[
https://issues.apache.org/jira/browse/HADOOP-14484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16036982#comment-16036982
]
Steve Loughran commented on HADOOP-14484:
-----------------------------------------
I think the patch of HADOOP-14457 will address it, because it will be creating
all the parent dirs. Would it make sense to include the test in that patch to
see if that's enough to make it go away?
tests are straightforward
{code}
Path d =path("testroot")
fs.mkdirs(d)
fs.delete(d)
touch(new Path(d, "child))
assertPathExists(d)
{code}
+, for rigorousness, the same where the touched path is 2+ levels down
{code}
> Ensure deleted parent directory tombstones are overwritten when implicitly
> recreated
> ------------------------------------------------------------------------------------
>
> Key: HADOOP-14484
> URL: https://issues.apache.org/jira/browse/HADOOP-14484
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/s3
> Reporter: Sean Mackrory
> Assignee: Sean Mackrory
>
> As discussed on HADOOP-13998, there may be a test missing (and possibly
> broken metadata store implementations) for the case where a directory is
> deleted but is later implicitly recreated by creating a file inside it, where
> the tombstone is not overwritten. In such a case, listing the parent
> directory would result in an error.
> This may also be happening because of HADOOP-14457, but we should add a test
> for this other possibility anyway and fix it if it fails with any
> implementations.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
