I have been trying for quite some time to launch a KVM VM
from a CEPH RBD volume using OpenNebula. I have gotten past the
permissions issues and to the point where kvm can actually
start a virtual machine, but we are getting a "Geom Error"
as soon as the virt console comes up.
(note, not a GRUB Geom error).
As far as I can tell this means that the kvm bios
can't understand enough of the geometry of the disk file as presented by
RBD (which is a RAW format image) to even find the boot sector
to get as far as GRUB. Have not seen anything with this specific
error in Google though.
The hardware that I am using for the Ceph test is going away
at the end of this week and if I can't get past this problem in
the next couple days I will have to recommend to my management
that Ceph is not ready for my production cloud environment. Appreciate
the help we have gotten thus far and hope the list can come through
one more time.
To rehash--I have posted this on other threads before:
Sci. Linux 6.5 (redhat clone) + Kernel 3.10 + ceph-compiled
qemu-kvm and qemu-img as downloaded from the Ceph site.
I can mount the rbd volume outside of qemu/kvm with
a normal rbd map on the same machine, verify the full
partition table structure is there and all the files are there.
KVM must be doing the rbd map and mount correctly because
I can see the device in question mapped to a /dev/rbd1.
So I must be dealing with some problem with the format
of the image, only question is what? Libvirt is presenting
the rbd volume to the virtual machine as /dev/vda
and it was installed on a classic virtual machine (i.e. vda
was pointing to a partition on a local hard disk) with the boot
sector on the first sector of vda.
Any help is appreciated.
Thanks
Steve Timm
------------------------------------------------------------------
Steven C. Timm, Ph.D (630) 840-8525
[email protected] http://home.fnal.gov/~timm/
Fermilab Scientific Computing Division, Scientific Computing Services Quad.
Grid and Cloud Services Dept., Associate Dept. Head for Cloud Computing
_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
