Re: Still problems with memory swapping and server load - Mailing list pgsql-general

From Tom Lane
Subject Re: Still problems with memory swapping and server load
Date
Msg-id 24894.1025184867@sss.pgh.pa.us
Whole thread Raw
In response to Re: Still problems with memory swapping and server load  (Alvar Freude <alvar@a-blast.org>)
List pgsql-general
Alvar Freude <alvar@a-blast.org> writes:
> -- Curt Sampson <cjs@cynic.net> wrote:
>> Some of your backends are getting pretty darn big. I wonder what
>> they're doing? It can't be sort memory at this point. But as you
>> can see, those five 200-250MB backends are killing you.

> no, not really: they use shared memory:

No, because all the backends will be accessing the *exact same* shared
memory; it's impossible for some to be using more than others.  Since we
see process sizes ranging from 17M to 251M, it's a pretty safe bet that
the latter guys are actually chewing a lot of private space.  I don't
know what the "share" column means in that ps display, but I wouldn't
trust it.

            regards, tom lane



pgsql-general by date:

Previous
From: "Markus Wollny"
Date:
Subject: One source of constant annoyance identified
Next
From: Jan Wieck
Date:
Subject: Re: Shared Memory Sizing