Nicholas Haggin wrote:
>     Greetings, all. My apologies if this problem has appeared
> previously and has already been solved; I've searched through the
> howtos, the admin guide, and the mailing list with no success.
>     I'm attempting to set up a replication cluster of four nodes,
> replicating two databases. Both sides are running Solaris 9 with a
> freshly-compiled copy of Postgres 8.1.3/Slony 1.1.5. Authentication
> in our environment requires passwords, but my .pgpass file is set up
> with the appropriate password for my replication user. The relevant
> snippets of my slonik script are presented below. foo_piper is the
> eventual master node; foo_swift is the eventual slave.
>
> DEFINE    documentation_piper        5;
> DEFINE    documentation_swift        6;
> ...
> DEFINE    ocblog_piper            17;
> DEFINE    ocblog_swift            18;
> ...
> CLUSTER = opcenter;
> ...
> NODE @documentation_piper ADMIN CONNINFO = 'dbname=documentation
> host=piper.cites.uiuc.edu user=ocdb';
> NODE @documentation_swift ADMIN CONNINFO = 'dbname=documentation
> host=swift.cites.uiuc.edu user=ocdb';
> ...
> NODE @ocblog_piper ADMIN CONNINFO = 'dbname=ocblog
> host=piper.cites.uiuc.edu user=ocdb';
> NODE @ocblog_swift ADMIN CONNINFO = 'dbname=ocblog
> host=swift.cites.uiuc.edu user=ocdb';
> ...
> INIT CLUSTER (id = @documentation_piper, comment = 'documentation on
> piper');
> STORE NODE (id = @documentation_swift, comment = 'documentation on
> swift');
> STORE NODE (id = @ocblog_piper, comment = 'ocblog on piper');
> STORE NODE (id = @ocblog_swift, comment = 'ocblog on swift');
>
>     The INIT CLUSTER runs just fine, but when slonik tries to run the
> STORE NODE for documentation_swift, I receive the following output:
>
> opcenter-initcluster.slon:13: PGRES_FATAL_ERROR select
> "_opcenter".getLocalNodeId('_opcenter'); - ERROR:  schema
> "_opcenter" does not exist
> opcenter-initcluster.slon:13: database specified in
> opcenter-preamble.slon:46 reports no_id -1
>
>     Line 13 of opcenter-initcluster.slon is the STORE NODE referred to
> earlier. Strangely enough, line 46 of opcenter-preamble.slon is the
> CLUSTER, and not the ADMIN CONNINFO for documentation_swift.
>     Can anyone shed some light on this, or slap me really hard if
> there's somewhere I ought to be looking that I haven't checked yet?
The problem is that the STORE NODE is trying to access node #1 in order
to draw out the existing configuration.

http://cbbrowne.com/info/stmtstorenode.html

You haven't got a node #1.

The three STORE NODE statements should be revised to...

STORE NODE (id = @whatever_id, comment='some comment', event
node=@documentation_piper);

Change @whatever_id and the comment as needed :-).