I had already started over via slonik_uninstall_nodes when I received
your post.  I set up Slony again with slonik_init_cluster, started slon
on both hosts, then ran slonik_create_set, then slonik_subscribe_set.
However still no replication.

With the logging verbosity turned up per your suggestion I got:

2006-10-16 19:17:13 EDT ERROR  remoteWorkerThread_1: "select
"_rep_atr_prod".setAddTable_int(1, 2,
'"public"."rpt_program_conf_pricing_b"',
'rpt_program_conf_pricing_b__Slony-I_rep_atr_prod_rowID_key', 'Table
public.rpt_program_conf_pricing_b without primary key'); "
PGRES_FATAL_ERROR ERROR:  Slony-I: setAddTable(): table
"public"."rpt_program_conf_pricing_b" has no index
rpt_program_conf_pricing_b__Slony-I_rep_atr_prod_rowID_key
2006-10-16 19:17:13 EDT WARN   remoteWorkerThread_1: data copy for set 1
failed - sleep 15 seconds

I then manually recreated the missing index in the slave database and
everything started working great.  Thanks, Casey!

Luke

-----Original Message-----
From: Casey Duncan
Sent: Monday, October 16, 2006 4:54 PM
To: Luke Morehead

Try turning up the log verbosity:

$ slon -d 4 rep_atr_prod "dbname=atr_test user=postgres"

And post the output. I assume the _rep_atr_prod schema exists on this  
db?

-Casey

On Oct 16, 2006, at 12:40 PM, Luke Morehead wrote:

> I've pasted an initial E-mail conversation below.  The actual slon
> launch command and output is as follows:
>
> postgres at dali:/$ slon rep_atr_prod "dbname=atr_test user=postgres"
> 2006-10-12 08:49:16 EDT CONFIG main: slon version 1.1.0 starting up
> 2006-10-12 08:49:16 EDT FATAL  main: Node is not initialized properly
> 2006-10-12 08:49:16 EDT DEBUG2 slon: exit(-1)
>
> Is it possible to get any more specific info about where and how the
> node is not initialized?  Could there be a problem with the
> _rep_atr_prod schema even though slon is running fine on the master  
> host
> and database?  If so is there a script that can actually remove all  
> the
> slony info (rowID columns, triggers, operators, etc.) from the master
> database so I can start over from scratch, or is that a manual  
> process?
>
> Thanks very much,
> Luke
>
> -----Original Message-----
> From: Christopher Browne
> Sent: Wednesday, October 11, 2006 1:48 PM
> To: Luke Morehead
> Subject: Re: Slony help
>
> On 10/10/06, Luke Morehead wrote:
>> Chris,
>>
>> I ran the slony1_extract_schema.sh script (using a test database) and
> used the output to populate a fresh slave database.  I then ran
> slonik_init_cluster.  After this slon_start ran fine for the master  
> node
> but failed for the slave node with the error, "FATAL  main: Node is  
> not
> initialized properly".
>>
>> The buzz online seems to indicate a problem with the (clustername)
> schema.  However the schema looks fine in the slave database.  What
> could this be?
>
> That message comes from the following code...
> 	/*
> 	 * Get our local node ID
> 	 */
> 	rtcfg_nodeid = db_getLocalNodeId(startup_conn);
> 	if (rtcfg_nodeid < 0)
> 	{
> 		slon_log(SLON_FATAL, "main: Node is not initialized
> properly\n");
> 		slon_exit(-1);
> 	}
>
> If there isn't a rtcfg_nodeid value, then that means the call to
> db_getLocalNodeId() failed.
>
> 	/*
> 	 * Select the last_value from the sl_local_node_id sequence
> 	 */
> 	snprintf(query, 1024, "select last_value::int4 from
> %s.sl_local_node_id",
> 			 rtcfg_namespace);
> 	res = PQexec(conn, query);
> 	if (PQresultStatus(res) != PGRES_TUPLES_OK)
> 	{
> 		slon_log(SLON_ERROR,
> 				 "cannot get sl_local_node_id - %s",
> 				 PQresultErrorMessage(res));
> 		PQclear(res);
> 		return -1;
> 	}
> 	if (PQntuples(res) != 1)
> 	{
> 		slon_log(SLON_ERROR,
> 				 "query '%s' returned %d rows (expected
> 1)\n",
> 				 query, PQntuples(res));
> 		PQclear(res);
> 		return -1;
> 	}
>
> The above are the expected failure cases, which would return negative
> numbers.  The slon should have had one of those messages.
>
> What I'm expecting is that perhaps the conn info to the subscriber
> node isn't right.
>
> So you might compare:
> a) What you type in to get a psql session to the subscriber node
> to
> b) The arguments passed to slon to get it to connect.
>
> You didn't say how you're launching slon processes; I'm suspicious
> that THAT is what is breaking down.
>
> I'd prefer to address the issue on the mailing list, if possible.
> _______________________________________________
> Slony1-general mailing list
> Slony1-general at gborg.postgresql.org
> http://gborg.postgresql.org/mailman/listinfo/slony1-general