[
https://issues.apache.org/jira/browse/HDFS-4248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13508033#comment-13508033
]
Kihwal Lee commented on HDFS-4248:
----------------------------------
Since the leases are used to track all {{INodeUnderConstruction}}, two going
out of sync is definitely not an acceptable behavior. I think this patch will
plug the hole in rename, regardless of the cause, i.e. dropping or failing to
update path. So +1 for it.
As for what happens to the leases after renaming, since the clients are not
notifies or there is no "real" inode, the lease becomes useless. I think
namenode should release them, through the existing lease release & block
recovery mechanism. Even if the old client was writing slowly to a block, they
will soon notice it when if the block gets recovered and are unable to update
pipeline. This sounds harsh, but can serve as a notification.
Does this jira supersede HDFS-2875?
> Renames may cause dangling leases
> ---------------------------------
>
> Key: HDFS-4248
> URL: https://issues.apache.org/jira/browse/HDFS-4248
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: namenode
> Affects Versions: 0.23.0, 2.0.0-alpha, 3.0.0
> Reporter: Daryn Sharp
> Assignee: Daryn Sharp
> Priority: Blocker
> Attachments: h4248_20121130_test_rename.patch,
> HDFS-4248.branch-0.23.patch, HDFS-4248.patch
>
>
> Renames of directories may incorrectly re-write the paths in leases under the
> tree being moved.
--
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
