Re: [PATCH] Add features to pg_stat_statements

Julien Rouhaud <rjuju123@gmail.com>

From: Julien Rouhaud <rjuju123@gmail.com>
To: Katsuragi Yuta <btkatsuragiyu@oss.nttdata.com>
Cc: pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2020-09-23T07:01:12Z
Lists: pgsql-hackers
On Wed, Sep 23, 2020 at 2:48 PM Katsuragi Yuta
<btkatsuragiyu@oss.nttdata.com> wrote:
>
> On 2020-09-18 18:49, Julien Rouhaud wrote:
> > Did you consider also adding the cumulated number of
> > evicted entries?  This could be useful to know how to configure
> > pg_stat_statements.max.
>
> Thank you for your comments!
> I overlooked the cumulated number of evicted entries.
> This statistic looks important. But, I am not sure
> if I should add this statistic to a view.
> This is because I am not sure how to utilize the cumulated
> number of evicted entries for configuring pg_stat_statements.max.

You're right, as the number of evicted entries isn't depending on the
number of entries that wouls been needed to contain the entirely
workload.

> Not only providing a view but also logging evictions
> along with the number of evicted entries might be a choice.
> This idea is from legrand legrand [1].

+1.  I'm wondering if logging each evicted entry, with its queryid,
would help to estimate the actual size of the normalised queries set.



Commits

  1. pg_stat_statements: Track number of times pgss entries were deallocated.