Re: Skipping logical replication transactions on subscriber side

On Tue, Jul 20, 2021 at 6:56 AM Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> wrote:
>
> On Mon, Jul 19, 2021 at 8:38 PM houzj(dot)fnst(at)fujitsu(dot)com
> <houzj(dot)fnst(at)fujitsu(dot)com> wrote:
> >
> > 3) For 0003 patch, if user set skip_xid to a wrong xid which have not been
> > assigned, and then will the change be skipped when the xid is assigned in
> > the future even if it doesn't cause any conflicts ?
>
> Yes. Currently, setting a correct xid is the user's responsibility. I
> think it would be better to disable it or emit WARNING/ERROR when the
> user mistakenly set the wrong xid if we find out a convenient way to
> detect that.
>

I think in this regard we should clearly document how this can be
misused by users. I see that you have mentioned about skip_xid but
maybe we can add more on how it could lead to skipping a
non-conflicting XID and can lead to an inconsistent replica. As
discussed earlier as well, users can anyway do similar harm by using
pg_replication_slot_advance(). I think if possible we might want to
give some examples as well where it would be helpful for users to use
this functionality.

--
With Regards,
Amit Kapila.