Re: BUG #17871: JIT during postgresql_fdw remote_estimates EXPLAIN have very negatively effect on planning time

On Tue, Mar 28, 2023 at 5:13 PM David Rowley <dgrowleyml@gmail.com> wrote:

On Tue, 28 Mar 2023 at 17:54, Maxim Boguk <maxim.boguk@gmail.com> wrote:
> During an attempt to create a minimal test case I discovered an effect for which I have no explanation:
> (all tests with jit=on):
>
> Run EXPLAIN on the single partition, so far expected results...
> explain select * from interview_review_info_archive_p028;
>                                           QUERY PLAN
> -----------------------------------------------------------------------------------------------
>  Seq Scan on interview_review_info_archive_p028  (cost=0.00..361819.08 rows=34546848 width=40)
>  JIT:
>    Functions: 2
>    Options: Inlining true, Optimization true, Expressions true, Deforming true

It could be something like a dropped column forcing projection of the
SELECT *.  You might want to try without SELECT * using some subset of
columns instead.

Check if columns have been dropped with:

select attnum,attname from pg_Attribute where
attrelid='interview_review_info_archive_p028'::Regclass;

David

Hi David,

You 100% right

negotiation_chat_archive=# explain select topic_id,review_id,move_to_invitation_state_time,no_interview_reply_time,review_suggestion_chat_message_creation_time from interview_review_info_archive_p028;
                                          QUERY PLAN                                          
-----------------------------------------------------------------------------------------------
 Seq Scan on interview_review_info_archive_p028  (cost=0.00..361921.00 rows=34563080 width=40)

Exceptionally curious case when select * not equivalent select list_all_columns from table...

Thank you.