Github user pwendell commented on a diff in the pull request:
https://github.com/apache/spark/pull/694#discussion_r12649770
--- Diff:
core/src/main/scala/org/apache/spark/scheduler/cluster/CoarseGrainedSchedulerBackend.scala
---
@@ -140,8 +141,21 @@ class CoarseGrainedSchedulerBackend(scheduler:
TaskSchedulerImpl, actorSystem: A
// Launch tasks returned by a set of resource offers
def launchTasks(tasks: Seq[Seq[TaskDescription]]) {
for (task <- tasks.flatten) {
- freeCores(task.executorId) -= scheduler.CPUS_PER_TASK
- executorActor(task.executorId) ! LaunchTask(task)
+ val ser = SparkEnv.get.closureSerializer.newInstance()
+ val serializedTask = ser.serialize(task)
+ if (serializedTask.limit >= akkaFrameSize - 1024) {
+ var msg = "Serialized task %s:%d was %d bytes which " +
+ "exceeds spark.akka.frameSize (%d bytes). " +
+ "Consider using broadcast variables for large values."
--- End diff --
FYI - this warning was originally because of performance considerations
(i.e. large tasks are expensive to ship around). It might make sense to
preserve the warning there anyways, because even if the user increases the akka
frame size, they will still suffer performance issues if they have large tasks.
We could also consolidate them though, just wanted to mention.
---
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.
---
