Github user tdas commented on a diff in the pull request:
https://github.com/apache/spark/pull/9181#discussion_r43063455
--- Diff:
streaming/src/main/scala/org/apache/spark/streaming/scheduler/ReceiverSchedulingPolicy.scala
---
@@ -163,40 +169,46 @@ private[streaming] class ReceiverSchedulingPolicy {
receiverId: Int,
preferredLocation: Option[String],
receiverTrackingInfoMap: Map[Int, ReceiverTrackingInfo],
- executors: Seq[String]): Seq[String] = {
+ executors: Seq[ExecutorCacheTaskLocation]): Seq[TaskLocation] = {
if (executors.isEmpty) {
return Seq.empty
}
// Always try to schedule to the preferred locations
- val scheduledExecutors = mutable.Set[String]()
- scheduledExecutors ++= preferredLocation
-
- val executorWeights = receiverTrackingInfoMap.values.flatMap {
receiverTrackingInfo =>
- receiverTrackingInfo.state match {
- case ReceiverState.INACTIVE => Nil
- case ReceiverState.SCHEDULED =>
- val scheduledExecutors =
receiverTrackingInfo.scheduledExecutors.get
- // The probability that a scheduled receiver will run in an
executor is
- // 1.0 / scheduledLocations.size
- scheduledExecutors.map(location => location -> (1.0 /
scheduledExecutors.size))
- case ReceiverState.ACTIVE =>
Seq(receiverTrackingInfo.runningExecutor.get -> 1.0)
- }
- }.groupBy(_._1).mapValues(_.map(_._2).sum) // Sum weights for each
executor
+ val scheduledLocations = mutable.Set[TaskLocation]()
+ // Note: preferredLocation could be `HDFSCacheTaskLocation`, so use
`TaskLocation.apply` to
+ // handle this case
+ scheduledLocations ++= preferredLocation.map(TaskLocation(_))
+
+ val executorWeights: Map[ExecutorCacheTaskLocation, Double] =
--- End diff --
nit: would be good to put a `{...}` around this whole thing.
---
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]
