On Wed, Dec 19, 2007 at 09:37:24PM -0500, Josh Harrison wrote:
> In the production environment is this okay? Anyone had tried this? Are there

Basically, you have two choices in doing this:

1.  Schedule outages for upgrades.  If there's nothing else going on, the
schema changes should go through very quickly.

2.  Do some timing trickery.  There was a problem report just yesterday or
today that described how someone normally does this.  Basically, you use the
replica as a read-only system, stop the slons, perform the EXECUTE SCRIPT on
the origin, and then move your application back.  Start the slons and
thereby allow the replica to catch up.  There are potential gotchas about
this; test your procedure carefully!

A