[
https://issues.apache.org/jira/browse/HDFS-4839?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13848453#comment-13848453
]
Chris Nauroth commented on HDFS-4839:
-------------------------------------
bq. Irrespective of how this is done, it changes the behaviour of local file://
operations when things fail.
>From a compatibility perspective, I think this means we'd have to be careful
>that the change doesn't start throwing {{IOException}} in cases that used to
>return {{false}}. As you said, just logging the information would be a way to
>handle this. I think this meets our goal to provide enough information for
>diagnosis. (I don't think anyone is proposing that we propagate the native
>{{errno}} all the way out to the caller of {{FileSystem#mkdirs}}.)
bq. In which case, this could maybe be implemented with reflection today
Sorry, I didn't follow this part. Could you explain? I've been thinking of
this change as calling C {{mkdir}}, then getting the {{errno}}, and then
pushing that back up to the Java layer for logging. How can reflection help
with this? Is the {{errno}} hidden somewhere in the JDK classes, and you're
proposing that we peek at it with reflection?
Thanks!
> add NativeIO#mkdirs, that provides an error message on failure
> --------------------------------------------------------------
>
> Key: HDFS-4839
> URL: https://issues.apache.org/jira/browse/HDFS-4839
> Project: Hadoop HDFS
> Issue Type: Improvement
> Affects Versions: 2.1.0-beta
> Reporter: Colin Patrick McCabe
> Priority: Minor
>
> It would be nice to have a variant of mkdirs that provided an error message
> explaining why it failed. This would make it easier to debug certain failing
> unit tests that rely on mkdir / mkdirs-- the ChecksumFilesystem tests, for
> example.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
