vanzin commented on a change in pull request #26218: [SPARK-29562][sql] Speed
up and slim down metric aggregation in SQL listener.
URL: https://github.com/apache/spark/pull/26218#discussion_r338848855
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/ui/SQLAppStatusListener.scala
##########
@@ -140,7 +146,16 @@ class SQLAppStatusListener(
override def onExecutorMetricsUpdate(event:
SparkListenerExecutorMetricsUpdate): Unit = {
event.accumUpdates.foreach { case (taskId, stageId, attemptId,
accumUpdates) =>
- updateStageMetrics(stageId, attemptId, taskId, accumUpdates, false)
+ updateStageMetrics(stageId, attemptId, taskId,
SQLAppStatusListener.UNKNOWN_PARTITION,
+ accumUpdates, false)
+ }
+ }
+
+ override def onTaskStart(event: SparkListenerTaskStart): Unit = {
+ Option(stageMetrics.get(event.stageId)).foreach { stage =>
+ if (stage.attemptId == event.stageAttemptId) {
+ stage.registerTask(event.taskInfo.taskId, event.taskInfo.index)
Review comment:
Man, if that's going to make you drop this discussion, I'll call them "pink
unicorns" if that's what you want. The variable name won't help you understand
the relationship between partitions and indices and why it's important to track
this.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
