Re: Loss of cluster status

"Christopher Kings-Lynne" <chriskl(at)familyhealth(dot)com(dot)au> writes:
>> A CLUSTER command issued just after table creation, while it's still
>> empty, would be cheap ... but we don't put the index in place until
>> we've loaded the data, do we? Darn.

> Maybe we should issue it after the CREATE INDEX and ADD CONSTRAINT has
> occurred and just bite it.

The real problem I think is that we've confused the notion of setting a
policy for CLUSTER (ie, marking the preferred thing to cluster on) with
the notion of actually doing a CLUSTER. Perhaps we need an ALTER
command that says "this is what to cluster on" without actually doing
it.

> Other potential problem - ALTER TABLE / SET STORAGE ?

Yeah, pg_dump should be dumping that too, probably.

regards, tom lane