Re: Columns correlation and adaptive query optimization

From: Yugo NAGATA <nagata(at)sraoss(dot)co(dot)jp>
To: Konstantin Knizhnik <k(dot)knizhnik(at)postgrespro(dot)ru>
Cc: Tomas Vondra <tomas(dot)vondra(at)enterprisedb(dot)com>, Rafia Sabih <rafia(dot)pghackers(at)gmail(dot)com>, David Steele <david(at)pgmasters(dot)net>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Columns correlation and adaptive query optimization
Date: 2021-03-22 02:29:25
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgsql-hackers

On Fri, 19 Mar 2021 19:58:27 +0300
Konstantin Knizhnik <k(dot)knizhnik(at)postgrespro(dot)ru> wrote:

> On 19.03.2021 12:17, Yugo NAGATA wrote:
> > On Wed, 10 Mar 2021 03:00:25 +0100
> > Tomas Vondra <tomas(dot)vondra(at)enterprisedb(dot)com> wrote:
> >
> >> What is being proposed here - an extension suggesting which statistics
> >> to create (and possibly creating them automatically) is certainly
> >> useful, but I'm not sure I'd call it "adaptive query optimization". I
> >> think "adaptive" means the extension directly modifies the estimates
> >> based on past executions. So I propose calling it maybe "statistics
> >> advisor" or something like that.
> > I am also agree with the idea to implement this feature as a new
> > extension for statistics advisor.
> >
> >> BTW Why is "qual" in
> >>
> >> static void
> >> AddMultiColumnStatisticsForQual(void* qual, ExplainState *es)
> >>
> >> declared as "void *"? Shouldn't that be "List *"?
> > When I tested this extension using TPC-H queries, it raised segmentation
> > fault in this function. I think the cause would be around this argument.
> >
> > Regards,
> > Yugo Nagata
> >
> Attached please find new version of the patch with
> AddMultiColumnStatisticsForQual parameter type fix and one more fix
> related with handling synthetic attributes.
> I can not reproduce the crash on TPC-H queries, so if the problem
> persists, can you please send me stack trace and may be some other
> information helping to understand the reason of SIGSEGV?

I also could not reproduce the segfault. I don't know why I observed it,
but it may be because I missed something when installing. Sorry for
annoying you.

Instead, I observed "ERROR: cache lookup failed for attribute 6 of
relation xxxx" in v8 patch, but this was fixed in v9 patch.

Yugo Nagata

Yugo NAGATA <nagata(at)sraoss(dot)co(dot)jp>

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Thomas Munro 2021-03-22 02:29:51 Re: Add client connection check during the execution of the query
Previous Message Amit Kapila 2021-03-22 02:27:15 Re: replication cleanup code incorrect way to use of HTAB HASH_REMOVE ?