Hello,

It is my intention to create a hybrid Windows/Linux Slony cluster. The
master will run MS windows, and each of the slaves will run Linux. The
master currently runs Postgres 8.3.1 with the latest version of Slony
(as supplied by the "stack builder") , and the slaves run 8.2.4 with a
slightly earlier version of Slony. I believe that configuring Slony is
difficult, with many variables to consider, so I think it wise to
approach this absolutely methodically and circumspectly.

According to the "best practices" section of the slony-I
documentation, "Running all of the slon daemons on a central server
for each network has proven preferable". It seems like a good idea to
manage all slon processes from a central location, so I'll aim to do
this. This cluster will be part of an application that will hopefully
be distributed widely, so ease of installation/ maintenance is
particularly important.

My questions are:

1. As I've said, my current set up has slightly different versions of
slony-I and PostgreSQL. Will this be problematic? Will it prove
essential, or prudent to harmonise versions?

2. According to the "Slon daemons" section of the Slony documentation,
"On Windows™ when running as a service things are slightly different.
One slon service is installed, and a separate configuration file
registered for each node to be serviced by that machine. The main
service then manages the individual slons itself". Would it be
possible to control the entire cluster through this one Windows/master
based slon process, with no trace of Slony on the slaves other then
the triggers and so on that the Slony configuration script leaves in
the node's database?

Regards,
Peter Geoghegan