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

On Thu, 2020-03-19 at 21:39 +1300, David Rowley wrote:
> > According to my reckoning, that is the remaining objection to the patch
> > as it is (with ordinary freezing behavior).
> >
> > How about a scale_factor od 0.005?  That will be high enough for large
> > tables, which seem to be the main concern here.
> 
> I agree with that, however, I'd thought 0.01, just so we're still
> close to having about 100 times less work to do for huge insert-only
> tables when it comes to having to perform an anti-wraparound vacuum.

Fine with me.

> > I am still sorry to see more proactive freezing go, which would
> > reduce the impact for truly insert-only tables.
> > After sleeping on it, here is one last idea.
> >
> > Granted, freezing with vacuum_freeze_min_age = 0 poses a problem
> > for those parts of the table that will receive updates or deletes.
> > But what if insert-triggered vacuum operates with - say -
> > one tenth of vacuum_freeze_min_age (unless explicitly overridden
> > for the table)?  That might still be high enough not to needlessly
> > freeze too many tuples that will still be modified, but it will
> > reduce the impact on insert-only tables.
> 
> I think that might be a bit too magical and may not be what some
> people want. I know that most people won't set
> autovacuum_freeze_min_age to 0 for insert-only tables, but we can at
> least throw something in the documents to mention it's a good idea,
> however, looking over the docs I'm not too sure the best place to note
> that down.

I was afraid that idea would be too cute to appeal.

> I've attached a small fix which I'd like to apply to your v8 patch.
> With that, and pending one final look, I'd like to push this during my
> Monday (New Zealand time).  So if anyone strongly objects to that,
> please state their case before then.

Thanks!

I have rolled your edits into the attached patch v9, rebased against
current master.

Yours,
Laurenz Albe