On Fri, Jul 22, 2011 at 4:17 PM, Josh Berkus <[email protected]> wrote:
>
>> * install on its own server or, ideally, set of servers (with heartbeat, 
>> etc.)
>
> This is preferred.  Installing pgpool on the database servers is pretty
> pointless.

Thanks Josh. How do you make pgpool highly available?

At first, I thought pgpool-ha existed for that reason, but it appears
to be more for failover from a primary backend Postgres node to a
standby one (a feature which pgpool-II now incorporates, right?).

But pgpool itself remains a single point of failure. I'm trying to
figure out the best way to make pgpool itself HA. The simplest seems
to be to setup heartbeat between two pgpool servers with a virtual IP
and point clients to the virtual IP.

Since it's stateless, it seems like sticking a loadbalancer in front
of two (or more) pgpool servers could also work. (Although your
loadbalancer now becomes the new SPOF :-), unless you have two.)

Thoughts?

johnny
_______________________________________________
Pgpool-general mailing list
[email protected]
http://pgfoundry.org/mailman/listinfo/pgpool-general

Reply via email to