Hi,
On Fri, Apr 24, 2009 at 10:06:24AM +0200, Andrew Beekhof wrote:
> On Fri, Apr 24, 2009 at 09:49, Juha Heinanen <j...@tutpro.com> wrote:
> > Andrew Beekhof writes:
> >
> > ?> + ? ? ? ?crm_config_err("NOTE: Clusters with shared data need STONITH to
> > ?> ensure data integrity");
Good! Now crm will also refuse to commit such a configuration.
> > is special stonith hardware a must or is there some poor man's stonith
> > solution, for example, for shared mysql storage?
>
> If you have a shared disk, SBD is an option.
> Failing that ssh is better than nothing but still not recommended.
With ssh they risk zero availability in case of network problems
between the hosts (i.e. what they are seeing right now). The
other day somebody expected the suicide plugin to fix split
brain. The ssh should be for testing only. And suicide just for
single-node clusters.
> The question you need to ask is: What is going to cost more?
> Data corruption (and the associated service outage) or a remote power switch?
Indeed. That's the basic question one should put in just about
any project. And I understand that UPS are really affordable
nowadays.
Thanks,
Dejan
> _______________________________________________
> Pacemaker mailing list
> Pacemaker@oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
_______________________________________________
Pacemaker mailing list
Pacemaker@oss.clusterlabs.org
http://oss.clusterlabs.org/mailman/listinfo/pacemaker
