I thank you very much for sharing your knowledge with me.
My VMs got defined by virt-manager with
<driver name='qemu' type='raw'/>
At first I'll give a try to Justin's configuration tweak and let you know what happen if I change it to
<driver name='qemu' type='raw' cache='none'/>
I'm also really happy to know how to dynamically add or remove an iscsi LUN from my server.
Have a nice day.
Le 28/10/2010 00:05, Justin Clift a écrit :
On 10/28/2010 06:06 AM, Mark van Walraven wrote:On Wed, Oct 27, 2010 at 11:10:25AM +0200, Gildas Bayard wrote:My understanding is that blade A got it's iscsi disk cache up and running and that when the VM comes back, blade A has no way to know that the VM got its disk altered by its stay on blade B for a while. Hence the corruption.Another possible reason is caching by KVM. Check in the <disk> element of the domain definition for one like: <driver name='qemu' cache='none'/> Without it, KVM defaults to 'writethrough' caching, which is not safe for migration.Gildas, are you ok to check this setting and let us know if it helps you? This one setting seems like it's what you might need, and could be a simple fix. It *might* affect your virtual machine's performance though. (not sure) Phil's end-to-end configuration sounds really useful too, but a bit more involved to adapt if there's an easy fix. :)
Responsable de l'équipe Assistance Informatique et Technique
Laboratoire HEUDIASYC - UMR CNRS 6599
UTC Centre de Recherches de Royallieu
BP 20529 - 60205 Compiègne Cedex
Tél. 03 44 23 46 71