Thread: Memory taken by FSM_relations
Folks: I'm not clear on where the memory needed by max_fsm_pages comes from. Is it taken out of the shared_buffers, or in addition to them? Further, Joe Conway gave me a guesstimate of 6k per max_fsm_pages which seems rather high ... in fact, the default settings for this value (10000) would swamp the memory used by the rest of Postgres. Does anyone have a good measurment of the memory load imposed by higher FSM settings? -- Josh Berkus Aglio Database Solutions San Francisco
Josh Berkus wrote: > Further, Joe Conway gave me a guesstimate of 6k per max_fsm_pages which seems > rather high ... in fact, the default settings for this value (10000) would > swamp the memory used by the rest of Postgres. I don't recall (and cannot find in my sent mail) ever making that guesstimate. Can you provide some context? Joe
Joe, > I don't recall (and cannot find in my sent mail) ever making that > guesstimate. Can you provide some context? Yeah, hold on ... hmmm ... no, your e-mail did not provide a figure. Sorry! Maybe I got it from Neil? In any case, it can't be the right figure ... -- Josh Berkus Aglio Database Solutions San Francisco
--On Monday, February 24, 2003 11:11:34 -0800 Joe Conway <mail@joeconway.com> wrote: > Josh Berkus wrote: >> Further, Joe Conway gave me a guesstimate of 6k per max_fsm_pages which >> seems rather high ... in fact, the default settings for this value >> (10000) would swamp the memory used by the rest of Postgres. > > I don't recall (and cannot find in my sent mail) ever making that > guesstimate. Can you provide some context? It may be 6 **BYTES** per max_fsm_pages. > > Joe > > > > > ---------------------------(end of broadcast)--------------------------- > TIP 6: Have you searched our list archives? > > http://archives.postgresql.org > -- Larry Rosenman http://www.lerctr.org/~ler Phone: +1 972-414-9812 E-Mail: ler@lerctr.org US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749
On Mon, 24 Feb 2003, Joe Conway wrote: > Josh Berkus wrote: > > Further, Joe Conway gave me a guesstimate of 6k per max_fsm_pages which seems > > rather high ... in fact, the default settings for this value (10000) would > > swamp the memory used by the rest of Postgres. > > I don't recall (and cannot find in my sent mail) ever making that > guesstimate. Can you provide some context? If I remember right, it was 6 BYTES per max fsm pages... not kbytes. That sounds about right anyway.
Scott, > If I remember right, it was 6 BYTES per max fsm pages... not kbytes. > That sounds about right anyway. So, does it come out of Shared_buffers or add to it? -- Josh Berkus Aglio Database Solutions San Francisco
Josh Berkus <josh@agliodbs.com> writes:
> I'm not clear on where the memory needed by max_fsm_pages comes from.
> Is it taken out of the shared_buffers, or in addition to them?
In addition to. Basically max_fsm_pages and max_fsm_relations are used
to ratchet up the postmaster's initial shared memory request to the kernel.
> Further, Joe Conway gave me a guesstimate of 6k per max_fsm_pages
> which seems rather high ...
Quite ;-). The correct figure is six bytes per fsm_page slot, and
I think about forty bytes per fsm_relation slot (recent versions of
postgresql.conf mention the multipliers, although for some reason
the Admin Guide does not).
regards, tom lane