On 11/01/2013 03:07 AM, nicolasc wrote:
Hi every one,
I finally and happily managed to get my Ceph cluster (3 monitors among 8
nodes, each with 9 OSDs) running on version 0.71, but the "rbd map"
command shows a weird behaviour.
I can list pools, create images and snapshots, alleluia!
However, mapping to a device with "rbd map" is not working. When I try
this from one of my nodes, the kernel says:
libceph: bad option at 'rw'
Which "rbd" translates into:
add failed: (22) Invalid argument
Any idea of what that could indicate?
Thanks for the report! The rw option was added in linux 3.7. In ceph
0.71, rbd map is passing the 'rw' and 'ro' options to tell the kernel
that to map the image as read-only or read/write. This will be fixed
in 0.72 by:
https://github.com/ceph/ceph/pull/807
Josh
I am using a basic config: no authentication, default crushmap (I just
changed some weights), and basic network config (public net, cluster
net). I have tried both image formats, different sizes and pools.
Moreover, I have a client running rbd from Ceph version 0.61.9, and from
there everything works fine with "rbd map" on the same image. Both nodes
(Ceph 0.61.9 and 0.71) are running Linux kernel 3.2 for Debian.
Hope you can provide some hints. Best regards,
Nicolas Canceill
Scalable Storage Systems
SURFsara (Amsterdam, NL)
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
