MIME-Version: 1.0
References: <tencent_8DEC9842690A9B6AFD52D4659EF0700E9409@qq.com>
 <ZrnlgJEH473Q1kTp@ip-10-97-1-34.eu-west-3.compute.internal> <tencent_925A991463194F3C97830C3BB7D0A2C2BD07@qq.com>
In-Reply-To: <tencent_925A991463194F3C97830C3BB7D0A2C2BD07@qq.com>
From: Masahiko Sawada <sawada.mshk@gmail.com>
Date: Thu, 3 Apr 2025 23:09:35 -0700
Message-ID: <CAD21AoCRD1=D2JOQmR444OttUf4ddbL-ChFLkzvQCCNfh1768w@mail.gmail.com>
Subject: Re: Historic snapshot doesn't track txns committed in
 BUILDING_SNAPSHOT state
To: cca5507 <cca5507@qq.com>
Cc: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>, Michael Paquier <michael@paquier.xyz>, 
	pgsql-hackers <pgsql-hackers@lists.postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: <https://www.postgresql.org/message-id/CAD21AoCRD1%3DD2JOQmR444OttUf4ddbL-ChFLkzvQCCNfh1768w%40mail.gmail.com>
Precedence: bulk

On Mon, Aug 12, 2024 at 9:25=E2=80=AFPM cca5507 <cca5507@qq.com> wrote:
>
> Hi,
>
> I refactor the code and fix the git apply warning according to [1].
>
> Here are the new version patches.
>

I've investigated the reported issue and reviewed the patch. IIUC to
fix this issue, what we need to do is to track catalog-change
transactions even in BUILDING_SNAPSHOT phase so that we can decode
transactions correctly that started in FULL_SNAPSHOT. My question is;
in order to track just catalog-change transactions, whether it's
sufficient to check if XLOG_XACT_COMMIT[_PREPARED] has the
XACT_XINFO_HAS_INVALS flag. If yes, we probably should change only
xact_decode() to check the commit records even in BUILDING_SNAPSHOT.
Otherwise, we would need to change mostly all paths where we mark the
transaction as catalog-change as the patch does.

Regards,

--=20
Masahiko Sawada
Amazon Web Services: https://aws.amazon.com