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

On Mon, 2020-03-02 at 14:57 +0100, I wrote:
> But I think it would be good to have *something* that addresses the immediate
> problem (INSERT-only tables are autovacuumed too late), as long as
> that does not have negative side-effects or blocks further improvements.
>
> I don't feel totally well with the very simplistic approach of this
> patch (use the same metric to trigger autoanalyze and autovacuum),
> but what about this:
>
> - a new table storage option autovacuum_vacuum_insert_threshold,
> perhaps a GUC of the same name, by default deactivated.
>
> - if tabentry->tuples_inserted exceeds this threshold, but not one
> of the others, lauch autovacuum with index_cleanup=off.

As a more substantial base for discussion, here is a patch that:

- introduces a GUC and reloption "autovacuum_vacuum_insert_limit",
default 10000000

- introduces a statistics counter "inserts_since_vacuum" per table
that gets reset to 0 after vacuum

- causes autovacuum to run without cleaning up indexes if
inserts_since_vacuum > autovacuum_vacuum_insert_limit
and there is no other reason for an autovacuum

No doc patch is included yet, and perhaps the new counter should
be shown in "pg_stat_user_tables".

Yours,
Laurenz Albe