Skip site navigation (1) Skip section navigation (2)

Re: Desaster Recovery from failover using SLONY-I

From: Andrew Sullivan <ajs(at)crankycanuck(dot)ca>
To: pgsql-admin(at)postgresql(dot)org
Subject: Re: Desaster Recovery from failover using SLONY-I
Date: 2007-05-30 15:47:32
Message-ID: 20070530154732.GG16260@phlogiston.dyndns.org (view raw or flat)
Thread:
Lists: pgsql-admin
On Wed, May 30, 2007 at 12:57:55PM +0200, Peter Hausmann wrote:
> Hi,
> 
> After a node is dropped due to failover, it takes days to recover the
> database, because it is build up from scratch.

This is from Slony?  You know that that's a built-in limitation, and
that the preferred method is switchover, right?  Failover should be
the last result.  (Also, you'll get better results if you discuss
this on the slony list.)

> We would prefer another approach:
> Make an Online Backup of the Provider database.
> The provider database continues writing new data and the Slony-Logs.
> The Backup is restored to the failed system.
> The slony cluster is dropped from the restored system
> and the system is defined as Slony-subscriber.
> How can we setup this system to recover from the status of the backup time?

Slony knows nothing about the WAL-archiving backup method.  When you
subscribe a datbaase, it automatically drops and recreates the target
from scratch.  There has been some discussion of how to improve that,
but AFAIK nobody has written the support for it yet.

A

-- 
Andrew Sullivan  | ajs(at)crankycanuck(dot)ca
A certain description of men are for getting out of debt, yet are
against all taxes for raising money to pay it off.
		--Alexander Hamilton

In response to

pgsql-admin by date

Next:From: Chris HooverDate: 2007-05-30 17:24:14
Subject: Fwd: Out of Memory on Reindex
Previous:From: Peter HausmannDate: 2007-05-30 10:57:55
Subject: Desaster Recovery from failover using SLONY-I

Privacy Policy | About PostgreSQL
Copyright © 1996-2014 The PostgreSQL Global Development Group