On Tue, 2004-11-09 at 13:27 -0500, Rod Taylor wrote:
> Should have said this earlier. pg_listener has a single page as reported
> by vacuum.

Replying to myself again.  I just noticed the index on log_1 has
changed, although I must have skipped over this while reading the newer
release notes.

create index sl_log_1_idx1 on @NAMESPACE at .sl_log_1
	(log_origin, log_xid @NAMESPACE at .xxid_ops, log_actionseq);

> On Tue, 2004-11-09 at 13:26 -0500, Rod Taylor wrote:
> > I've got slony (1.0.1) modified to run on PostgreSQL 7.2, and corrected
> > the LOG cursor to use an index (same plan as 7.3 and later).
> > 
> > The FETCH 100 FROM LOG; statement seems to take about 10 to 15 minutes,
> > gets run about 10 times, then a new cursor is defined and repeats.
> > 
> > Right now it's behind by about 50M tuples in log_1.
> > 
> > It appears it scans the entire table (of 50M entries) in order to find
> > the 1000 or so it wants for the current SYNC.
> > 
> > Is there a way to get it to catch up?
> 
-- 
Rod Taylor <rbt [at] rbt [dot] ca>

Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL
PGP Key: http://www.rbt.ca/signature.asc