Github user GrahamDennis commented on a diff in the pull request:
https://github.com/apache/spark/pull/1890#discussion_r16155353
--- Diff:
core/src/main/scala/org/apache/spark/deploy/worker/ExecutorRunner.scala ---
@@ -114,12 +115,12 @@ private[spark] class ExecutorRunner(
case other => other
}
- def getCommandSeq = {
+ def getCommandSeq(userJarClassPathEntries : Seq[String]) = {
val command = Command(
appDesc.command.mainClass,
appDesc.command.arguments.map(substituteVariables) ++ Seq(appId),
appDesc.command.environment,
- appDesc.command.classPathEntries,
+ appDesc.command.classPathEntries ++ userJarClassPathEntries,
--- End diff --
Good point. Also, even with the above code, `userJarClassPathEntries` will
be before the output of `bin/compute-classpath.sh` (see buildJavaOpts in
CommandUtils.scala). At present, appDesc.command.classPathEntries really comes
from spark.executor.extraClassPath.
What should be the order for the two values of
spark.files.userClassPathFirst?
If false, should it be `[<results of bin/compute-classpath.sh>,
spark.executor.extraClassPath, userJars]`? (Note that this is different to the
present behaviour where compute-classpath.sh comes after
spark.executor.extraClassPath). An alternative here would be
[spark.executor.extraClassPath, <results of bin/compute-classpath.sh>,
userJars]?
If true, should it be [userJars, spark.extraClassPath, <results of
bin/compute-classpath.sh>]?
---
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]
