[
https://issues.apache.org/jira/browse/HADOOP-4491?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12734079#action_12734079
]
Vinod K V commented on HADOOP-4491:
-----------------------------------
Broadly, there are two directory strucutres - system and users
- system directory will be owned by mapreduce, thereby protecting the contents.
- users is 755, owned by mapreduce
- users/$jobid is clearly 700 and owned by the user.
- system/$jobid/outputs can be directly $ttroot/ as was discussed offline. But
I've left it inside system/$jobid as the $jobid directory seemed reduntant to
me. In any case, the outputs once moved need to owned by the TT.
- all of the files localized by the TT are written into system/$jobid
- After job localization is done, all files under system/$jobid/userfiles are
moved to users/$jobid to be consumed by the user's task and so owned by the
user.
- After task localization is done, the whole directory system/$jobid/$taskid
is moved to users/$jobid/ and owned by the user.
- when the task finishes, the whole users/$user/$jobid/$attemptid/output
directory needs to be moved to outputs/$jobid/$attemptid.
- cleaning up of a task is removal of users//$jobid/$attemptid
- cleaning up a job is removal of users//$jobid, system/$jobid
These changs will be needed for both DefaultTaskController as well as the
LinuxTaskController.
LinuxTaskController uses the setuid binary to do the move operations as the
root and changing ownership of the target files to the user.
Distributed cache files and the log files still need to be baked into this
structure.
> Per-job local data on the TaskTracker node should have right access-control
> ---------------------------------------------------------------------------
>
> Key: HADOOP-4491
> URL: https://issues.apache.org/jira/browse/HADOOP-4491
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: security
> Reporter: Arun C Murthy
> Assignee: Vinod K V
> Attachments: HADOOP-4491-20090623-common.1.txt,
> HADOOP-4491-20090623-mapred.1.txt, HADOOP-4491-20090703-common.1.txt,
> HADOOP-4491-20090703-common.txt, HADOOP-4491-20090703.1.txt,
> HADOOP-4491-20090703.txt, HADOOP-4491-20090707-common.txt,
> HADOOP-4491-20090707.txt, HADOOP-4491-20090716-mapred.txt
>
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
