Re: Vacuum o/p with (full 1, parallel 0) option throwing an error

On Mon, 13 Apr 2020 at 18:25, Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>
> On Fri, Apr 10, 2020 at 7:05 PM Justin Pryzby <pryzby(at)telsasoft(dot)com> wrote:
> >
> >
> > No problem. I think I was trying to make my text similar to that from
> > 14a4f6f37.
> >
> > I realized that I didn't sq!uash my last patch, so it didn't include the
> > functional change (which is maybe what Robert was referring to).
> >
>
> I think it is better to add a new test for temporary table which has
> less data. We don't want to increase test timings to test the
> combination of options. I changed that in the attached patch. I will
> commit this tomorrow unless you or anyone else has any more comments.
>

Thank you for updating the patch!

I think we can update the documentation as well. Currently, the
documentation says "This option can't be used with the FULL option."
but we can say instead, for example, "VACUUM FULL can't use parallel
vacuum.".

Also, I'm concerned that the documentation says that VACUUM FULL
cannot use parallel vacuum and we compute the parallel degree when
PARALLEL option is omitted, but the following command is accepted:

postgres(1:55514)=# vacuum (full on) test;
VACUUM

Instead, we can say:

In plain VACUUM (without FULL), if the PARALLEL option is omitted,
then VACUUM decides the number of workers based on the number of
indexes that support parallel vacuum operation on the relation which
is further limited by max_parallel_maintenance_workers.

(it just adds "In plain VACUUM (without FULL)" to the beginning of the
original sentence.)

What do you think?

Regards,

--
Masahiko Sawada http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services