Hi,
On 2022-11-21 15:16:46 -0500, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
> > On 2022-11-21 14:21:35 -0500, Tom Lane wrote:
> >> pg_resetwal does seem like a better, more useful home for this; it'd
> >> allow you to adjust these numbers after initial creation which might be
> >> useful. I'm not sure how flexible it is right now in terms of where
> >> you can set the new values, but that can always be improved.
>
> > IIRC the respective pg_resetwal parameters are really hard to use for
> > something like this, because they don't actually create the respective
> > SLRU segments. We of course could fix that.
>
> Is that still true? We should fix it, for sure.
Sure looks that way to me. I think it might mostly work if you manage to
find nextXid, nextMulti, nextMultiOffset values that each point to the
start of a segment that'd then be created whenever those values are
used.
Greetings,
Andres Freund