On Wed, Feb 25, 2009 at 06:56:20AM +1300, Steve Holdoway wrote:
> 
> It works fine if I test it on an un replicated table, performing a bulk load either from a copy or multiple insert statements, which has sort of led me to dismiss that probability, and concentrate on environmental differences. If I remove the offending unique index, it fails on the next similar insert into a remote table, but there is still no data in the initial one...
> 

Well, without looking at the actual case, I'm not sure what else to
suggest.  But I'd bet a pretty good lunch that something is causing
the trigger to fire twice for the same data, which is what's causing
your duplicate problem.

A

-- 
Andrew Sullivan
ajs at crankycanuck.ca