Re: BUG #16109: Postgres planning time is high across version - 10.6vs 10.10 - Mailing list pgsql-bugs

From Mukesh Chhatani
Subject Re: BUG #16109: Postgres planning time is high across version - 10.6vs 10.10
Date
Msg-id CACnmOYJQVHhuaYe70z0yFr2rKy8jqe=8Fx92c7-HXShQOEy8qw@mail.gmail.com
Whole thread Raw
In response to Re: BUG #16109: Postgres planning time is high across version - 10.6vs 10.10  (Andres Freund <andres@anarazel.de>)
List pgsql-bugs
Thanks for response.

I analyzed after changing default_statistics_target but no improvement, are there any recommendations around this parameter because as far as I have seen increasing this parameter in the past lets optimizer choose better plans and has never caused me this problem related to high planning time, but I am open to suggestions since every problem is a new problem.

I know changing partitioned to unpartitioned and then changing columns types makes no sense in terms of resolving the issue but that is what is working now.

I will go ahead and change the parameter - default_statistics_target to 100 and analyze whole database and then wait for couple of hours and then run my queries. Let me know if this approach is good.

Regards,
Mukesh

On Wed, Nov 13, 2019 at 12:42 PM Andres Freund <andres@anarazel.de> wrote:
Hi,

On 2019-11-13 11:37:30 -0700, Mukesh Chhatani wrote:
> FYI.. default_statistics_target was set to 10000 but I changed it 100 and
> even to 1000 and still planning time was high.

Note that you'd need to ANALYZE the involved tables before that change
actually would effect planning time.


> Update I was able to resolve the problem by changing - partitioned tables to
> single table and changing data type of 2 columns used in the joins from
> varchar to varchar(50).

That's not going to be the fix. There's no efficiency difference between
those. It's more likely that, that the different statistics target would
have taken effect after the alter table etc.

- Andres

pgsql-bugs by date:

Previous
From: Andres Freund
Date:
Subject: Re: BUG #16104: Invalid DSA Memory Alloc Request in Parallel Hash
Next
From: Tom Lane
Date:
Subject: Re: [BUG] Uninitializaed configOut.leafType used.