Github user pwendell commented on a diff in the pull request:
https://github.com/apache/spark/pull/186#discussion_r11964660
--- Diff: core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala
---
@@ -322,6 +289,14 @@ class DAGScheduler(
missing.toList
}
+ private def startDAGSchedulerActors() {
+ dagSchedulerActorSupervisor = env.actorSystem.actorOf(Props(
+ new DAGSchedulerActorSupervisor(this)))
+ while (dagSchedulerActorSupervisor == null || eventProcessActor ==
null) {
--- End diff --
how can `dagSchedulerActorSupervisor` possibly be null here? Busy-waiting
here is also not ideal, it would be better to use proper synchronization.
I think the issue is that actors in general are supposed to be
self-contained, but here inside of an actor we're setting global state. I'm not
following this patch closely enough to understand why it's necessary, but it
does seem a bit ugly.
---
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.
---
