Github user tgravescs commented on a diff in the pull request:
https://github.com/apache/spark/pull/14673#discussion_r75128458
--- Diff:
core/src/main/scala/org/apache/spark/ui/jobs/JobProgressListener.scala ---
@@ -137,6 +139,17 @@ class JobProgressListener(conf: SparkConf) extends
SparkListener with Logging {
)
}
+ /** If Tasks is too large, remove and garbage collect old tasks */
+ private def trimTasksIfNecessary(taskData: HashMap[Long, TaskUIData]) =
synchronized {
+ if (taskData.size > retainedTasks) {
+ val toRemove = (taskData.size - retainedTasks)
+ val oldIds =
taskData.map(_._2.taskInfo.taskId).toList.sorted.take(toRemove)
--- End diff --
is there a reason we are using _._2.taskInfo.taskId instead of just the
Long ( _._1), which is the taskId also?
I'm a bit concerned with the sorting here. If someone is using really
large number of tasks , say 1 million I'm afraid this sorting might be
expensive, especially if we are doing it for every single task that comes in
after hitting the limit, we have to keep resorting every time.
---
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]
