glynastill@yahoo.co.uk (Glyn Astill) writes:
> It it possible to get a system that does syncronous replication and
> also allows slaves to catch up if they're down for a period of time
> like you can with asyncronous?
Well, a "modal approach" is possible - that's what Postgres-R tries to
do.
Of course, once you drop into a mode that "allows slaves to catch up,"
then you have given up on synchronicity, and have fallen back to
asynchronous replication.
If you systematically have a way to do that, then you no longer have a
replication system that can honestly be called "synchronous."
If it's *truly* synchronous, then when nodes fall over, the system
MUST stop accepting transactions.
--
output = reverse("ofni.secnanifxunil" "@" "enworbbc")
http://www3.sympatico.ca/cbbrowne/linux.html
Signs of a Klingon Programmer - 2. "Specifications are for the weak
and timid!"