The right way to keep the CentOS and ovirt up to date is to update it regularly. It keeps the element of surprise low and guarantees the full attention of the community in case of an issue. I have done so over the last 5 year and the most tasks run out of the box. Admittedly it is always exciting to do it again, but you get used to it over time and get a thick skin. No risk no fun ;-) In case of the Data(Master) Domain is down , first Up your Hosts and then Activate DataCenter/Default/Data(Master) domain. For everything else you can buy support for RHV from RH to support the community as well. By the way, it the past it was a CentOS kernel that panicked at 3ware controller, the switch to firewalld and the upgrade to Postgres 10.x are important to me. Cluster Compatibility Levels are always a nasty thing: * At cluster level (all hosts maintanance), at running VM (shutdown all VM's) and in stored VM’s (have old levels, may not be changeable (UI) (recreate new machine configs). * Maybe all clients need a new Ovirt Agent. You see, there is a downtime (e.g. changing cluster Level), but no breaks! The many improvements to Ovirt outweigh the work. :-) I don't use HE and no Gluster, that's probably where I missed the most. ;-) * Centos and Ovirt are rock solid, give it a try. * or Never touch a running system. * and/or wait for Ovirt on CentOS 8 and reinstall everything. At this time it would be appropriate to test ovirt 4.3.5 node with gluster 6 and HE on new hardware. Talk to your boss. Best Regards, Oliver