Re: Skipping logical replication transactions on subscriber side

On Fri, Oct 1, 2021 at 6:30 AM Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> wrote:
>
> On Fri, Oct 1, 2021 at 5:05 AM Peter Eisentraut
> <peter(dot)eisentraut(at)enterprisedb(dot)com> wrote:
> >
> > Also, what happens when you forget to reset the xid after it has passed?
> > Will it get skipped again after wraparound?
>
> Yes.
>

Aren't we resetting the skip_xid once we skip that transaction in
stop_skipping_changes()? If so, it shouldn't be possible to skip it
again after the wraparound. Am I missing something?

Now, if the user has wrongly set some XID which we can't skip as that
is already in past or something like that then I think it is the
user's problem and that's why it can be done only by super users. I
think we have even thought of protecting that via cross-checking with
the information in view but as the view data is lossy, we can't rely
on that. I think users can even set some valid XID that never has any
error and we will still skip it which is what can be done today also
by pg_replication_origin_advance(). I am not sure if we can do much
about such scenarios except to carefully document them.

--
With Regards,
Amit Kapila.