[jira] Commented: (HADOOP-432) support undelete, snapshots, or other mechanism to recover lost files

["Doug Cutting (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HADOOP-432?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12468438 ] 

Doug Cutting commented on HADOOP-432:
-------------------------------------

Here's a proposed design:

- change the FS shell's 'rm' command to move files to /Trash/New.  For example, 'rm foo' might move the file /user/cutting/foo to /Trash/New/user/cutting/foo.

- add an FS shell 'expunge' command that removes everything in /Trash.

- add a TrashDumper Runnable that:
   - sleeps a configured interval (1 day)
   - checkpoint: rename /Trash/New to be /Trash/YYMMDDHHMMSS.
   - flush: remove any /Trash/YYMMDDHHMMSS directories older than a configured interval (2 days)

- change NameNode to run the TrashDumper

The two intervals might be defined with a single parameter: the flush interval could always be twice the checkpoint interval.  TrashDumper might have a main() which takes a filesytem named on the command line and runs itself for that FS.

For extra credit, the Dumper might wake more frequently, and check if the filesystem is overly full, and, when it is, empty the trash sooner.  This would require a way to determine the FileSystem's fullness, which is not currently part of the FileSystem API.


> support undelete, snapshots, or other mechanism to recover lost files
> ---------------------------------------------------------------------
>
>                 Key: HADOOP-432
>                 URL: https://issues.apache.org/jira/browse/HADOOP-432
>             Project: Hadoop
>          Issue Type: Improvement
>          Components: dfs
>            Reporter: Yoram Arnon
>         Assigned To: Doug Cutting
>         Attachments: undelete12.patch, undelete16.patch, undelete17.patch
>
>
> currently, once you delete a file it's gone forever.
> most file systems allow some form of recovery of deleted files.
> a simple solution would be an 'undelete' command.
> a more comprehensive solution would include snapshots, manual and automatic, with scheduling options.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.