From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andy Fan <zhihui.fan1213@gmail.com>
cc: Robert Haas <robertmhaas@gmail.com>,
        Amit Langote <amitlangote09@gmail.com>,
        Justin Pryzby <pryzby@telsasoft.com>,
        "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Subject: Re: FailedAssertion("pd_idx == pinfo->nparts",
 File: "execPartition.c", Line: 1689)
In-reply-to: 
 <CAKU4AWpEMxre9zDs-Odx7PnYd7tW3MiAyYBaP=9QrjuD7ps=MQ@mail.gmail.com>
References: <20200802181131.GA27754@telsasoft.com>
 <CA+TgmoawZkbuvP5BDBNNVp18F4tuwaZtqKUFxtH61QX7CiAV=Q@mail.gmail.com>
 <20200803161133.GA21372@telsasoft.com>
 <CA+HiwqETTgTh5SG9bcAdYRahEeTgpPHy+Xi=b93icoTX7j2EFg@mail.gmail.com>
 <2528807.1596648630@sss.pgh.pa.us>
 <CA+TgmoYMqqqHaOpfMYs8R6VJekykfRABCD=o4ovN2eUz-efYMA@mail.gmail.com>
 <2550497.1596651719@sss.pgh.pa.us>
 <CAKU4AWpEMxre9zDs-Odx7PnYd7tW3MiAyYBaP=9QrjuD7ps=MQ@mail.gmail.com>
Comments: In-reply-to Andy Fan <zhihui.fan1213@gmail.com>
	message dated "Thu, 06 Aug 2020 11:49:36 +0800"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <2598265.1596686548.1@sss.pgh.pa.us>
Date: Thu, 06 Aug 2020 00:02:28 -0400
Message-ID: <2598266.1596686548@sss.pgh.pa.us>
Precedence: bulk

Andy Fan <zhihui.fan1213@gmail.com> writes:
> On Thu, Aug 6, 2020 at 2:22 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> In the longer term, it's annoying that we have no test methodology
>> for this other than "manually set a breakpoint here".

> One of the methods I see is we can just add some GUC variable for some
> action injection.   basically it adds some code based on the GUC like this;

See my straw-man proposal downthread.  I'm not very excited about putting
things like this into the standard build, because it's really hard to be
sure that there are no security-hazard-ish downsides of putting in ways to
get at testing behaviors from standard SQL.  And then there's the question
of whether you're adding noticeable overhead to production builds.  So a
loadable module that can use some existing hook to provide the needed
behavior seems like a better plan to me, whenever we can do it that way.

In general, though, it seems like we've seldom regretted investments in
test tooling.

			regards, tom lane