I ran into this random problem with virtualisation recently. For some reason, I just could not manage the LVM storage pools on my virtualisation server from my workstation. My workstation was running on Kubuntu 10.04 and my server was running Debian 5.04 using virt-manager and libvirt on each.
This was a very weird problem because I could access the LVM if there were no allocated logical volumes in them. However, the moment there was anything in them, virt-manager would fail to start the storage pool. This was a really weird problem because I did not have this problem on some of my other installations.
After spending days digging into it, I found out the cause of the problem.
It seems that the libvirt people changed the protocol in version 0.5.0 and swapped the colon delimeter to a comma delimeter. The workstation had a newer version of virt-manager while the server had the older version of libvirt. So, all I had to do was upgrade the libvirt from lenny-backports and that fixed the problem entirely.
The reason why I had not seen this in some other machines is because of the hardware was different. On this particular server, the harddisk was not seen as /dev/sdaX but parked under /dev/blocks/XXX:X instead. So, that is why the confusion with the “:” (colon) came into the picture between the two different versions.