Hi , Without the engine - this will be hard.The engine knows the configuration and the disks of each VM so the Hypervisor can start it. In your case I would focus on the engine as this could be (or not) the easier approach. Otherwise you need to run qemu-img against each image (just an uuid, no meta or lease ) , find your image and copy it to a standalone KVM. Once you have the disk, you should try to recall what was the host CPU type (should be the same on the whole cluster) and create a new VM using a COPY of your image . If KVM manages to start and the machine boots - you are done. As you can see , fixing the engine might be easier - especially with the help of the community. Best Regards, Strahil NikolovOn Mar 20, 2019 18:01, siovelrm(a)gmail.com wrote: