Re: Minimal logical decoding on standbys - Mailing list pgsql-hackers

Hi,

On 4/4/23 3:43 PM, Amit Kapila wrote:
> On Tue, Apr 4, 2023 at 6:05 PM Drouvot, Bertrand
> 
> I think we might want to consider slot's effective_xmin instead of
> data.xmin as we use that to store xmin_horizon when we build the full
> snapshot.
> 

Oh, I did not know about the 'effective_xmin' and was going to rely only on the catalog xmin.

Reading the comment in the ReplicationSlot struct about the 'effective_xmin' I do think it makes sense to use it
(instead of data.xmin).

Please find attached v59 doing so.

Regards,

-- 
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com