Description of problem: After running hosted-engine --deploy on one host properly, running it on the second fails with the error that the UUID of the current computer is already in the database. Version-Release number of selected component (if applicable): How reproducible: Very Steps to Reproduce: 1. Use multiple Supermicro based systems (same Board revision) 2. Try to add them into the same hosted-engine setup. Actual results: The hosted-engine --deploy process fails. Expected results: The hosted-engine --deploy process succeeds and the following nodes are visible to the engine. Additional info: Used CentOS 7 with oVirt 3.5 nightly snapshots (through repository).
Workaround (so it seems): Use uuidgen > /etc/vdsm/vdsm.id to generate a new UUID. For restoring a current oVirt installation, please take measures to backup this file, as this BECOMES your new UUID for this machine.
As discussed in mailing list, closing as wontfix since this is a corner case hitting only systems with bugged bioses. http://lists.ovirt.org/pipermail/users/2014-September/027292.html Workaround for the issue is available in comment #1.