[Linux-cluster] A few GFS newbie questions: journals, etc

[Lon Hohberger]

On Thu, 2005-06-09 at 11:48 -0500, Nate Carlson wrote:

> My case is a little different - we're not talking physical hardware here, 
> it's virtual machines.

I think you said that originally, and I completely forgot about that
fact.

> I've got 4 physical boxes, and will be running ~15-20 virtual machines on 
> these physical boxes. I'd like to be able to tolerate any of the virtual 
> machines going down, and still have the remaining VM's able to get to GFS. 
> (Many of the VM's are just development toys, so will be rebooted / halted 
> / started on a regular basis.)

Interesting.

If you had 5 physical machines, you could have xen0 be the lock_gulm
master and only have xenU access the GFS volumes acting as clients.
Doing it this way should theoretically allow 2 physical boxes to fail
and any number of xenUs fail...

(Rik?)

-- Lon