From: Tom Lane <tgl@sss.pgh.pa.us>
To: Michael Paquier <michael.paquier@gmail.com>
cc: Dmitry Shalashov <skaurus@gmail.com>,
        "pgsql-performance@postgresql.org" <pgsql-performance@postgresql.org>
Subject: Re: Query became very slow after 9.6 -> 10 upgrade
In-reply-to: 
 <CAB7nPqQkVkEmQ29bWE8i_D7Sg1zEx9YptEuKR7_N0o7p_K7BNQ@mail.gmail.com>
References: 
 <CAKPeCUGAeHgoh5O=SvcQxREVkoX7UdeJUMj1F5=aBNvoTa+O8w@mail.gmail.com>
 <13426.1511363963@sss.pgh.pa.us>
 <CAKPeCUGkCcp37oa3n=-9EDAujx_f190wFdmbfy_K4OF5H62JyQ@mail.gmail.com>
 <9755.1511392027@sss.pgh.pa.us>
 <CAKPeCUEy6HMm=Kn=V82xjycZf9KYnqzPLVQ8ngjoLXvO97zx4g@mail.gmail.com>
 <5549.1511456440@sss.pgh.pa.us>
 <CAKPeCUF_k9dLot=Lb0anZZQZmjoS72ScSWr6oj9GCj_C3+Ciaw@mail.gmail.com>
 <28356.1511541575@sss.pgh.pa.us>
 <CAKPeCUH72oDs4BqkcXYaaGFjygykdifn0xsPF-YPuVuOP1FcaA@mail.gmail.com>
 <CAB7nPqQkVkEmQ29bWE8i_D7Sg1zEx9YptEuKR7_N0o7p_K7BNQ@mail.gmail.com>
Comments: In-reply-to Michael Paquier <michael.paquier@gmail.com>
	message dated "Sat, 25 Nov 2017 21:13:42 +0900"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1664.1511624534.1@sss.pgh.pa.us>
Date: Sat, 25 Nov 2017 10:42:14 -0500
Message-ID: <1665.1511624534@sss.pgh.pa.us>
Precedence: bulk

Michael Paquier <michael.paquier@gmail.com> writes:
> On Sat, Nov 25, 2017 at 8:54 PM, Dmitry Shalashov <skaurus@gmail.com> wrote:
>> Is it completely safe to use manually patched version in production?

> Patching upstream PostgreSQL to fix a critical bug is something that
> can of course be done. And to reach a state where you think something
> is safe to use in production first be sure to test it thoroughly on a
> stage instance. The author is also working on Postgres for 20 years,
> so this gives some insurance.

It's not like there's some magic dust that we sprinkle on the code at
release time ;-).  If there's a problem with that patch, it's much more
likely that you'd discover it through field testing than that we would
notice it during development (we missed the original problem after all).
So you can do that field testing now, or after 10.2 comes out.  The
former seems preferable, if you are comfortable with building a patched
copy at all.  I don't know what your normal source of Postgres executables
is, but all the common packaging technologies make it pretty easy to
rebuild a package from source with patch(es) added.  Modifying your
vendor's SRPM (or equivalent concept if you're not on Red Hat) is a
good skill to have.

			regards, tom lane