Thanks! I completely understood what I want :)
I'll try to use CephFS.
Best,
Ryosuke
On Mon Dec 15 2014 at 12:17:12 PM Christian Balzer <chibi@xxxxxxx> wrote:
Hello,
On Mon, 15 Dec 2014 02:39:25 +0000 riywo wrote:
> Hi ceph users,
>
> I'm a new user of ceph, just POC-ing for a shred file system. I'm
> struggling that a file on top of a same image of RBD is not synced
> between two machines.
>
You're looking for a shared filesystem (like NFS, CephFS), but RBD (like
DRBD, or iSCSI) provides replicated, shared BLOCK devices.
You need to either run a cluster aware filesystem like ocfs2 on top of the
block device or use CephFS.
> I've just setup a ceph cluster (giant on ubuntu14.04), one mon on
> machine1 and one osd on machine2, created a image(foo) and
> mapped/mounted the image on two machines like below.
>
> machine1$ mount
> /dev/rbd1 on /mnt/foo type ext4 (rw,sync,_netdev)
> machine2$ mount
> /dev/rbd1 on /mnt/foo type ext4 (rw,sync,_netdev)
>
> Then, I've just touched a file(/mnt/foo/bar) on machine1, but it didn't
> appear on machine2 even I waited for a few minutes.
>
> machine1$ sudo touch /mnt/foo/bar
> machine1$ ls -l /mnt/foo
> -rw-r--r-- 1 root root 0 Dec 15 01:35 bar
> drwx------ 2 root root 16384 Dec 12 16:05 lost+found
> machine2$ ls -l /mnt/foo
> drwx------ 2 root root 16384 Dec 12 16:05 lost+found
>
> Though, it appeared after flushing block device buffers.
>
> machine1$ sudo blockdev --flushbufs /dev/rbd1
> machine2$ sudo blockdev --flushbufs /dev/rbd1
> machine2$ ls -l /mnt/foo
> -rw-r--r-- 1 root root 0 Dec 15 01:35 bar
> drwx------ 2 root root 16384 Dec 12 16:05 lost+found
>
> Am I wrong something? Are there something to be configured? Is it
> impossible to use RBD as a shared network block device?
>
Shared block device != shared file system.
Christian
--
Christian Balzer Network/Systems Engineer
chibi@xxxxxxx Global OnLine Japan/Fusion Communications
http://www.gol.com/
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
