[ovirt-users] Hosted engine setup shooting dirty pool

[Jamie Lawrence]

> On Apr 12, 2017, at 1:31 AM, Evgenia Tokar <etokar at redhat.com> wrote:
> 
> Hi Jamie, 
> 
> Are you trying to setup hosted engine using the "hosted-engine --deploy" command, or are you trying to migrate existing he vm? 
>  
> For hosted engine setup you need to provide a clean storage domain, which is not a part of your 4.1 setup, this storage domain will be used for the hosted engine and will be visible in the UI once the deployment of the hosted engine is complete.
> If your storage domain appears in the UI it means that it is already connected to the storage pool and is not "clean”.

Hi Jenny,

Thanks for the response.

I’m using `hosted-engine —deploy`, yes. (Actually, the last few attempts have been with an answerfile, but the responses are the same.)

I think I may have been unclear.  I understand that it wants an unmolested SD. There just doesn’t seem to be a path to provide that with an Ovirt-managed Gluster cluster.

I guess my question is how to provide that with an Ovirt-managed gluster installation. Or a different way of asking, I guess, would be how do I make Ovirt/VDSM ignore a newly created gluster SD so that `hosted-engine` can pick it up? I don’t see any options to tell the Gluster cluster to not auto-discover or similar. So as soon as I create it, the non-hosted engine picks it up. This happens within seconds - I vainly tried to time it with running the installer.

This is why I mentioned dismissing the idea of using another Gluster installation, unattached to Ovirt. That’s the only way I could think of to give it a clean pool. (I dismissed it because I can’t run this in production with that sort of dependency.)

Do I need to take this Gluster cluster out of Ovirt control (delete the Gluster cluster from the Ovirt GUI, recreate outside of Ovirt manually), install on to that, and then re-associate it in the GUI or something similar?

-j