Re: Making clausesel.c Smarter

From: Robert Haas <robertmhaas(at)gmail(dot)com>
To: David Rowley <david(dot)rowley(at)2ndquadrant(dot)com>
Cc: PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Making clausesel.c Smarter
Date: 2017-02-26 06:41:41
Message-ID: CA+Tgmoagx4vke2Eg9JR6QMSOWTYpe6nUkBLws55=4zabNXxhCg@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Fri, Feb 24, 2017 at 3:30 PM, David Rowley
<david(dot)rowley(at)2ndquadrant(dot)com> wrote:
> It would be good to improve the situation here in the back branches
> too, but I'm thinking that the attached might be a little invasive for
> that?

My experience has been that customers - at least EnterpriseDB
customers - do not appreciate plan changes when they upgrade to a new
minor release. Most people have production installations that are
basically working; if not, they wouldn't be in production. Even if
they're getting a good plan only by some happy accident, they're still
getting it, and a change can cause a good plan to flop over into a bad
plan, which can easily turn into a service outage. The people who had
a bad plan and get flipped to a good plan will be happy once they
realize what has changed, of course, but that's not really enough to
make up from the panicked calls from customers whose stuff falls over
when they try to apply the critical security update.

I think the basic think you are trying to accomplish is sensible,
though. I haven't reviewed the patch.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Robert Haas 2017-02-26 06:51:53 Re: GUC for cleanup indexes threshold.
Previous Message Robert Haas 2017-02-26 06:31:59 Re: Proposal : Parallel Merge Join