Github user pwendell commented on a diff in the pull request:
https://github.com/apache/spark/pull/433#discussion_r11788225
--- Diff:
yarn/common/src/main/scala/org/apache/spark/deploy/yarn/ClientBase.scala ---
@@ -361,55 +361,48 @@ object ClientBase {
val LOG4J_CONF_ENV_KEY: String = "SPARK_LOG4J_CONF"
val LOCAL_SCHEME = "local"
- // Based on code from org.apache.hadoop.mapreduce.v2.util.MRApps
- def populateHadoopClasspath(conf: Configuration, env: HashMap[String,
String]) {
- val classpathEntries = Option(conf.getStrings(
- YarnConfiguration.YARN_APPLICATION_CLASSPATH)).getOrElse(
- getDefaultYarnApplicationClasspath())
- for (c <- classpathEntries) {
- YarnSparkHadoopUtil.addToEnvironment(env,
Environment.CLASSPATH.name, c.trim,
+ def populateHadoopClasspath(conf: Configuration, env: HashMap[String,
String]) = {
+ val classPathElementsToAdd = getYarnAppClasspath(conf) ++
getMRAppClasspath(conf)
+ for (c <- classPathElementsToAdd.flatten) {
+ YarnSparkHadoopUtil.addToEnvironment(
+ env,
+ Environment.CLASSPATH.name,
+ c.trim,
File.pathSeparator)
}
+ classPathElementsToAdd
+ }
- val mrClasspathEntries = Option(conf.getStrings(
- "mapreduce.application.classpath")).getOrElse(
- getDefaultMRApplicationClasspath())
- if (mrClasspathEntries != null) {
- for (c <- mrClasspathEntries) {
- YarnSparkHadoopUtil.addToEnvironment(env,
Environment.CLASSPATH.name, c.trim,
- File.pathSeparator)
- }
+ private def getYarnAppClasspath(conf: Configuration):
Option[Array[String]] =
+ Option(conf.getStrings(YarnConfiguration.YARN_APPLICATION_CLASSPATH))
match {
+ case s: Some[Array[String]] => s
+ case None => getDefaultYarnApplicationClasspath
}
- }
- def getDefaultYarnApplicationClasspath(): Array[String] = {
- try {
- val field =
classOf[MRJobConfig].getField("DEFAULT_YARN_APPLICATION_CLASSPATH")
- field.get(null).asInstanceOf[Array[String]]
- } catch {
- case err: NoSuchFieldError => null
- case err: NoSuchFieldException => null
+ private def getMRAppClasspath(conf: Configuration):
Option[Array[String]] =
+ Option(conf.getStrings("mapreduce.application.classpath")) match {
+ case s: Some[Array[String]] => s
+ case None => getDefaultMRApplicationClasspath
}
- }
+
+ def getDefaultYarnApplicationClasspath: Option[Array[String]] =
Try[Array[String]] {
+ val field =
classOf[YarnConfiguration].getField("DEFAULT_YARN_APPLICATION_CLASSPATH")
+ field.get(null).asInstanceOf[Array[String]]
+ }.toOption
/**
* In Hadoop 0.23, the MR application classpath comes with the YARN
application
* classpath. In Hadoop 2.0, it's an array of Strings, and in 2.2+ it's
a String.
* So we need to use reflection to retrieve it.
*/
- def getDefaultMRApplicationClasspath(): Array[String] = {
- try {
- val field =
classOf[MRJobConfig].getField("DEFAULT_MAPREDUCE_APPLICATION_CLASSPATH")
- if (field.getType == classOf[String]) {
- StringUtils.getStrings(field.get(null).asInstanceOf[String])
- } else {
- field.get(null).asInstanceOf[Array[String]]
- }
- } catch {
- case err: NoSuchFieldError => null
- case err: NoSuchFieldException => null
+ def getDefaultMRApplicationClasspath: Option[Array[String]] =
Try[Array[String]] {
+ val field =
classOf[MRJobConfig].getField("DEFAULT_MAPREDUCE_APPLICATION_CLASSPATH")
+ if (field.getType == classOf[String]) {
+ StringUtils.getStrings(field.get(null).asInstanceOf[String]).toArray
+ } else {
+ field.get(null).asInstanceOf[Array[String]]
}
- }
+ }.toOption
--- End diff --
There are a few issues here:
1. Can we change this API? Yes, we can change it - the whole deploy package
is hidden from our docs and not supposed to be used by users. In a separate PR
we should make this private.
2. Should we swallow all exceptions here? No, we should swallow only the
ones we expect which is the missing field. Otherwise we should just throw it up
the chain.
3. Should we use scala.util.Try instead of a try catch block? In the Spark
code we use try {} catch {} blocks typically unless it is a very simple case
where we want to swallow every possible exception.
---
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.
---
