Re: Berserk Autovacuum (let's save next Mandrill)

On Mon, 2020-03-16 at 13:13 -0700, Andres Freund wrote:
> > Freezing tuples is the point of this patch.
>
> Sure. But not hurting existing installation is also a goal of the
> patch. Since this is introducing potentially significant performance
> downsides, I think it's good to be a bit conservative with the default
> configuration.
>
> I'm gettin a bit more bullish on implementing some of what what I
> discussed in
> https://www.postgresql.org/message-id/20200313213851.ejrk5gptnmp65uoo%40alap3.anarazel.de
> at the same time as this patch.
>
> In particularl, I think it'd make sense to *not* have a lower freezing
> horizon for insert vacuums (because it *will* cause problems), but if
> the page is dirty anyway, then do the freezing even if freeze_min_age
> etc would otherwise prevent us from doing so?

I don't quite see why freezing tuples in insert-only tables will cause
problems - are you saying that more WAL will be written compared to
freezing with a higher freeze_min_age?

> > As I have said, if you have a table where you insert many rows in few
> > transactions, you would trigger an autovacuum that then ends up doing nothing
> > because none of the rows have reached vacuum_freeze_table_age yet.
> > Then some time later you will get a really large vacuum run.
>
> Well, only if you don't further insert into the table. Which isn't that
> common a case for a table having a "really large vacuum run".

Ah, yes, you are right.
So it actually would not be worse if we use the normal freeze_min_age
for insert-only vacuums.

So do you think the patch would be ok as it is if we change only that?

Yours,
Laurenz Albe