Github user tdas commented on a diff in the pull request: https://github.com/apache/spark/pull/15307#discussion_r82855608 --- Diff: sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/StreamExecution.scala --- @@ -516,12 +563,127 @@ class StreamExecution( """.stripMargin } - private def toInfo: StreamingQueryInfo = { - new StreamingQueryInfo( - this.name, - this.id, - this.sourceStatuses, - this.sinkStatus) + /** + * Report row metrics of the executed trigger + * @param triggerExecutionPlan Execution plan of the trigger + * @param triggerLogicalPlan Logical plan of the trigger, generated from the query logical plan + * @param sourceToDF Source to DataFrame returned by the source.getBatch in this trigger + */ + private def reportNumRows( + triggerExecutionPlan: SparkPlan, + triggerLogicalPlan: LogicalPlan, + sourceToDF: Map[Source, DataFrame]): Unit = { + // We want to associate execution plan leaves to sources that generate them, so that we match + // the their metrics (e.g. numOutputRows) to the sources. To do this we do the following. + // Consider the translation from the streaming logical plan to the final executed plan. + // + // streaming logical plan (with sources) <==> trigger's logical plan <==> executed plan + // + // 1. We keep track of streaming sources associated with each leaf in the trigger's logical plan + // - Each logical plan leaf will be associated with a single streaming source. + // - There can be multiple logical plan leaves associated a streaming source. + // - There can be leaves not associated with any streaming source, because they were + // generated from a batch source (e.g. stream-batch joins) + // + // 2. Assuming that the executed plan has same number of leaves in the same order as that of + // the trigger logical plan, we associate executed plan leaves with corresponding + // streaming sources. + // + // 3. For each source, we sum the metrics of the associated execution plan leaves. + // + val logicalPlanLeafToSource = sourceToDF.flatMap { case (source, df) => + df.logicalPlan.collectLeaves().map { leaf => leaf -> source } + } + val allLogicalPlanLeaves = triggerLogicalPlan.collectLeaves() // includes non-streaming sources + val allExecPlanLeaves = triggerExecutionPlan.collectLeaves() + val sourceToNumInputRows: Map[Source, Long] = + if (allLogicalPlanLeaves.size == allExecPlanLeaves.size) { + val execLeafToSource = allLogicalPlanLeaves.zip(allExecPlanLeaves).flatMap { --- End diff -- See the condition in the previous line :)
--- 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 infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org