With Maverick's UEC / Euclayptus, there is no way to mount EBS volumes because libvirt gives the error:
sdh uses scsi:1, rather than scsi:0.
The error and the patch applied upstream can be seen at: http://osdir.com/ml/libvir-list/2011-01/msg01268.html
As you can see it's a single line change, the affected file for the maverick version is: src/qemu/qemu_driver.c (seems like they did some code refactoring on the version used on that email )
I applied the patch by hand, recompiled and installed it and the ebs mount worked.
===============================
SRU Justification:
1. impact: VMs are not able to mount drives
2. How bug was addressed: a patch was cherrypicked from upstream which fixes the mistaken assignment of 0 rather than the chosen controller.
3. Minimal patch: see the linked bzr tree
4. To reproduce: mount a scsi device as /dev/sdh.
1. # virsh attach-disk --target sdb ...
2. # virsh attach-disk --target sdh ...
5. Regression potential: only one line is changed, following a patch from upstream.
===============================
Confirmed based on the fact that it is confirmed upstream :)
Patch needs to be re-generated against qemu_driver.c as qemu_hotplug.c doesn't yet exist in maverick's libvirt. I will create the patch and SRU.
For SRU justification, from the linked upstream email:
Steps to reproduce this bug:
> 1. # virsh attach-disk --target sdb ...
> 2. # virsh attach-disk --target sdh ...
> error: Failed to attach disk
> error: operation failed: target scsi:0 already exists
>
> sdh uses scsi:1, rather than scsi:0.
>