Github user skonto commented on a diff in the pull request:
https://github.com/apache/spark/pull/14511#discussion_r73864247
--- Diff:
core/src/main/scala/org/apache/spark/scheduler/cluster/mesos/MesosClusterScheduler.scala
---
@@ -358,14 +358,12 @@ private[spark] class MesosClusterScheduler(
.orElse(desc.command.environment.get("SPARK_EXECUTOR_URI"))
}
- private def getDriverEnvironment(desc: MesosDriverDescription):
Environment = {
- val env = {
- val executorOpts = desc.conf.getAll.map { case (k, v) => s"-D$k=$v"
}.mkString(" ")
- val executorEnv = Map("SPARK_EXECUTOR_OPTS" -> executorOpts)
--- End diff --
From what I understand SPARK_EXECUTOR_OPTS is not used because it is
actually reconstructed at the driver side
[here:](https://github.com/apache/spark/blob/master/core/src/main/scala/org/apache/spark/scheduler/cluster/mesos/MesosCoarseGrainedSchedulerBackend.scala#L180)
from spark.executor.extraJavaOptions.
I dont see that property set at the Dispacther side.
Two things here:
I would prefer the process of reconstructing the env to be transparent to
the user, the same for client and cluster mode, no need for special prefixes.
Secondly, we need to make sure there is continuity with all properties we
pass from process to process.
---
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]
