Re: Scaling shared buffer eviction

On 2014-10-02 20:04:58 +0300, Heikki Linnakangas wrote:
> On 10/02/2014 05:40 PM, Robert Haas wrote:
> >On Thu, Oct 2, 2014 at 10:36 AM, Andres Freund <andres@2ndquadrant.com> wrote:
> >>>OK.
> >>
> >>Given that the results look good, do you plan to push this?
> >
> >By "this", you mean the increase in the number of buffer mapping
> >partitions to 128, and a corresponding increase in MAX_SIMUL_LWLOCKS?
> 
> Hmm, do we actually ever need to hold all the buffer partition locks at the
> same time? At a quick search for NUM_BUFFER_PARTITIONS in the code, I
> couldn't find any place where we'd do that. I bumped up
> NUM_BUFFER_PARTITIONS to 128, but left MAX_SIMUL_LWLOCKS at 100, and did
> "make check". It passed.

Do a make check-world and it'll hopefully fail ;). Check
pg_buffercache_pages.c.

I'd actually quite like to have a pg_buffercache version that, at least
optionally, doesn't do this, but that's a separate thing.

Greetings,

Andres Freund

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