Github user tnachen commented on a diff in the pull request:
https://github.com/apache/spark/pull/8610#discussion_r45656480
--- Diff: core/src/main/scala/org/apache/spark/deploy/master/Master.scala
---
@@ -639,10 +640,11 @@ private[deploy] class Master(
// in the queue, then the second app, etc.
for (app <- waitingApps if app.coresLeft > 0) {
val coresPerExecutor: Option[Int] = app.desc.coresPerExecutor
+ val coresPerTask = app.desc.coresPerTask
// Filter out workers that don't have enough resources to launch an
executor
val usableWorkers = workers.toArray.filter(_.state ==
WorkerState.ALIVE)
.filter(worker => worker.memoryFree >=
app.desc.memoryPerExecutorMB &&
- worker.coresFree >= coresPerExecutor.getOrElse(1))
+ worker.coresFree >= math.max(coresPerExecutor.getOrElse(1),
coresPerTask))
--- End diff --
IMO it's not clear what the relationship between coresPerTask and
coresPerExecutor. Looking at the scheduling requirement we just need the
maximum of the set to be able to schedule, and I thought cores per task is
additional cpu resources on top of the executor.
Can we perhaps comment where this is introduced what these two are? Or
point to documentation?
---
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]
