On Wednesday 02,February,2011 11:04 AM, Daniel Hokka Zakrisson wrote:
> It sounds like you're doing it wrong. Why don't you fail-over on a
> guest-by-guest basis?
We run 2-node heartbeat (version 1) clusters with all the vservers on
the same shared storage (DRBD). We can't fail-over individual guests -
the hosts and the shared storage are primary/secondary.
And yes, we could go to a more elaborate setup of primary/primary,
heartbeat version 2, and a cluster file system or separate LVM
partitions for every guest, etc., etc. But this way is very simple to
set up and maintain, and it works well for our needs (once we order the
startup of the vservers, of course). ;-)
Jeff Jansen
Received on Wed Feb 2 03:26:31 2011