Re: query_id: jumble names of temp tables for better pg_stat_statement UX

On Wed, 16 Jul 2025 at 01:39, Michael Paquier <michael@paquier.xyz> wrote:

>
> > create schema s1;
> > create table s1.t as select id from generate_series(1, 10) as id;
> > create schema s2;
> > create table s1.t as select id from generate_series(1, 1000000) as id;
>
> I suspect that you mean s2.t and not s1.t here.
>

Yes.


> Yes, we had this argument upthread, and it is still possible to
> differentiate both cases by using a different alias in the FROM
> clause, as of:
> select count(id) from s1.t as t1;
> select count(id) from s2.t as t2;
>
> The new behavior where we do not need to worry about temporary tables,
> which is not that uncommon because some workloads like using these for
> JOIN patterns as a "temporary" anchor in a session, has more benefits
> IMO, particularly more if the connections have a rather higher
> turnover.


Yes, I've seen this argument and know that aliases will make these queries
look different.
However, we regularly hear from many different customers that they *don't
control queries* sent by application or *can't modify these queries*.
Such kinds of workloads are also not that uncommon and this change makes it
impossible to monitor them.

I would somewhat understand when a table in the query is used without
specifying schema and such queries are merged together:
s1: SET search_path s1; select count(*) from t;
s2: SET search_path s2; select count(*) from t;

But, even this case doesn't feel right, because these tables are still
different and therefore queries.

Regards,
--
Alexander Kukushkin