Re: BUG #13750: Autovacuum slows down with large numbers of tables. More workers makes it slower.

Hi,

On 03/18/2016 11:13 PM, Alvaro Herrera wrote:
> David Gould wrote:
>
>> I have some thoughts for a different approach. In short, the stats collector
>> actually knows what needs vacuuming because queries that create dead tuples
>> tell it. I'm considering have the stats collector maintain a queue of
>> vacuum work and that autovacuum request work from the stats collector. When I
>> have something more concrete I'll post it on hackers.
>
> Hm, ideally we would have something backpatchable, but this new idea
> doesn't sound so to me. I think that having some variant of the patch
> you proposed previously would be better for the stable branches.

I believe implementing such queue in the pgstat will be rather
difficult, for example because you can tweak the limits in reloptions,
and those are only accessible in the database. So pgstat does not know
that and thus can't really decide what needs autovacuuming etc.

regards

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