[Libguestfs] [libguestfs] Problem using virt-sysprep with RBD images

[Richard W.M. Jones]

On Wed, Nov 06, 2013 at 02:52:06PM -0700, Jon wrote:
> Hello Rich,
> 
> Interesting.  Thanks for the explanation.
> 
> When you specify an rbd on the command line for virt-sysprep, do you expect
> the path to include the monitor address?
> 
> e.g.:
> >> virt-sysprep -a rbd://host-name/pool-name/device-name

Note that these URIs are not Ceph URIs.

They are something that libguestfs virt tools invent, implemented in
fish/uri.{c,h} and described here:

  http://libguestfs.org/guestfish.1.html#adding-remote-storage

Below that layer is the interface to libguestfs-as-a-library,
described here:

  http://libguestfs.org/guestfs.3.html#remote-storage

So if I understand your question correctly, then I guess the answer is
"no", although I'm not sure what a "monitor address" is.  Because
libguestfs virt tools URIs are nothing like and unrelated to Ceph URIs.

> If I understand correctly, libvirt is able to understand the ceph
> configuration, so when I create a device with qemu-img I only specify the
> protocol and pool/device.
> 
> e.g.:
> >> qemu-img create rbd:pool-name/device-name 5G
> 
> (there is some voodoo that I don't understand, I've got a whole thread on
> trying to get qemu-img to create format 2 rbds by default... but that's for
> another thread)
> 
> Would it be possible to specify rbds like this instead?  Or is the scope
> bigger than I'm understanding and that would cause issues with other disk
> types specified for the --add parameter.  It seems like --add can take
> either a URI or a physical disk path.

I don't know, but see above.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
virt-top is 'top' for virtual machines.  Tiny program with many
powerful monitoring features, net stats, disk stats, logging, etc.
http://people.redhat.com/~rjones/virt-top