Franco Pieressa <fpieressa at gmail.com> writes:
> Hi!? I have a simple question, how can I do to know which is the
> local node ID of a DB that is members of a slony cluster?? Are there
> some SQL to find it?? I'm playing with failovers, and I want to know
> if it's running OK...

It is stored in a sequence in the special schema, in sl_local_node_id.

The function (again, in the special Slony-I schema), getlocalnodeid()
is the "portable" way you're intended to get at the value.

Thus, if the cluster is called T1, then you could do...

select "_T1".getlocalnodeid('_T1');

> And now, the difficult question :-) I read the article in
> http://www.linuxjournal.com/article/7834, and I test the script
> promote.sh.? After running it, clear the orginal master db and
> replicate the schema from the original slave, how can I do the
> switchback, witch a slonik script? May be with a "move set" command,
> but how?

The slonik script would involve two steps:

1.  Lock the replication set

  lock set (id=1);

2.  Move the set origin

  move set (id=1, old origin = 2, new origin = 1);

This would all go into one slonik script consisting of preamble and
contents, something like:

----------------------------------------------------
cluster name = 'T1';
node 1 admin conninfo = 'some dsn';
node 2 admin conninfo = 'some dsn';
lock set (id=1);
move set (id=1, old origin = 2, new origin = 1);
----------------------------------------------------

You'll obviously have to rename nodes, connection info, and the
cluster to correspond with your system's configuration.
-- 
(reverse (concatenate 'string "ofni.sailifa.ac" "@" "enworbbc"))
<http://dba2.int.libertyrms.com/>
Christopher Browne
(416) 673-4124 (land)