Re: Using pg_upgrade on log-shipping standby servers

On Wed, Jul 18, 2012 at 09:36:51AM -0400, Robert Haas wrote:
> On Tue, Jul 17, 2012 at 6:02 PM, Bruce Momjian <bruce(at)momjian(dot)us> wrote:
> > However, I have two ideas. First, I don't know _why_ the
> > primary/standby would be any different after pg_upgrade, so I added the
> > documentation mention because I couldn't _guarantee_ they were the same.
> > Actually, if people can test this, we might be able to say this is safe.
> >
> > Second, the user files (large) are certainly identical, it is only the
> > system tables (small) that _might_ be different, so rsync'ing just those
> > would add the guarantee, but I know of no easy way to rsync just the
> > system tables.
>
> I'm scratching my head in confusion here. After pg_upgrade, the
> master is a completely new cluster. The system catalog contents are
> completely different, and so are things like the database system
> identifier and the WAL position - yeah, the latter is approximately
> the same, but almost doesn't count except in horseshoes. Obviously
> any attempt to replay WAL from the new cluster on the old cluster is
> doomed to failure, at least unless we do a bunch more engineering here
> that hasn't really been thought about yet.

No, the point is they run pg_upgrade on the stopped primary and stopped
standbys. Are those the same? I am not really sure.

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ It's impossible for everything to be true. +