Date: Wed, 1 Jun 2011 09:06:41 -0400
From: Ray Stell <stellr@cns.vt.edu>
To: Greg Smith <greg@2ndquadrant.com>
Cc: pgsql-admin@postgresql.org
Subject: Re: patching the OS of a 9.0.4 db with hot standby
Message-ID: <20110601130640.GD22815@cns.vt.edu>
References: <20110519134009.GA16689@cns.vt.edu>
	<4DE618D7.7000102@2ndquadrant.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <4DE618D7.7000102@2ndquadrant.com>
User-Agent: Mutt/1.5.17 (2007-11-01)

On Wed, Jun 01, 2011 at 06:47:51AM -0400, Greg Smith wrote:
> Ray Stell wrote:
>> I want to do OS patching on prod and hot standby.  What is the advised
>> procedure for maintenance on the two systems?  I would think patching
>> the standby should come first in case it blows something up. Oracle has
>> a defer switch you can throw on prod to tell it to relax on log shipping
>> to the standby.  Do we have that here?
>>   
>
> Your intuition here is reasonable--test on the standby, if things go wrong 
> consider reversing the patch, with no impact on the master.  It's not 
> covered in the main because that tends not to cross over into OS management 
> considerations.
>
> The main parameter you could consider tweaking in this case is 
> wal_keep_segments, which will let you adjust how long the master queues 
> things for the standby before it gives up.  If you really do lose the 
> standby, you can run out of disk space on the master eventually if you 
> haven't set this properly.
>
> -- 
> Greg Smith   2ndQuadrant US    greg@2ndQuadrant.com   Baltimore, MD
> PostgreSQL Training, Services, and 24x7 Support  www.2ndQuadrant.us
> "PostgreSQL 9.0 High Performance": http://www.2ndQuadrant.com/books


Thanks for the suggestion.  It's great that the keep parm can be
adjusted on the fly.