Horton

Thread

Commits

  1. doc: Fix documenation for snapshot export in logical decoding.

  2. Don't force-assign transaction id when exporting a snapshot.

  1. Hot standbys now support exporting snapshots while creating a logical replication slot

    The Post Office <noreply@postgresql.org> — 2025-05-22T11:53:24Z

    The following documentation comment has been logged on the website:

Page: https://www.postgresql.org/docs/17/logicaldecoding-explanation.html
Description:

Hi,
I'm referring to this bit at the very end of the page: Creation of a
snapshot is not always possible. In particular, it will fail when connected
to a hot standby. Applications that do not require snapshot export may
suppress it with the NOEXPORT_SNAPSHOT option.
It's not clear what the intention of this is (is it referring only to
physical replication slots?) but as it stands you can export snapshots while
creating a logical replication slot. This should have been possible starting
PG16, I've tested this on PG17.
postgres=# CREATE_REPLICATION_SLOT export LOGICAL pgoutput (SNAPSHOT
'export');
-[ RECORD 1 ]----+--------------------
slot_name        | export
consistent_point | 0/36C9F28
snapshot_name    | 00000072-0000000A-1
output_plugin    | pgoutput
postgres=# SELECT pg_is_in_recovery();
-[ RECORD 1 ]-----+--
pg_is_in_recovery | t
postgres=# SHOW hot_standby;
-[ RECORD 1 ]---
hot_standby | on
As an aside, I think NOEXPORT_SNAPSHOT is from the older syntax of
CREATE_REPLICATION_SLOT and should be swapped with SNAPSHOT 'nothing'.
Thanks,
Kevin

  2. Re: Hot standbys now support exporting snapshots while creating a logical replication slot

    Fujii Masao <masao.fujii@oss.nttdata.com> — 2025-05-23T01:42:13Z

    
On 2025/05/22 20:53, PG Doc comments form wrote:
> The following documentation comment has been logged on the website:
> 
> Page: https://www.postgresql.org/docs/17/logicaldecoding-explanation.html
> Description:
> 
> Hi,
> I'm referring to this bit at the very end of the page: Creation of a
> snapshot is not always possible. In particular, it will fail when connected
> to a hot standby. Applications that do not require snapshot export may
> suppress it with the NOEXPORT_SNAPSHOT option.
> It's not clear what the intention of this is (is it referring only to
> physical replication slots?) but as it stands you can export snapshots while
> creating a logical replication slot. This should have been possible starting
> PG16, I've tested this on PG17.

The description "Creation of a snapshot is not always possible. In particular,
it will fail when connected to a hot standby." no longer seems accurate.
Since v10 (commit 6c2003f8a1b), exporting snapshots on a standby has been supported.

This description was accurate when the first patch [1] adding that description
was written, but I think we just overlooked updating it later.


> As an aside, I think NOEXPORT_SNAPSHOT is from the older syntax of
> CREATE_REPLICATION_SLOT and should be swapped with SNAPSHOT 'nothing'.

Yes, so how about the following change?

      <para>
-     Creation of a snapshot is not always possible.  In particular, it will
-     fail when connected to a hot standby.  Applications that do not require
-     snapshot export may suppress it with the <literal>NOEXPORT_SNAPSHOT</literal>
+     Applications that do not require
+     snapshot export may suppress it with the <literal>SNAPSHOT 'nothing'</literal>
       option.
      </para>

Regards,

[1] https://www.postgresql.org/message-id/CAMsr+YFjxv0T8Yi1Q=3tVdgViU2bm+fb_-XUbTfxHsnLDNSkzg@mail.gmail.com

Regards,

-- 
Fujii Masao
Advanced Computing Technology Center
Research and Development Headquarters
NTT DATA CORPORATION

  3. Re: Hot standbys now support exporting snapshots while creating a logical replication slot

    Kevin K Biju <kevinkbiju@gmail.com> — 2025-05-23T10:52:13Z

    Hi Fuiji,

I wasn't aware that support for exporting snapshots goes as far back as
v10. The change looks good.

