Date: Tue, 10 Oct 2023 20:29:08 -0700
From: Noah Misch <noah@leadboat.com>
To: Peter Geoghegan <pg@bowt.ie>
Cc: pgsql-hackers@postgresql.org
Subject: Re: interval_ops shall stop using btequalimage (deduplication)
Message-ID: <20231011032908.35@rfd.leadboat.com>
References: <20231011013317.22.nmisch@google.com>
 <CAH2-Wzm_OFH4Aip3uinAJ-AoBgX0wGjB9XBhrhkTnjCoZ_i8NA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: 
 <CAH2-Wzm_OFH4Aip3uinAJ-AoBgX0wGjB9XBhrhkTnjCoZ_i8NA@mail.gmail.com>
User-Agent: Mutt/2.2.12 (2023-09-09)
Archived-At: 
 <https://www.postgresql.org/message-id/20231011032908.35%40rfd.leadboat.com>
Precedence: bulk

On Tue, Oct 10, 2023 at 08:12:36PM -0700, Peter Geoghegan wrote:
> On Tue, Oct 10, 2023 at 6:33 PM Noah Misch <noah@leadboat.com> wrote:
> > interval_ops, however, recognizes equal-but-distinguishable values:
> 
> > Fails with:
> >
> >   2498151 2023-10-10 05:06:46.177 GMT DEBUG:  building index "ti" on table "t" serially
> >   2498151 2023-10-10 05:06:46.178 GMT DEBUG:  index "ti" can safely use deduplication
> >   TRAP: failed Assert("!itup_key->allequalimage || keepnatts == _bt_keep_natts_fast(rel, lastleft, firstright)"), File: "nbtutils.c", Line: 2443, PID: 2498151
> 
> Nice catch.
> 
> Out of curiosity, how did you figure this out? Did it just occur to
> you that interval_ops had a behavior that made deduplication unsafe?
> Or did the problem come to your attention after running amcheck on a
> customer database? Or was it something else?

My friend got an amcheck "lacks matching index tuple" failure, and they asked
me about it.  I ran into the assertion failure while reproducing things.

> I'm a little surprised that it took this long to notice
> the interval_ops issue.

Agreed.  I don't usually store interval values in tables, and I'm not sure
I've ever indexed one.  Who knows.

> Do we really need to change the catalog contents when backpatching?

Not really.  I think we usually do.  On the other hand, unlike some past
cases, there's no functional need for the catalog changes.  The catalog
changes just get a bit of efficiency.  No strong preference here.