Brad Nicholson <bnichols at ca.afilias.info> writes:
> I wonder though if there will be a limiting factor in Slony itself
> though.  I recall that node numbers had some sort of artificial ceiling
> around 9000 or so, due to how they are used internally. 

The trouble with node numbers is that there is an in-memory array set
up storing node information, because that needs to be referenced
frequently.

There isn't any need to keep sequence values in memory "en masse," so
the query that pulls sequence updates does so simply via a query.

There is no array "denominated by" sequence numbers, so this oughtn't
be an issue.

It's a good thing to consider, but it shouldn't be a problem to have
sequences numbered "quite a bit higher."  Actually, I'll see about
changing the sequence test to validate all of this ;-).

[... time passes ...]

I've got a sample running right now that generates 2000-odd sequences,
with fairly large sequence ID values, and that seems to be working
AOK.

Thus:
  sequences seq400000 thru seq402000

with IDs
             23400000 thru 23402000

The test runs for a mighty long time, basically because it takes
*some* time to add each object to replication, and the test adds
thousands of objects.  But within that stipulation, all seems to be
working fine.
-- 
(format nil "~S@~S" "cbbrowne" "ca.afilias.info")
<http://dba2.int.libertyrms.com/>
Christopher Browne
(416) 673-4124 (land)
"Bother,"  said Pooh,  "Eeyore, ready  two photon  torpedoes  and lock
phasers on the Heffalump, Piglet, meet me in transporter room three"