Skip to content

[SPARK-3596][YARN]Support changing the yarn client monitor interval #5292

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
Sephiroth-Lin wants to merge 1 commit into from
Closed

[SPARK-3596][YARN]Support changing the yarn client monitor interval #5292

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
3 changes: 2 additions & 1 deletion yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnClientSchedulerBackend.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -125,6 +125,7 @@ private[spark] class YarnClientSchedulerBackend(
*/
private def asyncMonitorApplication(): Unit = {
assert(client != null && appId != null, "Application has not been submitted yet!")
val interval = conf.getLong("spark.yarn.client.progress.pollinterval", 1000)
Copy link
Contributor

@sryza sryza Apr 1, 2015

Choose a reason for hiding this comment

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

I would change this to something like spark.yarn.client.amPollInterval.

Copy link
Contributor Author

@Sephiroth-Lin Sephiroth-Lin Apr 1, 2015

Choose a reason for hiding this comment

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

Thank you, but as it is the client to get the application report from the RM, so maybe "spark.yarn.client.progress.pollinterval" is better.

Copy link
Contributor

@sryza sryza Apr 1, 2015

Choose a reason for hiding this comment

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

My main concern is that we don't really have a notion of "progress" in Spark on YARN. In MapReduce, progress traditionally referred to how far along the app was, but Spark doesn't provide a similar notion. YARN itself does have a notion of app progress, which is not related to what's here.

Copy link
Contributor Author

@Sephiroth-Lin Sephiroth-Lin Apr 2, 2015

Choose a reason for hiding this comment

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

Yeah, you are right. In PR #5305 I use the client.monitorApplication, then we can use "spark.yarn.report.interval" to changing the yarn client monitor interval. Thank you.

Copy link
Member

@srowen srowen Apr 2, 2015

Choose a reason for hiding this comment

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

OK, is this superseded by #5305 ? it sounds like you have two different changes related to SPARK-3596 and I'm wondering whether they are mutually exclusive, or one or both is preferred, or what.

Copy link
Contributor Author

@Sephiroth-Lin Sephiroth-Lin Apr 2, 2015

Choose a reason for hiding this comment

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

@srowen Yes, #5305 can solve this issue, may be we can close this PR first.

Copy link
Member

@srowen srowen Apr 2, 2015

Choose a reason for hiding this comment

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

Up to your judgment; you're able to open and close PRs.

Copy link
Contributor Author

@Sephiroth-Lin Sephiroth-Lin Apr 2, 2015

Choose a reason for hiding this comment

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

OK

val t = new Thread {
override def run() {
while (!stopping) {
Expand All @@ -143,7 +144,7 @@ private[spark] class YarnClientSchedulerBackend(
sc.stop()
stopping = true
}
Thread.sleep(1000L)
Thread.sleep(interval)
}
Thread.currentThread().interrupt()
}
Expand Down