Thank you! That does do some interesting things, but doesn't appear to work. I've added a new storage domain, merged the differences in the metadata file back to the old one and powered it up. When I start up the nodes they endlessly report: VDSM ovirt-reqa1 command failed: (-226, 'Unable to read resource owners', 'Sanlock exception') I tried reinstalling one of them and the error message continues. At least for now I've restored the old config and the error is gone. Regards, Logan ----- On Jul 22, 2016, at 3:06 AM, Milan Zamazal mzamazal@redhat.com wrote: | Logan Kuhn <logank@wolfram.com> writes: | |> Am I correct in the assumption that importing a previously master data domain |> into a fresh engine without a current master domain is supported? | | It's supported only in case the master domain was previously correctly | detached from the data center. | | In case of an unexpected complete disaster, when a fresh engine is | installed and used, it's still possible to recover the master domain in | theory. You must find `metadata' file in the master domain and edit it | for the new engine. It's completely unsupported and it may or may not | work. We don't have guidelines how to do it, but you may try to create | a new master domain, then detach it and compare the two metadata files.