"Ed L." <pgsql (AT) bluepolka (DOT) net> writes:
I don't think so. Since sequences are inherently not transactional,
it seems wrong to fire a transactional behavior as a side effect of
nextval(). What exactly are you expecting the trigger to do, and
how is it going to do it in a way that won't get rolled back if the
calling transaction fails later?

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match

On Sunday October 31 2004 8:25, Tom Lane wrote:
I don't know how to implement a solution to my problem. What seems evident
is that polling hundreds of sequence objects to tell if they've changed is
an unscalable aspect of current async replication solutions. My only
thought in raising the idea is that notification of a change to a sequence
value is highly preferrable over performance-intensive polling. Whether
that's done via triggers inside a transaction, or some other mechanism, I
don't care much as long as polling is not required.

I wonder if these async replication triggered procedures might be able to
detect changes to their related sequence objects, and issue a NOTIFY for
the updated sequence? The replicator could be listening on the NOTIFY and
update accordingly. The NOTIFY wouldn't go out until/unless the
transaction completed, and would obviate the need for sequence polling.
That'd save us 100-200 queries/second if it were possible if we're
attempting to replicate with a very short (1s) sync interval.

Ed


---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?

http://www.postgresql.org/docs/faqs/FAQ.html