On 2009-05-05T14:33:50, Juha Heinanen <j...@tutpro.com> wrote:
> > Yes, this happens because your replica loses quorum, you need to set
> > the cluster property no-quorum-policy to "ignore" for two node
> > clusters.
> i just tried this kind of reboot using heartbeat 2.99.2+sle11r3-2 and
> pacemaker 1.0.2-2 and drbd resource and resource group on top of it got
> migrated to the node that was still running and stayed there. i don't
> have any no-quorum-policy property set. how is that possible if the
> above is true?
Because you aren't using OpenAIS, apparently. Heartbeat (or CCM,
rather), for two node clusters,pretends that even in a 1:1 split
situation, each side has quorum. That is "safe", because resource
integrity is still protected via fencing/STONITH.
Yet, it is also not true per se, because _exactly_ 50% is not majority.
OpenAIS adheres more strictly according to that definition of quorum,
hence the option needs to be set to "ignore".
(OpenAIS - or rather its next version of the infrastructure, called
"corosync" - gains a more powerful quorum framework which will allow the
plugin of tie-breaker modules such as a scsi reservation etc.)
Regards,
Lars
--
SuSE Labs, OPS Engineering, Novell, Inc.
SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
"Experience is the name everyone gives to their mistakes." -- Oscar Wilde
_______________________________________________
Linux-HA mailing list
Linux-HA@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
