[ovirt-users] Is it possible to recover from a failed Engine host?

[Alexander Wels]

On Wednesday, May 2, 2018 8:03:53 AM EDT ~Stack~ wrote:
> Greetings,
> 
> I have a dev environment where it seems the hard drive on our Engine
> host kicked the bucket (Yeah, I know. Smartmon. I watch it closely on
> the systems I care about - this was a learning environment for me so I
> didn't).
> 
> The Hypervisors are fine and the VM's running on the Hypervisors are
> fine...But I can't manage any of the Hypervisors. To make things a bit
> more tricky, the SQL and the backups were on the drive that died. I
> really don't have anything from that host. It's dev. I can rebuild. But
> it is also a learning environment for me so might as well use this to learn.
> 
> Is it possible for me to build a new Engine host and attach it to an
> existing hypervisor environment? Better yet, would this be something I
> could do as a hosted-engine-deploy? (something I haven't experimented
> with yet.)
> 
> Again, this is a play ground so if it goes horrifically wrong...oh well.
> But I would really like to try to recover it for the learning
> experience. I've been poking around in the documentation but I haven't
> seen anything that seems to address this issue directly.
> 
> Thoughts?
> 
> Thanks!
> ~Stack~

As long as the storage domain is in tact you should be able to recover 
everything. And it does sound like this is the case as the VMs are still 
running. Basically you just install a new engine somewhere and then do the 
following:

- Create new Data Center
- Create new Cluster
- You will need a host to add to your cluster. Add this host.
- Create a small temporary storage domain, this will allow you to bring up the 
data center which in turn will allow you to IMPORT the existing storage 
domain.
- Once the DC is up, you can 'import' the existing storage domain, it will 
warn you that the storage domain is still attached to another DC, but since 
that engine is gone, you can ignore that.
- Once the new DC is imported you can stop/detach/remove the small temporary 
storage domain, which will make the imported storage domain, the master 
domain.

Once all that is done, you can simply go to the storage domain, and 'import' 
whatever VM/template you have stored on the storage domain, and it will show 
up in the VM/template list. Then you add all your hosts and you should have a 
running environment again.