Hi,
Tom Lane wrote:
I think you need to see about getting this rowcount estimate to be more
accurate:
- Index Scan using idx_link_1 on link
(cost=0.00..680.51 rows=13477 width=26) (actual time=5.707..12.043
rows=126 loops=1)
Hi,
we experience some strange performance problems, we've already found a
workaround for us, but are curious if it's a known problem of the optimizer.
Tested with the following Postgres Version: 8.2.15 and 8.3.9
AUTOVACUUM is enabled, explicit VACUUM and REINDEX both tables and the
whole DB.