Why don't you just monitor pg_stat_user_tables.n_dead_tup on a regular basis and increase autovacuum aggressiveness based on that at the global level (postgresql.conf - thresholds) or set autovacuum parms at the table level for customized cases.
Regards, Michael Vitale
Joseph Hammerman wrote on 3/12/2023 4:34 PM:
Hi all,
Apologies for any confusion I may have caused. What I am imagining is per table tuning that buckets the tables based on their relative sizes.
Something like:
Up to 1Gb - Small
Up to 4Gb - Medium
Up to 8Gb - L
Bigger - XL
And an accordant autovacuum_scale_factor associated with each size.
The motivation for this is to make sure large tables get regularly vacuumed.
I think Mr Hammerman is referring to T-shirts for user stories.
But even when I'm right I don't get what is meant with that. I don't get what Joe means with "autovacuuming profiles per table". Joe, can you elaborate on that?
Yours
Wolfgang
Am Sonntag, 12. März 2023 um 14:47:42 MEZ hat Laurenz Albe <laurenz.albe@cybertec.at> Folgendes geschrieben:
On Sat, 2023-03-11 at 10:49 -0800, Joseph Hammerman wrote:
> I would like to define t-shirt sizes and have an autovacuuming profile associated with each t-shirt size. > > Is there any tooling out there that assists in the execution side of this? Or are all of you rolling your own?