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

On Fri, Mar 13, 2020 at 10:48:27PM +0100, Laurenz Albe wrote:
> On Fri, 2020-03-13 at 13:44 -0500, Justin Pryzby wrote:
> > Possible it would be better to run VACUUM *without* freeze_min_age=0 ?  (I get
> > confused and have to spend 20min re-reading the vacuum GUC docs every time I
> > deal with this stuff, so maybe I'm off).
> > 
> > As I understand, the initial motivation of this patch was to avoid disruptive
> > anti-wraparound vacuums on insert-only table.  But if vacuum were triggered at
> > all, it would freeze the oldest tuples, which is all that's needed; especially
> > since fd31cd2651 "Don't vacuum all-frozen pages.", those pages would never need
> > to be vacuumed again.  Recently written tuples wouldn't be frozen, which is ok,
> > they're handled next time.
> 
> Freezing tuples too early is wasteful if the tuples get updated or deleted
> soon after, but based on the assumption that an autovacuum triggered by insert
> is dealing with an insert-mostly table, it is not that wasteful.

You're right that it's not *that* wasteful.  If it's a table that gets 90%
inserts/10% updates, then only 10% of its tuples will be frozen.  In the worst
case, it's the same tuples every time, and that's somewhat wasteful.  In the
best case, those tuples are clustered on a small number of pages.

-- 
Justin