Bill Moran <wmoran at collaborativefusion.com> writes:
> The problem in question turned out to be pilot error, but the error message
> from Slony was pretty misleading.
>
> The mistake:
> In the slonik script, 2 nodes are defined, but later in the script I attempt
> to subscribe a new set to 3 nodes (the 3rd one never existed).
>
> The problem:
> Slony reports:
> PGRES_FATAL_ERROR select "_sl_schema".subscribeSet(2, 2, 3, 't');  - ERROR:  Slony-I: subscribeSet(): set 2 not found
>
> Set #2 does exist.  The error is that node #3 does not exist.  Took me a
> little longer than it should have to discover the actual problem.
>
> This is with slony 1.2.9.  From the release notes, it doesn't look like
> 1.2.10 has done anything to change this, but I'll try to reproduce it as
> soon as I get an opportunity to upgrade.  We're running against Postgres
> 8.1.

It's too late this week to do anything about it, but given a
repeatable test case, this sounds like something that might need
change.

Mind you, looking at the code in subscribeSet(), I don't see what
looks like a bug.  Here's where that message is issued:

	-- ----
	-- Check that the origin and provider of the set are remote
	-- ----
	select set_origin into v_set_origin
			from @NAMESPACE at .sl_set
			where set_id = p_sub_set;
	if not found then
		raise exception ''Slony-I: subscribeSet(): set % not found'', p_sub_set;
	end if;

Are you _certain_ that set #2 exists?

Run the query...
-- 
let name="cbbrowne" and tld="cbbrowne.com" in String.concat "@" [name;tld];;
http://linuxfinances.info/info/rdbms.html
Now is a good time to spend a year dead for tax purposes.