On Tue, Mar 12, 2013 at 2:35 AM, Adrian Allen <[email protected]> wrote: > I suspect that you may not get an answer from anyone on this issue, because > I don't think anyone else has yet gotten this working on disparate physical > hosts. I was trying for a while, and although I did get some suggestions > that might work, I haven't had time to try again to implement it. > > To be more specific, using fence_virt to manage KVM guests on different
Btw. fence_virsh != fence_virt I've made that mistake often enough too :) > physical hosts apparently hasn't been done yet. I think it would work in a cman/pacemaker stack, but not corosync 2.x > But it would be useful to > know what the behavior would be in that instance. > > It's possible that you might be able to sidestep the issue by treating your > guests as resources managed by Pacemaker, so that the guests themselves > never even know they are part of a cluster. I was going to look into this > next if I couldn't figure out how to get qpid working. You could also try David's brand new whitebox resources. You define the VMs as resources and then use pacemaker_remoted (no corosync required) to manage other resources that run inside them. _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
