Message-ID: <506220A6.5020907@pinpointresearch.com>
Date: Tue, 25 Sep 2012 14:22:46 -0700
From: Steve Crawford <scrawford@pinpointresearch.com>
User-Agent: Mozilla/5.0 (X11; Linux x86_64;
	rv:15.0) Gecko/20120827 Thunderbird/15.0
MIME-Version: 1.0
To: "pgsql-admin@postgresql.org" <pgsql-admin@postgresql.org>
Subject: Upgrading and streaming replication
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer-Encoding: quoted-printable

I'm planning my upgrade from 9.1.5 to 9.1.6 on some master and=20
streaming-replication standby pairs and have some questions:

Generally:

http://www.postgresql.org/docs/9.1/static/warm-standby.html section=20
25.2.1 says:

"...it is likely that running different minor release levels on primary=20
and standby servers will work successfully. However, no formal support=20
for that is offered and you are advised to keep primary and standby=20
servers at the same release level as much as possible. When updating to=20
a new minor release, the safest policy is to update the standby servers=20
first =97 a new minor release is more likely to be able to read WAL files=
=20
from a previous minor release than vice versa."

My mental translation of this is "...try updating the standby first and=20
hope it works - if not and you at some point find your standby is scrod,=20
you're on your own...we warned you..."

Is this the correct interpretation? I'd feel better if there were a=20
documented approved upgrade process for systems using streaming=20
replication and that the release notes would specify when a minor=20
upgrade involves WAL file changes that require special handling (and it=20
would be nice if PostgreSQL would detect dangerous format mismatch=20
issues and refuse to start).

Specifically:

Neither the update release notes on the wiki=20
(http://wiki.postgresql.org/wiki/20120924updaterelease as of this=20
writing) nor the release notes=20
http://www.postgresql.org/docs/9.1/static/release-9-1-6.html nor news=20
http://www.postgresql.org/about/news/1416/ indicate that any special=20
handling is required on standby servers. Can anyone confirm if that is=20
correct or, if not, what additional steps are prudent?

Cheers,
Steve