Github user srowen commented on a diff in the pull request:
https://github.com/apache/spark/pull/14371#discussion_r72389845
--- Diff: core/src/main/scala/org/apache/spark/SparkContext.scala ---
@@ -1410,10 +1410,12 @@ class SparkContext(config: SparkConf) extends
Logging with ExecutorAllocationCli
val scheme = new URI(schemeCorrectedPath).getScheme
if (!Array("http", "https", "ftp").contains(scheme)) {
val fs = hadoopPath.getFileSystem(hadoopConfiguration)
- if (!fs.exists(hadoopPath)) {
- throw new FileNotFoundException(s"Added file $hadoopPath does not
exist.")
+ val isDir = try {
+ fs.getFileStatus(hadoopPath).isDirectory
+ } catch {
+ case f: FileNotFoundException =>
+ throw new FileNotFoundException(s"Added file $hadoopPath does
not exist.").initCause(f)
--- End diff --
I think the theme of most of these changes is making one call to the HDFS
NN rather than two. `getFileStatus` is the one call now. But it will signal
that the file isn't found with an exception. It seems OK to me. Here the
catch-throw is even pretty redundant.
---
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]
