On Thu, 2004-10-21 at 11:55 -0400, Jan Wieck wrote:

> You should have RTFM before being that nasty to your database - Slony 
> provides a DDL script execution mechanism especially to do that for you 
> in a safe way. See
> 
> http://gborg.postgresql.org/project/slony1/genpage.php?slonik_commands#stmt_ddl_script
> 
> The best way would be to rebuild your Slony cluster, now that you have 
> done it wrong already. And it would be the only way in case the table 
> got already updated with that new attributes in place, because those 
> items did not get replicated.
> 
> If you are 100% sure that this hasn't happened and that the table is 
> still in sync and the same attributes have been added in the same order 
> on all subscribers, you could execute a sequence of
> 
>      start transaction;
>      select _<clustername>.alterTableRestore(<table-id>);
>      select _<clustername>.alterTableForReplication(<table-id>);
>      commit;
> 
> on all nodes.
> 
> 
> Jan
> 

Ah well,

I wanted to avoid rebuilding the cluster as it took rather too long the
first time, and I figured if it did break I could just rebuild.

As it happens, it is actually replicating correctly without any extra
commands. Maybe it shouldn't...

-- 
Peter Nelson                                       http://www.bucks.net
Systems Administrator                              Tel: +44 1296 432486
Bucks Net                                          Fax: +44 1296 398795