>I can pretty much guarantee that whatever you're building won't work.
>There's no such thing as simple multimaster replication.
I know, but I don't tend to build something generally usable. I suppose to create it just for this small project. Very, very specific and tied to concrete tables, functions and so on.
>You'd be much better off to look into adopting one of the existing
>solutions that someone else has already written and debugged.
I've already tried to look into any similiar solution, but only one suitable is pgcluster. Which is synchronous and in my projects are many separated db's which must contain same content but, are not 100% connected via internet.
DF