Kyle Hanson wrote:

>Thanks for all the info...  I'm a little nervous about starting over from
>scratch as I've never installed and configured slony, it was done by our
>previous DBA and I have to pick up the broken pieces.
>
>How much work is it to start from scratch, my largest concern is
>affecting(breaking) our production postgres database?
>  
>
You could submit one or two UNINSTALL NODE scripts, which would drop out
replication configuration, cleaning things out, and then rebuild the nodes.

The notable risk is that the C/SPI trigger functions disappear out from
under things; you may need to recycle connections to the DB, otherwise
updates to the replicated tables will all start failing.  This is a well
known issue; if you restart the DB, that'll certainly "clear things up,"
forcing apps to start up new connections.

It's not *all* that scary to drop and rebuild replication.

You probably want to grab a dump of the table sl_table/sl_sequence;
that'll give a good idea as to what objects were being replicated.  With
only two nodes, that's the main complexity that arises...