Have you vacuum analyzed one of the databases and not the other?
The optimizer in the first case thinks 391 rows will match and in the
second something like 50000. That's what's probably driving the move
toward sequence scan in the second case.
Stephan Szabo
sszabo@bigpanda.com
On 12 Oct 2000, Martin Tomes wrote:
>
> I have Redhat 6.2 installed on two very different machines. One is a Dual
> processor xeon with 512Mb RAM and RAID disks, the other is for development and
> is a PII-266 with an IDE disk. I have PostgreSQL 7.0 installed on both from
> the same RPMS and two identical databases (I used pg_dump on the production
> machine and psql -e to read it into the development machine). However there
> is one query which uses an index on the Big Machine but not on the Small
> Machine.
>
