Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 2 Dec 2004 04:41:44 +0000 (GMT)
From:      Rob DeMarco <r.p.demarco@att.net>
To:        freebsd-questions@freebsd.org
Message-ID:  <20041202044145.80CCB43D39@mx1.FreeBSD.org>

next in thread | raw e-mail | index | archive | help
From: "Rob DeMarco" <r.p.demarco@att.net>
To: Kris Kennaway <kris@obsecurity.org>
Cc: freebsd-questions@freebsd.org
Bcc: 
Subject: Re: Can 10M Buffer Ceiling be lowere?
Reply-To: 
In-Reply-To: <20041201035316.GA20429@xor.obsecurity.org>

On Tue, Nov 30, 2004 at 07:53:16PM -0800, Kris Kennaway wrote:
> On Wed, Dec 01, 2004 at 03:27:11AM +0000, r.p.demarco@att.net wrote:
> >      A technical question:
> > 
> >      I have an old NEC computer (c. 1997) running 5.3-RELEASE with
> > 48M of RAM.  Getting a new computer isn't an option right now, but
> > I would like to get as much out of my memory as possible.
> >      My /boot/kernel/kernel file is about 3M, and from the initial
> > boot: 
> > 		real memory  = 50331648 (48 MB)
> > 		avail memory = 43896832 (41 MB)
> > it appears this kernel takes up about 7M of memory with one screen saver
> > kld loaded.  With a few unneeded services (cron, sendmail) disabled, I
> > start off with about 26M free after a fresh reboot with just root logged in,
> > running `top'.  Looking at top, I noticed:
> > 
> >                 Mem: 4320K Active, 15M Inact, 12M Wired, 10M Buf, 11M Free
> >                                                          ^^^
> > 
> >      From TOP(1):
> > 
> >                 Buf: number of pages used for BIO-level disk caching
> > 
> >      Actually, the 10M is after some disk usage (it starts ~6M).
> > It never gets above 10M.  Is there anyway to adjust this, to
> > (say) a maximum of 5M?  Yes, a new 256 MB RAM system would be nice,
> > but until then, I would like to avoid serious paging running xclock :)
> > Thanks,
> 
> There's no point, that memory will be used if demanded.  Note that you
> still have 11M free in your example, so throwing away 6MB that is used
> for caching would only *reduce* performance.
> 
> Kris

    Thanks for the info.  The thing that got my attention was how,
after enough processes were spawned and (presumably, some of the cache
could have been used before needing to page) the 10M remained for use for
the "Buf" only -- or at least it seemed like it according to `top'.
But maybe I'm misreading that (I seemed to remember reading that
the "Wired" info always included the "Buf")  But then I am
definitely way out of my league here.  I can't even make sense out
of the SIZE / RES columns, neither which seem to add up to the actual
memory/swap used.

     Oh, well.  Regards,

     -Rob



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20041202044145.80CCB43D39>