Thanks a lot! I appreciate the quick response and comments.

Indeed, the key=serial stuff seems to work fine. (I missed it from the
pgbench tutorial)

Unfortunally I'm not fully authorized to make db schema changes at the
moment. However, I could write a little Recommedations-like document
for future works.

Thanks again and best regards,
Manuel.

On 9/12/06, Christopher Browne <cbbrowne at ca.afilias.info> wrote:
> Manuel V?zquez Acosta wrote:
> > Hi all,
> >
> > I've been asigned to set up a postgresql replication system with
> > slony. I've read all the docs, run all the tests (with the pgbench
> > database) over two machines.
> >
> > Now I'm trying to set it up with the actual db schema for which the
> > replica must work.
> >
> > I have create some slonik scripts:
> >
> > 1) init_cluster: Which sets up the cluster: init cluster, store node,
> > store path and store listen clauses only.
> > 2) create_tablekeys: Which creates all "missing" keys: table add key
> > clauses only.
> > 3) create_repsets: Which creates the replications sets: create set,
> > set add table and set add sequence clauses.
> >
> > The 1st and 2nd scripts works fine. But when I run the 3rd one, slonik
> > says that:
> >
> > <stdin>:25: PGRES_FATAL_ERROR select
> > "_saime_cluster".determineIdxnameUnique('public.workflow_type_map',
> > NULL);  - ERROR:  Slony-I: table "public"."workflow_type_map" has no
> > primary key
> >
> >
> > However, the 2nd scripts DOES have a clause for it:
> > table add key (node id = 1, fully qualified name = 'public.workflow_type_map');
> >
> > And when I inspect the db schema (after running the 1st and 2nd
> > script) I see the slony-created key:
> >
> > CREATE TABLE workflow_type_map
> > (
> >  "type" varchar(255) NOT NULL DEFAULT ''::character varying,
> >  wid int4 NOT NULL DEFAULT 0,
> >  "_Slony-I_saime_cluster_rowID" int8 DEFAULT
> > nextval('_saime_cluster.sl_rowid_seq'::regclass)
> > )
> > WITH OIDS;
> >
> >
> > I really don't find a clue in the docs. Can anyone help me?
> >
> Drew's reservations about using "TABLE ADD KEY" are well founded, and
> I'd agree with him in general.  There is a common opinion amongst
> Slony-I developers (I'd not call it universal, but there are multiple of
> us...) that TABLE ADD KEY was a mistaken feature to add.
>
> The particular problem you run into is that it is not sufficient to
> merely request the extra column; the SET ADD TABLE request needs to
> point out that  this column should be used.
>
> notably, it should include the option 'key = SERIAL'.
>
> Thus, you shouldn't have just the statement:
>    SET ADD TABLE (set id=1, origin = 1, id=4, fully qualified name =
> 'public.workflow_type_map');
>
> You need to have...
>    SET ADD TABLE (set id=1, origin = 1, id=4, fully qualified name =
> 'public.workflow_type_map', key = serial);
>
> The 'key = serial' part is probably missing, and that, I expect, is the
> problem.
>