Re: VACUUM PARALLEL option vs. max_parallel_maintenance_workers

On Mon, Sep 21, 2020 at 12:45 PM Peter Eisentraut
<peter(dot)eisentraut(at)2ndquadrant(dot)com> wrote:
>
> On 2020-09-21 05:48, Amit Kapila wrote:
> > What according to you should be the behavior here and how will it be
> > better than current?
>
> I think if I write VACUUM (PARALLEL 5), it should use up to 5 workers
> (up to the number of indexes), even if max_parallel_maintenance_workers
> is 2.
>

So you want it to disregard max_parallel_maintenance_workers but all
parallel operations have to regard one of the max_parallel_* option so
that it can respect max_parallel_workers beyond which the system won't
allow more parallel workers. Now, if we won't respect one of the
max_parallel_* option, it will unnecessarily try to register those
many workers even though it won't be able to start those many workers.
I think it is better to keep the limit for workers for scans and
maintenance operations separately so that the user is allowed to
perform different parallel operations in the system.

--
With Regards,
Amit Kapila.