen srowen Aug 25, 2015

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There are a number of recursive calls that then don't get this value then and recompute anyway? To avoid complication, is it simpler to just have all callers provide this info? I don't think it's but one more change.

Copy link
Contributor

@andrewor14 andrewor14 Sep 1, 2015

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

+1, it actually simplifies the code down there because now you don't need the getOrElse in L756 anymore

val jobId = activeJobForStage(stage)
if (jobId.isDefined) {
logDebug("submitStage(" + stage + ")")
if (!waitingStages(stage) && !runningStages(stage) && !failedStages(stage)) {
val missing = getMissingParentStages(stage).sortBy(_.id)
val missing = missingStages.getOrElse(getMissingParentStages(stage)).sortBy(_.id)
logDebug("missing: " + missing)
if (missing.isEmpty) {
logInfo("Submitting " + stage + " (" + stage.rdd + "), which has no missing parents")
Expand Down