Thanks,
Kevin

On Fri, May 23, 2025 at 7:12 AM Fujii Masao <masao.fujii@oss.nttdata.com>
wrote:

>
>
> On 2025/05/22 20:53, PG Doc comments form wrote:
> > The following documentation comment has been logged on the website:
> >
> > Page:
> https://www.postgresql.org/docs/17/logicaldecoding-explanation.html
> > Description:
> >
> > Hi,
> > I'm referring to this bit at the very end of the page: Creation of a
> > snapshot is not always possible. In particular, it will fail when
> connected
> > to a hot standby. Applications that do not require snapshot export may
> > suppress it with the NOEXPORT_SNAPSHOT option.
> > It's not clear what the intention of this is (is it referring only to
> > physical replication slots?) but as it stands you can export snapshots
> while
> > creating a logical replication slot. This should have been possible
> starting
> > PG16, I've tested this on PG17.
>
> The description "Creation of a snapshot is not always possible. In
> particular,
> it will fail when connected to a hot standby." no longer seems accurate.
> Since v10 (commit 6c2003f8a1b), exporting snapshots on a standby has been
> supported.
>
> This description was accurate when the first patch [1] adding that
> description
> was written, but I think we just overlooked updating it later.
>
>
> > As an aside, I think NOEXPORT_SNAPSHOT is from the older syntax of
> > CREATE_REPLICATION_SLOT and should be swapped with SNAPSHOT 'nothing'.
>
> Yes, so how about the following change?
>
>       <para>
> -     Creation of a snapshot is not always possible.  In particular, it
> will
> -     fail when connected to a hot standby.  Applications that do not
> require
> -     snapshot export may suppress it with the
> <literal>NOEXPORT_SNAPSHOT</literal>
> +     Applications that do not require
> +     snapshot export may suppress it with the <literal>SNAPSHOT
> 'nothing'</literal>
>        option.
>       </para>
>
> Regards,
>
> [1]
> https://www.postgresql.org/message-id/CAMsr+YFjxv0T8Yi1Q=3tVdgViU2bm+fb_-XUbTfxHsnLDNSkzg@mail.gmail.com
>
> Regards,
>
> --
> Fujii Masao
> Advanced Computing Technology Center
> Research and Development Headquarters
> NTT DATA CORPORATION
>
>

  4. Re: Hot standbys now support exporting snapshots while creating a logical replication slot

    Fujii Masao <masao.fujii@oss.nttdata.com> — 2025-05-23T14:46:36Z

    
On 2025/05/23 19:52, Kevin K Biju wrote:
> Hi Fuiji,
> 
> I wasn't aware that support for exporting snapshots goes as far back as v10. The change looks good.

Thanks for the review!

I've prepared two patches: one for v15 and later, and another for v13 and v14.
Unless there are any objections, I'll go ahead and commit and back-patch the patch
to all supported branches.

Regards,

-- 
Fujii Masao
NTT DATA Japan Corporation

  5. Re: Hot standbys now support exporting snapshots while creating a logical replication slot

    Fujii Masao <masao.fujii@oss.nttdata.com> — 2025-05-26T03:56:38Z

    
On 2025/05/23 23:46, Fujii Masao wrote:
> 
> 
> On 2025/05/23 19:52, Kevin K Biju wrote:
>> Hi Fuiji,
>>
>> I wasn't aware that support for exporting snapshots goes as far back as v10. The change looks good.
> 
> Thanks for the review!
> 
> I've prepared two patches: one for v15 and later, and another for v13 and v14.
> Unless there are any objections, I'll go ahead and commit and back-patch the patch
> to all supported branches.

I've pushed the patch. Thanks!

Regards,

-- 
Fujii Masao
NTT DATA Japan Corporation