Re: Resetting spilled txn statistics in pg_stat_replication

On Fri, Oct 23, 2020 at 7:42 AM Masahiko Sawada
<masahiko(dot)sawada(at)2ndquadrant(dot)com> wrote:
>
> On Thu, 22 Oct 2020 at 20:34, Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
> >
> > > > I have modified the description of spill_count and spill_txns to make
> > > > things clear. Any suggestions?
> > >
> > > Thank you for the patch.
> > >
> > > - logical decoding exceeds
> > > <literal>logical_decoding_work_mem</literal>. The
> > > - counter gets incremented both for toplevel transactions and
> > > - subtransactions.
> > > + logical decoding of changes from WAL for this exceeds
> > > + <literal>logical_decoding_work_mem</literal>. The counter gets
> > > + incremented both for toplevel transactions and subtransactions.
> > >
> > > What is the word "this" in the above change referring to?
> > >
> >
> > 'slot'. The word *slot* is missing in the sentence.
> >
> > > How about
> > > something like:
> > >
> >
> > > Number of transactions spilled to disk after the memory used by
> > > logical decoding of changes from WAL exceeding
> > > logical_decoding_work_mem. The counter gets incremented both for
> > > toplevel transactions and subtransactions.
> > >
> >
> > /exceeding/exceeds. I am fine with your proposed text as well but if
> > you like the above after correction that would be better because it
> > would be more close to spill_count description.
>
> yeah, I agree with the correction.
>

Okay, thanks, attached is an updated patch. I'll push this early next
week unless you or someone else has any comments/suggestions.

--
With Regards,
Amit Kapila.