Re: When do vacuumed pages/tuples become available for reuse? - Mailing list pgsql-general

From Jeff Janes
Subject Re: When do vacuumed pages/tuples become available for reuse?
Date
Msg-id CAMkU=1w5tuRcA_BuNgs+cd-VgpZ71RBw=RyOykvDo0+JeDCyfQ@mail.gmail.com
Whole thread Raw
In response to Re: When do vacuumed pages/tuples become available for reuse?  (rihad <rihad@mail.ru>)
List pgsql-general
On Thu, Apr 11, 2019 at 11:14 AM rihad <rihad@mail.ru> wrote:

> autovacuum_vacuum_scale_factor = 0.01
> autovacuum_vacuum_threshold = 50  

This seems counterproductive.  You need to make the vacuum more efficient, not more frantic.


autovacuum_vacuum_cost_delay = 10ms
autovacuum_vacuum_cost_limit = 400
 
Anything more than that and we risk impacting the performance of user
queries.

Well, unbounded bloat will also impact the user queries--eventually.  Was this an experimental determination?  Can you tell what about the autovac most impacts the user queries, the reading or the writing?

You might just have more workload than your hardware can handle.  There is always going to be some fundamental limit, and while you can tune your way up to that limit, you can't tune your way past it.

Cheers,

Jeff

pgsql-general by date:

Previous
From: rihad
Date:
Subject: Re: When do vacuumed pages/tuples become available for reuse?
Next
From: Jeff Janes
Date:
Subject: Re: When do vacuumed pages/tuples become available for reuse?