Re: [ovirt-users] Greetings and observations from an oVirt noob
On Fri, Dec 4, 2015 at 11:35 PM, Ken Marsh <kenskyfish(a)gmail.com> wrote:
Hi again,
Sorry for not getting back to you sooner.
I did read all that stuff, but what was totally unclear to me I finally
understood from the Red Hat documentation. It's just the concept that an
oVirt node's storage isn't used for VMs. I never would have guessed that in
a million years.
Using a host for both is generally called "hyperconvergence". As I wrote
above, that's not supported at all in 3.5, and partially supported in 3.6.
Official (or at least better) support is expected in 4.0.
Here's my situation - I have a bunch of Xeon workstations, each with 2TB
HDD. I'd like to use them as both RHEV nodes AND use their HDD for storage.
If I understand correctly, I can't do that. I'll need to set some up with
RHEL 6.5 and share the storage, and I'll need to set others up as RHEV nodes
to run the VMs. On the RHEL machines the CPU will go to waste and on the
RHEV nodes the disk space goes to waste. Is this right?
That's the simplest way I guess.
IIUC (Fabian?) you might be able to boot the compute nodes from the network,
so move their disks to the storage nodes.
You are also welcome to try hyperconvergence, now that 3.6 is out, but I
admit I do not know what problems, if at all, you should expect. You might
read some relevant pages on the wiki/bugzilla and/or lists archives etc.
Search for stuff like: ovirt hyperconvergence HC glusterfs
Also adding Federico.
The confusion I had about the hypervisor version is because when you are
booting the 3.5 hypervisor, it says in the splash screen "oVirt 3.2.1".
I'd
much rather it said "Press F1 to see something useful," but now that I've
figured this out I'm very happy with it.
I'm still learning and the more I learn the more I like. It's just so
conceptually different from VMWare, and the doco and naming conventions seem
to take the general concept for granted. Is the VM's disk really used over
the network and not locally copied to the oVirt node where it runs? I guess
in a purely datacenter environment that makes sense but I was hoping to make
use of the storage on nodes too.
I think better I keep learning about it and when I understand completely,
then I'll make suggestions on how to add/change the documentation. That
would probably be more helpful, right?
I guess so :-)
Thanks again - amazing product. I'm enjoying this despite the initial
confusion and trial-and-error learning.
Thanks :-)
Best,
--
Didi