"Pshem Kowalczyk" <pshem.k at gmail.com> writes:
> First I would like to thank you for great product - slony1 is a great
> master-slave replication solution :-)
>
> Due to historical reasons we have multiply databases running on a
> single machine (using the same db engine). One of those databases is
> replicated using slony1 to other machines. Now we have to replicate
> the other database (across the same physical hardware)
> Should I create new node for every database and the create a new
> independent set?
> and have something like this:
>
> master db:
> node 1 - database 1
> node 11 - database 2
>
> slave 1
> node 2 - database 1
> node 12 - database 2
>
> slave 2
> node 3 - database 1
> node 13 - database 2
>
> or is there another way of doing this?

Could you be a little more precise about this?  I suspect that the
answer is "you need a new node each time," but it is a bit ambiguous
based on what you have said.

You need a node for each distinct database that you are using, for sure.

Thus, if the conninfo for nodes 1 and 11 are different, then they
definitely need to be separate nodes.

Suppose, for node 1, the DB is identified as:
  dbname=db1 port=5432 host=masterhost

and for node 11, the DB is identified as:
  dbname=db2 port=5432 host=masterhost

then it's quite clear that you need 2 nodes corresponding to your
"node 1" and "node 11."

It is less obvious that nodes 2 and 12 need to be distinct, ditto for
3 and 13.  They may be able to coexist happily as one node on each of
those servers.

But if these databases are quite independent, I'd be inclined to have
them be *completely* separate, as two Slony-I clusters that don't
necessarily share much of anything.
-- 
select 'cbbrowne' || '@' || 'ca.afilias.info';
<http://dba2.int.libertyrms.com/>
Christopher Browne
(416) 673-4124 (land)