On Tue, Jul 26, 2022 at 11:43 AM Jonathan S. Katz <jkatz@postgresql.org> wrote:
>
...
> That said, this introduces a new restriction for this particular
> scenario that doesn't exist on other scenarios. Instead, I would
> advocate we document how to correctly set up the two-way replication
> scenario (which we have a draft!), document the warnings around the
> conflicts, perhaps include Vignesh's instructions on how to remediate a
> conflict on initial sync, and consider throwing a WARNING as you suggested.
>
> Thoughts?
Perhaps a WARNING can be useful if the SUBSCRIPTION was created with
enabled=false because then the user still has a chance to reconsider,
but otherwise, I don't see what good a warning does if the potentially
harmful initial copy is going to proceed anyway; isn't that like
putting a warning sign at the bottom of a cliff?
------
Kind Regards,
Peter Smith.
Fujitsu Australia