On Tue, Mar 14, 2006 at 02:40:23PM -0000, Victoria Parsons wrote:
> its not a problem. I started doing this when I got errors from programs
> with a connection to postgres complaining because it could no longer
> find table with oid XXX.

It wasn't a table, I'll bet.  The problem was likely that the
_function_ disappeared.  The query plans for the trigger functions
are cached per connection.  When you remove any of those functions,
even if you re-add them, the old query plan is still there, and you
get this error.  The reference is by oid, and when you remove and
re-add the trigger function, the oid reference now points to the
wrong entry.

> I've got to the stage where I'm adding new sets and tables to an
> established cluster. After adding a new set and getting all current
> subscribers to subscribe to this, will something funny happen to my new
> tables that requires postgres connections to be re-established? To try

Shouldn't do, no.

A


-- 
Andrew Sullivan  | ajs at crankycanuck.ca
The plural of anecdote is not data.
		--Roger Brinner