From: | Bruce Momjian <bruce(at)momjian(dot)us> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | Julien Rouhaud <rjuju123(at)gmail(dot)com>, Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org>, Robert Haas <robertmhaas(at)gmail(dot)com>, Michael Paquier <michael(at)paquier(dot)xyz>, torikoshia <torikoshia(at)oss(dot)nttdata(dot)com>, Atsushi Torikoshi <atorik(at)gmail(dot)com>, Tatsuro Yamada <tatsuro(dot)yamada(dot)tf(at)nttcom(dot)co(dot)jp>, Tomas Vondra <tomas(dot)vondra(at)2ndquadrant(dot)com>, Evgeny Efimkin <efimkin(at)yandex-team(dot)ru>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
Subject: | Re: Feature improvement: can we add queryId for pg_catalog.pg_stat_activity view? |
Date: | 2021-03-17 16:13:23 |
Message-ID: | 20210317161323.GC16650@momjian.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Wed, Mar 17, 2021 at 12:01:38PM -0400, Tom Lane wrote:
> Bruce Momjian <bruce(at)momjian(dot)us> writes:
> > On Wed, Mar 17, 2021 at 11:28:38AM -0400, Tom Lane wrote:
> >> I still say that it's a serious mistake to sanctify a query ID calculation
> >> method that was designed only for pg_stat_statement's needs as the one
> >> true way to do it. But that's what exposing it in a core view would do.
>
> > OK, I am fine with creating a new method, and maybe having
> > pg_stat_statements use it. Is that the direction we should be going in?
>
> The point is that we've understood Query.queryId as something that
> different extensions might calculate differently for their own needs.
> In particular it's easy to imagine extensions that want an ID that is
> less fuzzy than what pg_stat_statements wants. We never had a plan for
> how two such extensions could co-exist, but at least it was possible
> to use one if you didn't use another. If this gets moved into core
> then there will basically be only one way that anyone can do it.
Well, the patch docs say:
Enables or disables in core query identifier computation.arameter. The
<xref linkend="pgstatstatements"/> extension requires a query
--> identifier to be computed. Note that an external module can
--> alternatively be used if the in core query identifier computation
specification doesn't suit your need. In this case, in core
computation must be disabled. The default is <literal>off</literal>.
> Maybe what we need is a design for allowing more than one query ID.
>
> > I do think we need _some_ method in core if we are going to be exposing
> > this value in pg_stat_activity and log_line_prefix.
>
> I'm basically objecting to the conclusion that we should do either
> of those. There is no way around the fact that it will break every
> user of Query.queryId other than pg_stat_statements, unless they
> are okay with whatever definition pg_stat_statements is using (which
> is a moving target BTW).
I thought the above doc patch feature avoided this problem because an
extension can override the build-in query id.
--
Bruce Momjian <bruce(at)momjian(dot)us> https://momjian.us
EDB https://enterprisedb.com
If only the physical world exists, free will is an illusion.
From | Date | Subject | |
---|---|---|---|
Next Message | Pavel Stehule | 2021-03-17 16:16:50 | Re: Feature improvement: can we add queryId for pg_catalog.pg_stat_activity view? |
Previous Message | Tomas Vondra | 2021-03-17 16:12:17 | Re: WIP: WAL prefetch (another approach) |