[Pgpool-general] uptime for maintenance

[Steven Crandell]

Heartbeat would definitely work, but for my tastes it's a bit heavy for what
you need.
In general, I don't use heartbeat for anything that is IP dependent.
e.g. if you can get away with pointing servers and services to another
destination/domain name to do the same task, don't bother with heartbeat.
If you absolutely must pass a specific IP around between servers, then H.B.
is your answer.

Pgpool doesn't ask too terribly much from a machine.
Consider bringing a second instance of Pgpool live on another machine and
flipping the DNS entry during your maintenance period.

My production HA configuration involves multiple instances of pgpool running
on different servers.
These services are effectively 0 load until they're used.
My monitoring solution executes a script that both checks for availability,
and upon failing that check, flips the service domain name (via an update to
the PDNS backend db) over to another of the already available instances.

YMMV
-s


On Thu, Aug 27, 2009 at 5:46 PM, Janos Erdelyi <janos.erdelyi at gmail.com>wrote:

> Is there a fairly standard way to do maintenance on a pgpool instance
> while not interrupting service too much for end-users?
> I was thinking of using heartbeat with a shared ip between two pgpool
> instances so that i could fail one out, do machine updates, upgrade
> the pgpool version, etc.
>
> is this a fairly typical (and successful) route?
>
> -janos
> _______________________________________________
> Pgpool-general mailing list
> Pgpool-general at pgfoundry.org
> http://pgfoundry.org/mailman/listinfo/pgpool-general
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://pgfoundry.org/pipermail/pgpool-general/attachments/20090829/c9098bc1/attachment.html>