[Pacemaker] Two resource nodes + one quorum node

[Jacek Konieczny]

On Thu, 13 Jun 2013 15:50:26 +0400
Andrey Groshev <greenx at yandex.ru> wrote:
> 11.06.2013, 22:52, "Michael Schwartzkopff" <misch at clusterbau.com>:
> 
> Am Dienstag, 11. Juni 2013, 22:33:32 schrieb Andrey Groshev:
> 
> > Hi,
> 
> > I want to make Postgres cluster.
> 
> > As far as I understand, for the proper functioning of the cluster
> > must use a
> 
> > quorum (ie, at least three nodes).
> 
> No. Two nodes are enough. See: no-quorum-policy="ignore".
>  
> Very big thanks, but I know it. :) 

And what is wrong with a 2-node quorum provided by the corosync '2 node'
mode?

AFAIK it is better than no-quorum-policy="ignore", as it prevents the
'fence loop' – node won't fence the other node or do anything dangerous 
just after booting up after being fenced, because it cannot get quorum
without the other, but when the cluster is booted properly with both
nodes, any of the two can fail and the cluster would continue in
degraded mode.

I agree that using three nodes when two would be ok is an overkill. When
we sell a 'high availability' solution, the customer expects two devices
- one 'normal' and one 'backup'. It would be hard to sell him the third
machine (same high-end server as the other two? or some 'stupid little
box' just for the quorum-keeping?), that 'does nothing'. And it is not
only the purchase cost, but also power and rack space.

And even in a three-node cluster there are still many things that may go
wrong (with very little probablility) so even if it is slightly more
reliable it is probably not worth it. There is no fail-proof solution
for any IT problem. If a 2-node cluster handles gracefully 95% of
possible failures and 3-node cluster would handle 98%, is it worth
installing the third machine?

Greets,
	Jacek