[Gluster-users] ownership of link file changed to root:root after reboot brick service

Brian Candler B.Candler at pobox.com
Wed Jan 16 07:32:35 UTC 2013


[Aside: please do not cross-post. If you have a problem using glusterfs, and
you are not looking at the source code and proposing a specific path, then
I suggest "glusterfs-users" is the appropriate place]

On Wed, Jan 16, 2013 at 02:53:40PM +0800, huangql wrote:
>           I have encountered a strange problem that the ownership of
>    linkto file was changed to root:root after we reboot server. And I
>    reproduced the problem many times.

Can you show "gluster volume info" output please? Is this a distributed
volume, a replicated volume, or something else?  Are you doing anything else
at the same time, e.g. rebalance?

>    For example: ff_2
...
>    From server side, we have 2 bricks and we can see the ff_2 status:
> 
>    brick1:
> 
>    -rw-rw-r-- 1 test test  0 Jan 12 11:38 ff_1
>    ---------T 1 test test   0 Jan 13 00:43 ff_2
> 
>    brick2:
> 
>    -rw-rw-r-- 1 test test 5242880 Jan 12 11:31 f_3
>    -rw-rw-r-- 1 test test 5242880 Jan 12 11:31 f_4
>    -rw-rw-r-- 1 test test 5242880 Jan 13 00:43 ff_2

This seems wrong either way:

* If it's a distributed volume, there should not be ff_2 on both.

* If it's a replicated volume, there should be ff_1, ff_2, f_3 and f_4 on
  both.

On the bricks, please show the xattrs on these files, and also their
parent directories. e.g.

    getfattr -d -m'^trusted\.' /path/to/ff_2
    getfattr -d -m'^trusted\.' /path/to

Somebody on this list may then be able to tell you what state they are in
and how they got into it (but not me I'm afraid, I don't know how the
internals work)

Regards,

Brian.



More information about the Gluster-users mailing list