[Pacemaker] How to make a redundant structure of arbitrator?

[Jiaju Zhang]

On Tue, 2012-03-27 at 16:56 +1100, Andrew Beekhof wrote:
> On Tue, Mar 27, 2012 at 2:40 PM, Gao,Yan <ygao at suse.com> wrote:
> > On 03/27/12 10:33, Andrew Beekhof wrote:
> >> On Tue, Mar 27, 2012 at 2:34 AM, Jiaju Zhang <jjzhang at suse.de> wrote:
> >>> On Mon, 2012-03-26 at 11:50 +0900, Yuichi Seino wrote:
> >>>> Hi Jiaju,
> >>>>
> >>>> I have a question about booth.
> >>>> I would like to know if a redundant structure of arbitrator is possible.
> >>>> If it is possible, Please tell me how to the approach.
> >>>
> >>> If I understand the question correctly, you mean that the arbitrator
> >>> should be redundant in case it might be down. Well, this can be resolved
> >>> by adding more arbitrators, for example, you have 2 sites, and configure
> >>> 3 arbitrators.
> >>
> >> The arbitrator is already redundant isn't it?
> >> Because you need a cluster node to fail before an arbitrator failure
> >> has any ill-effect.
> > I think the term "arbitrator" that Jiaju referred to is the booth daemon
> > running on a single machine, which is added to make sure the amount of
> > booth daemons are uneven number.
> 
> Right, but making it redundant is only useful if it and one of the
> real sites is already down.
> If you've lost two sites (real + tie-breaker), you already have bigger
> issues than validating the sites are really gone and manually granting
> a ticket.

Yes, if 2 of 3 is down, the ticket will be revoked by booth.

Thanks,
Jiaju