Re: Parallel heap vacuum

On Mon, Mar 10, 2025 at 5:03 PM Melanie Plageman
<melanieplageman(at)gmail(dot)com> wrote:
>
> On Sat, Mar 8, 2025 at 1:42 AM Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> wrote:
> >
> >
> > I've attached the updated version patches.
>
> I've started trying to review this and realized that, while I'm
> familiar with heap vacuuming code, I'm not familiar enough with the
> vacuumparallel.c machinery to be of help without much additional
> study. As such, I have mainly focused on reading the comments in your
> code.

Thank you for looking at the patch.

>
> I think your comment in vacuumlazy.c describing the design could use
> more detail and a bit of massaging.
>
> For example, I don't know what you mean when you say:
>
> * We could require different number of parallel vacuum workers for each phase
> * for various factors such as table size and number of indexes.
>
> Does that refer to something you did implement or you are saying we
> could do that in the future?

It referred to the parallel heap vacuum implementation that I wrote.
Since the parallel degrees for parallel heap scan and parallel index
vacuuming are chosen separately based on different factors, we launch
a different number of workers for each phase and they exit at the end
of each phase.

Regards,

--
Masahiko Sawada
Amazon Web Services: https://aws.amazon.com