Re: unstable query plan on pg 16,17,18

On Thu, 2026-02-26 at 17:15 +0100, Attila Soki wrote:
> Increasing the statistics of dp_end_dat did not helped. With statistic 1000
> I was not able to get a good plan without setting join_collapse_limit=7

Reducing "join_collapse_limit" dumbs down the optimizer, so you are getting a good plan
by accident. I mean, you can try to rewrite the query so that the tables are written
in the order in which they should be joined in the good plan, then set "join_collapse_limit"
to 1. That may be a solution if you cannot find a better one.

Still, the bad estimate that I indicated in [1] is worrysome, and I don't quite
understand it. Could you show the result of the simplified query that I suggested?

EXPLAIN (ANALYZE, BUFFERS, SETTINGS)
SELECT * FROM schema1.table_k AS kal
WHERE dp_end_dat < current_date;

If I were you, I'd focus on getting PostgreSQL to estimate that correctly.

Yours,
Laurenz Albe

[1]: https://postgr.es/m/b2e372392b8a022da81b95b7c823a5729d7fd70f.camel%40cybertec.at