MIME-Version: 1.0
From: Kevin K Biju <kevinkbiju@gmail.com>
Date: Tue, 30 Sep 2025 17:21:05 +0530
Message-ID: 
 <CAM45KeHa32soKL_G8Vk38CWvTBeOOXcsxAPAs7Jt7yPRf2mbVA@mail.gmail.com>
Subject: Replace O_EXCL with O_TRUNC for creation of state.tmp in
 SaveSlotToPath
To: pgsql-hackers@lists.postgresql.org
Content-Type: multipart/mixed; boundary="000000000000a6c2690640035e90"
Archived-At: 
 <https://www.postgresql.org/message-id/CAM45KeHa32soKL_G8Vk38CWvTBeOOXcsxAPAs7Jt7yPRf2mbVA%40mail.gmail.com>
Precedence: bulk

--000000000000a6c2690640035e90
Content-Type: multipart/alternative; boundary="000000000000a6c2680640035e8e"

--000000000000a6c2680640035e8e
Content-Type: text/plain; charset="UTF-8"

Hi,

We have encountered a few instances where logical replication errors out
during SaveSlotToPath() after creating the state.tmp file, but before it
was renamed (due to ENOSPC, for example). In these cases, since state.tmp
is not cleaned up and is created with the O_EXCL flag, further invocations
of SaveSlotToPath() for this slot will error out on OpenTransientFile()
with EEXIST, completely blocking slot metadata persistence. The only
explicit cleanup for state.tmp occurs during server startup as part of
RestoreSlotFromDisk().

It doesn't seem that this function relies on data written to state.tmp
previously, so O_EXCL is unnecessary. Attaching a patch that swaps O_EXCL
for O_TRUNC, ensuring a fresh state.tmp is available for writing.

Thanks,
Kevin

--000000000000a6c2680640035e8e
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Hi,<br><br>We have encountered a few instances where logic=
al replication errors out during SaveSlotToPath() after creating the state.=
tmp file, but before it was renamed (due to ENOSPC, for example). In these =
cases, since state.tmp is not cleaned up and is created with the O_EXCL fla=
g, further invocations of SaveSlotToPath() for this slot will error out on =
OpenTransientFile() with EEXIST, completely blocking slot metadata persiste=
nce. The only explicit cleanup for state.tmp occurs during server startup a=
s part of RestoreSlotFromDisk().<br><br>It doesn&#39;t seem that this funct=
ion relies on data written to state.tmp previously, so O_EXCL is unnecessar=
y. Attaching a patch that swaps O_EXCL for O_TRUNC, ensuring a fresh state.=
tmp is available for writing.=C2=A0<br><br>Thanks,<br>Kevin =C2=A0<br></div=
>

--000000000000a6c2680640035e8e--

--000000000000a6c2690640035e90
Content-Type: application/octet-stream;
	name="replslot_state_tmp_otrunc.patch"
Content-Disposition: attachment; filename="replslot_state_tmp_otrunc.patch"
Content-Transfer-Encoding: base64
Content-ID: <f_mg6gzcj40>
X-Attachment-Id: f_mg6gzcj40
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--000000000000a6c2690640035e90--