Re: Question of Parallel Hash Join on TPC-H Benchmark

On 13/10/2024 15:57, Ba Jinsheng wrote:
> Thanks for looking into it!
> We are working a research project to automatically look for such cases.
> If it matters for PostgreSQL, I will report more such cases.
I made an attempt last year [1]. However, at least in production, it 
provided too many false-positive cases to trigger automatic 
re-optimisation [2] on each poorly estimated query plan.
I wonder if you have any progress in that area. Living nearby Singapore 
(Chon Buri, Thailand) I would like to visit meetup related to this 
problem, if you organised it.
Of course, reporting performance issues is always beneficial.

[1] https://github.com/danolivo/pg_track_optimizer
[2] https://postgrespro.com/docs/enterprise/16/realtime-query-replanning

-- 
regards, Andrei Lepikhov