Github user yaooqinn commented on a diff in the pull request:
https://github.com/apache/spark/pull/19663#discussion_r149272398
--- Diff:
resource-managers/yarn/src/main/scala/org/apache/spark/deploy/yarn/Client.scala
---
@@ -705,6 +705,19 @@ private[spark] class Client(
}
}
+ val confDir =
+ sys.env.getOrElse("SPARK_CONF_DIR", sys.env("SPARK_HOME") +
File.separator + "conf")
+ val dir = new File(confDir)
+ if (dir.isDirectory) {
+ val files = dir.listFiles(new FileFilter {
+ override def accept(pathname: File): Boolean = {
+ pathname.isFile && pathname.getName.endsWith("xml")
+ }
+ })
+ files.foreach { f => hadoopConfFiles(f.getName) = f }
--- End diff --
yes, we follow that order to build classpath. plz check
https://github.com/apache/spark/blob/12ab7f7e89ec9e102859ab3b710815d3058a2e8d/launcher/src/main/java/org/apache/spark/launcher/AbstractCommandBuilder.java#L134
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
