[ https://issues.apache.org/jira/browse/HDFS-4619?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13664129#comment-13664129 ]
Steve Loughran commented on HDFS-4619: -------------------------------------- RawLocalFS could look for the dir existing and throw the exception if it exists -and we were happy that this is the expected behaviour. I'm not sure it is. The FS will throw {{FileAlreadyExistsException}} if a parent directory entry exists and is not a directory. This is the same behaviour as -on a first glance- HDFS does. It is not an error to create a filesystem that is already there. # we should add a test, if there is not one already, that creating a dir under a file throws an exception. # {{FileSystemContractBaseTest#testMkdirsFailsForSubdirectoryOfExistingFile()}} verifies that trying to create a subdir of a file fails, while {{testMkdirs()}} verifies that {{mkdirs()}} is idempotent, that it does not fail if called repeatedly. Which makes me think that {{RawLocalFileSystem}} is behaving according to its specification -that just doesn't match the expectations of where it is called > No Exception is thrown if we try to create directory using FileContext > (create) on RawLocalFileSystem. Even if the directory already exists. > --------------------------------------------------------------------------------------------------------------------------------------------- > > Key: HDFS-4619 > URL: https://issues.apache.org/jira/browse/HDFS-4619 > Project: Hadoop HDFS > Issue Type: Bug > Reporter: Omkar Vinit Joshi > Priority: Critical > > as per FileContext create should fail if the file/directory is already > present. However if the underlying filesystem is RawLocalFileSystem then it > doesn't throw any exception. > FileContext.create() > RawLocalFileSystem.mkdirs() -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira