[
https://issues.apache.org/jira/browse/HDFS-3045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13226655#comment-13226655
]
Eli Collins commented on HDFS-3045:
-----------------------------------
Looks good. This doesn't unwind the directory creation in lost+found but on 2nd
thought I think that's better (might as well salvage what blocks we can).
Nits:
- Better if the IO references the file it failed to create, eg
{code}
throw new IOException(errmsg + ": could not create " + target + "/" + chain);
{code}
- Not your change but lets add brackets to the if where the new IOE is thrown
> fsck move should bail on a file if it can't create a block file
> ---------------------------------------------------------------
>
> Key: HDFS-3045
> URL: https://issues.apache.org/jira/browse/HDFS-3045
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: name-node
> Reporter: Eli Collins
> Assignee: Colin Patrick McCabe
> Attachments: HDFS-3045.001.patch
>
>
> NamenodeFsck#lostFoundMove, when it fails to create a file for a block
> continues on to the next block (There's a comment "perhaps we should bail out
> here..." but it doesn't). It should instead fail the move for that particular
> file (unwind the directory creation and not delete the original file).
> Otherwise a transient failure speaking to the NN means this block is lost
> forever.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
