[
https://issues.apache.org/jira/browse/HADOOP-16429?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Steve Loughran resolved HADOOP-16429.
-------------------------------------
Resolution: Done
seems to be done; as well as the deleteSubtree doing leaf nodes first,
HADOOP-16430 is moving to a list of all children and delete, which is now
linear rather than treewalk: leaf nodes are all you get
> DynamoDBMetaStore deleteSubtree to delete leaf nodes first
> ----------------------------------------------------------
>
> Key: HADOOP-16429
> URL: https://issues.apache.org/jira/browse/HADOOP-16429
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/s3
> Affects Versions: 3.3.0
> Reporter: Steve Loughran
> Priority: Major
>
> in {{deleteSubtree(path)}}, the DynamoDB metastore walks down the tree,
> returning elements to delete. But it will delete parent entries before
> children, so if an operation fails partway through, there will be orphans
> Better: DescendantsIterator to return all the leaf nodes before their parents
> so the deletion is done bottom up
> Also: push the deletions off into their own async queue/pool so that they
> don't become the bottleneck on the process
--
This message was sent by Atlassian Jira
(v8.3.2#803003)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
