hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Craig Macdonald (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-5542) Group changes cause FUSE-DFS I/O error
Date Fri, 20 Mar 2009 23:14:50 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-5542?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12685384#action_12685384
] 

Craig Macdonald commented on HADOOP-5542:
-----------------------------------------

Brian,

This would be same for any long-running Hadoop process in general. It's not specific to fuse-dfs.
Its possible the case that fuse-dfs processes are most long running.

> Group changes cause FUSE-DFS I/O error
> --------------------------------------
>
>                 Key: HADOOP-5542
>                 URL: https://issues.apache.org/jira/browse/HADOOP-5542
>             Project: Hadoop Core
>          Issue Type: Bug
>    Affects Versions: 0.19.1
>            Reporter: Brian Bockelman
>            Assignee: Pete Wyckoff
>
> If a user utilizes the FUSE mount, then has a group change, they will be unable to write
into FUSE using the new group information.
> To duplicate (assuming user brian starts only in group brian):
> 1) Write file into FUSE (or do any other action)
> 2) Add user brian to group brian2
> 3) chown anything brian owns to brian:brian2.  The error message that gets passed along
is:
> [brian@red ~]$ chown brian:brian2 /mnt/hadoop/user/brian/test_group_perms2
> chown: changing ownership of `/mnt/hadoop/user/brian/test_group_perms2': Input/output
error
> I believe this is due to the fact that group information is only looked up at NN connection
time?
> If you then remount the FUSE mount, the chown command succeeds.

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


Mime
View raw message