Description of problem: I used F19 with old libvirt and the setup crashed (taking screen session with it), VM with running engine was killed and all of this because the injected host reported that is does not support required Cluster level. I fixed this on the hypervisor node and tried to find a way to repeat just the last steps of ovirt-hosted-engine-setup.. and I discovered that there is no way to do that apart from installing the VM again. RFE: Create setup recovery scenarios (documentation, code or both) to recover from a situation like this without having to reinstall the whole thing.
or this use case we have most of the infrastructure in place. Targeting to 3.5.
An answer file is generated at validation stage before changing the system state, including before starting the VM. If the user run "hosted-engine --deploy --config-append=/etc/ovirt-hosted-engine/answer.conf" question are answered automatically. If the disk image has not been deleted from the within the storage domain, it will be reused by the VM preserving the content if boot is pxe or cdrom, overwritten in boot is from ovf. User just need to confirm that OS have been installed on first checkpoint and rhevm has been installed at second checkpoint and if rhevm was really installed and able to run everything is already in place without any more action needed. Is this enough? If so, we can move this to closed, current release, target 3.4.0.
Closing as current release as per comment #2. Please reopen if current solution is not considered good enough, proposing a different solution.