Re: autovacuum_work_mem

On 11 December 2013 22:23, Peter Geoghegan <pg(at)heroku(dot)com> wrote:
> On Wed, Dec 11, 2013 at 1:28 PM, Simon Riggs <simon(at)2ndquadrant(dot)com> wrote:
>> But nobody has given a sensible answer to my questions, other than to
>> roll out the same general points again. In practice, its a knob that
>> does not do very much of use for us. At best it is addressing the
>> symptoms, not the cause. IMHO.
>
> It's just a usability feature. It lessens the intellectual overhead of
> managing maintenance_work_mem. I understand that it isn't very
> impressive from a technical point of view. However, in many
> environments, it actually will make a significant difference, because
> non-autovacuum maintenance operations are very rare compared to
> autovacuum workers vacuuming, and therefore I can now afford to be
> much less conservative in setting maintenance_work_mem globally on
> each of 8 Postgres instances hosted on a single box. These are
> precisely the kinds of Postgres instances where users are very
> unlikely to have heard of maintenance_work_mem at all. These users are
> not even using an admin tool in many cases, preferring to rely on ORM
> migrations. Having said that, it's also something I'll find useful on
> a day to day basis, because it's a chore to set maintenace_work_mem
> manually, and sometimes I forget to do so, particularly when under
> pressure to relieve a production performance issues on a random
> customer's database.

My view remains that, yes, we have a problem setting maint work men
usefully, my conclusion is that having two parameters we don't know
how to set won't improve things and doesn't constitute an improvement
in usability.

That being said, I acknowledge everybody else's viewpoints here and
will commit this feature as is.

--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services