MIME-Version: 1.0
From: Akashkiran Shivakumar <akashkiran.foss@gmail.com>
Date: Tue, 29 Jul 2025 15:37:54 -0700
Message-ID: <CADTnmPjcbW07j26AxdqAsMbSJEAdHtva7Kdazmo99twuF59whw@mail.gmail.com>
Subject: Can postgres replication slot using pgoutput release multiple CDC
 records for a single update to a particular row
To: pgsql-general@lists.postgresql.org
Content-Type: multipart/alternative; boundary="000000000000caf7eb063b190ff2"
Archived-At: <https://www.postgresql.org/message-id/CADTnmPjcbW07j26AxdqAsMbSJEAdHtva7Kdazmo99twuF59whw%40mail.gmail.com>
Precedence: bulk

--000000000000caf7eb063b190ff2
Content-Type: text/plain; charset="UTF-8"

Hello,
I have a postgres database (major version 13) and am doing CDC by using a
replication slot with pgoutput. In our data lake, we see that there are
multiple updates (3 in this case) happening to the same row as part of the
same transaction. This doesn't make sense if we look at them as separate
updates. The expectation was that the row was updated once and a single CDC
record was pushed out.  I haven't completely ruled out whether multiple
updates happened in that transaction but I wanted to ask the community if
it is possible that a single update statement could spill over as multiple
CDC update records by pgoutput / postgres ?

If yes, could you possibly point to the testcases or code where this might
happen? Any blogs or suggestions are welcome. LMK if you need further
information

P.S: Each update seen in the data lake changes at least one field in the
row. The entire row with all the columns are pushed out for the 3 updates
seen in the data lake. From the application perspective, it makes sense if
the 3 updates were merged into a single update record

Regards,
Akashkiran

--000000000000caf7eb063b190ff2
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Hello,=C2=A0<br>I have a postgres database (major version =
13) and am doing CDC by using a replication slot with pgoutput. In our data=
 lake, we see that there are multiple updates (3 in this case) happening to=
 the same row as part of the same transaction. This doesn&#39;t make=C2=A0s=
ense if we look at them as separate updates. The expectation was that the r=
ow was updated once and a single CDC record was pushed out.=C2=A0=C2=A0I ha=
ven&#39;t completely ruled out whether multiple updates happened in that tr=
ansaction but I wanted to ask the community if it is possible that a single=
 update statement could spill over as multiple CDC update=C2=A0records=C2=
=A0by pgoutput=C2=A0/ postgres=C2=A0?=C2=A0<br><div><br>If yes, could you p=
ossibly point to the=C2=A0testcases or code where this might happen? Any bl=
ogs or suggestions are welcome. LMK if you need further information<br><br>=
</div><div>P.S: Each update seen in the data lake changes at least one fiel=
d in the row. The entire row with all the columns are pushed out for the 3 =
updates seen in the data lake. From the application perspective, it makes s=
ense if the 3 updates were merged into a single update record<br><div><br><=
/div><div>Regards,</div><div>Akashkiran</div></div></div>

--000000000000caf7eb063b190ff2--