On 08/23/2013 12:00 AM, Olivier Desport wrote:
Le 22/08/2013 17:04, Joe Julian a écrit :
On 08/22/2013 07:58 AM, Olivier Desport wrote:
Le 22/08/2013 16:34, Joe Julian a écrit :
On 08/22/2013 06:31 AM, Olivier Desport wrote:
Le 22/08/2013 15:07, Olivier Desport a écrit :
I've removed a volume and I can't re-create it :
gluster volume create gluster-export gluster-6:/export
gluster-5:/export gluster-4:/export gluster-3:/export
/export or a prefix of it is already part of a volume
I've formatted the partition and reinstalled the 4 gluster
servers and the error still appears.
Any idea ?
Some more information:
I've formatted in OCFS2 and GlusterFS version is 3.3.1-1.
I've tried to set attributes with setfattr command but it still
doesn't work.
Isn't running a clustered filesystem on top of a clustered
filesystem a little bit redundant?
Perhaps but can I format my isci shared volume in xfs and mount it
on several machines to share with GlusterFS ?
Wait... so are you saying that /export is the same shared filesystem
on each of those servers?
Yes. I want to use GlusterFS to have an HA network share.
GlusterFS *is* a clustered filesystem. Provided you use replication, it
is a HA network share all by itself. You don't need to put it on top of
another clustered filesystem. You especially cannot use the /same/
filesystem for multiple bricks.
In your other email, you suggested that you could scp to the same target
from two clients when your bricks were xfs. If you're mounting the
glusterfs volume on your client(s) (mount -t glusterfs) and accessing
your same file through those mountpoints, that wouldn't happen. Standard
posix locking would happen and only one client could write to the file
at once.
_______________________________________________
Gluster-users mailing list
[email protected]
http://supercolony.gluster.org/mailman/listinfo/gluster-users
