On Tue, 2018-11-06 at 13:21 +0000, Tony Brian Albers wrote:
On Tue, 2018-11-06 at 13:47 +0100, Karli Sjöberg wrote:
> >
On Tue, 2018-11-06 at 12:36 +0000, Tony Brian Albers wrote:
> > Hi guys,
> >
> > I have 3 machines that I'd like to test oVirt/gluster on.
> >
> > The idea is that I want two of them to be oVirt nodes and the
> > third
> > to
> > run the oVirt engine.
> >
> > They all have 8 hdd's, and I'd like to use these in gluster for
> > storing
> > the vm images. (on all 3 machines)
> >
> > So,
> > machine1 is a gluster server with oVirt engine installed
> > machine2 is a gluster server and oVirt node
> > machine3 is a gluster server and oVirt node
> >
> > Is this doable?
>
> Usually you´d want to deploy oVirt/Gluster "hyperconverged"[*] with
> three machines. Hyperconverged means that the servers are deployed
> both
> for compute and storage at the same time, and the engine is a VM in
> it´s own cluster (it doesn´t have to be a separate server).
>
> Or is this something you really don´t want? If so, why?
>
> [*]:
>
>
https://ovirt.org/documentation/gluster-hyperconverged/Gluster_Hyperc
> onverged_Guide/
>
> /K
>
Machine1 only has a very limited amount of RAM, so it's not usable as
a
node.
And I'd like to keep the engine outside, at least for now.
OK, no that is not supported. I mean, it was a long time ago but not
any more. I wouldn´t worry about the engine being a VM, it´s acutally
very stable, net positive compared to having it physical, in my
opinion. And more available compared to having hardware outage.
What I would do is set it up as three node hyperconverged and then just
remove "Machine1" from the "Hosts" interface, so it´s not used for
compute.
/K
/tony
--
--
Tony Albers
Systems Architect
Systems Director, National Cultural Heritage Cluster
Royal Danish Library, Victor Albecks Vej 1, 8000 Aarhus C, Denmark.
Tel: +45 2566 2383 / +45 8946 2316