[
https://issues.apache.org/jira/browse/HADOOP-15940?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16693074#comment-16693074
]
Steve Loughran commented on HADOOP-15940:
-----------------------------------------
If there's two possible "valid" actions for the rename to self, then really
there should be another contract option which declares which codepath an FS
takes, and the standard test changed to reading it and requiring the specific
outcome
* makes the behaviour chosen explicit in the XML
* makes it easy for other stores to adopt the same behaviour
* by adding an assert that the new outcome must occur, catches regressions.
Interesting that the handling of a missing source didn't raise an exception
-I'd have expected that.
FWIW, part of the history of those contract tests is that the s3 and swift
connectors didn't check for a rename under self (who would do that?) and after
copying the files to the destination, would then recursively delete the entire
source tree, so losing the renamed data. Luckily, nobody ever seems to hae
tried to do this in the wild, which shows how our filesystem mental model
prevents us wandering into dangerous places
> ABFS: For HNS account, avoid unnecessary get call when doing Rename
> -------------------------------------------------------------------
>
> Key: HADOOP-15940
> URL: https://issues.apache.org/jira/browse/HADOOP-15940
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/azure
> Affects Versions: 3.2.0
> Reporter: Da Zhou
> Assignee: Da Zhou
> Priority: Major
> Attachments: HADOOP-15940-001.patch
>
>
> When rename, there is always a GET dst file status call, this is not
> necessary.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
