Re: Parallel heap vacuum

On Mon, Feb 17, 2025 at 12:11 PM Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com>
wrote:

> > If the idea is to never allow parallelism in vacuum, then I think
> > disabling eager scanning during manual parallel vacuum seems
> > reasonable. People could use vacuum freeze if they want more freezing.
>
> IIUC the purpose of parallel vacuum is incompatible with the purpose
> of auto vacuum. The former is aimed to execute the vacuum as fast as
> possible using more resources, whereas the latter is aimed to execute
> the vacuum while not affecting foreground transaction processing. It's
> probably worth considering to enable parallel vacuum even for
> autovacuum in a wraparound situation, but the purpose would remain the
> same.
>

That's assuming that the database is running with autovacuum_cost_delay >
0. There's presumably some number of systems that intentionally do not
throttle autovacuum. Another consideration is that people are free to
set vacuum_cost_page_miss = 0; in those cases it would still be useful to
parallelize at least phase 1 and 2. Of course, folks can also
set vacuum_cost_page_dirty = 0, in which case parallelization would help
phase 2 and 3.

In terms of less hypothetical scenarios, I've definitely run into cases
where 1 table accounts for 90%+ of the space used by a database.
Parallelism would help ensure the single table is still processed in a
timely fashion. (Assuming non-default settings for vacuum throttling.)