[
https://issues.apache.org/jira/browse/HADOOP-16860?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Gabor Bota updated HADOOP-16860:
--------------------------------
Description:
When you prune a directory in S3Guard the children entries are not pruned with
the tombstoned parent (directory).
I propose the solution to remove all children entries (the whole hierarchy) to
avoid orphaned entries once the parent (directory) tombstone is removed.
Also the pruning itself in
{{org.apache.hadoop.fs.s3a.s3guard.DynamoDBMetadataStore#prune(org.apache.hadoop.fs.s3a.s3guard.MetadataStore.PruneMode,
long, java.lang.String)}} does not make sense without removing the children
entries - once a directory is pruned then all children will became orpans.
That is particularly dangerous if the same directory is created again in auth
mode - then we have entries that may not exist in S3 but got picked up from the
old orphans.
was:
When you prune a directory in S3Guard the children entries are not pruned with
the tombstoned parent (directory).
I propose the solution to remove all children entries (the whole hierarchy) to
avoid orphaned entries once the parent (directory) tombstone is removed.
> Prune -tombstones to remove children entries
> --------------------------------------------
>
> Key: HADOOP-16860
> URL: https://issues.apache.org/jira/browse/HADOOP-16860
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/s3
> Affects Versions: 3.3.0
> Reporter: Gabor Bota
> Assignee: Gabor Bota
> Priority: Major
>
> When you prune a directory in S3Guard the children entries are not pruned
> with the tombstoned parent (directory).
> I propose the solution to remove all children entries (the whole hierarchy)
> to avoid orphaned entries once the parent (directory) tombstone is removed.
> Also the pruning itself in
> {{org.apache.hadoop.fs.s3a.s3guard.DynamoDBMetadataStore#prune(org.apache.hadoop.fs.s3a.s3guard.MetadataStore.PruneMode,
> long, java.lang.String)}} does not make sense without removing the children
> entries - once a directory is pruned then all children will became orpans.
> That is particularly dangerous if the same directory is created again in auth
> mode - then we have entries that may not exist in S3 but got picked up from
> the old orphans.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
