From owner-freebsd-stable Tue Apr 29 01:06:31 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id BAA03532 for stable-outgoing; Tue, 29 Apr 1997 01:06:31 -0700 (PDT) Received: from zibbi.mikom.csir.co.za (zibbi.mikom.csir.co.za [146.64.24.58]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id BAA03526 for ; Tue, 29 Apr 1997 01:06:14 -0700 (PDT) Received: (from jhay@localhost) by zibbi.mikom.csir.co.za (8.8.5/8.8.5) id KAA18509; Tue, 29 Apr 1997 10:03:57 +0200 (SAT) From: John Hay Message-Id: <199704290803.KAA18509@zibbi.mikom.csir.co.za> Subject: Re: Memory usage on NFS server In-Reply-To: <199704290720.AAA11888@implode.root.com> from David Greenman at "Apr 29, 97 00:20:09 am" To: dg@root.com Date: Tue, 29 Apr 1997 10:03:57 +0200 (SAT) Cc: freebsd-stable@freebsd.org X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stable@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > > 2.2-970420-RELENG on an NFS server that does only that. The ~20 > >processes normally running on the machine add up to less than 10MB, > >yes `top' reports that 78MB is "active". I would have though most of > >the memory would be allocated as cache (which is only reported at > >25MB). Am I interpreting the numbers incorrectly? > ... > >Mem: 78M Active, 4984K Inact, 17M Wired, 25M Cache, 8343K Buf, 616K Free > > The numbers lie; don't believe them. :-) It's not as simple as looking > at the "cache" number - pages in "active" and "inactive" are also part of > the file cache. I know this makes it very difficult to see how much memory > is actually available for caching...I have the same problem on wcarchive. > We need to provide an additional metric, but I don't know at the moment > how to create the desired information (it's difficult and perhaps impossible > in the current architecture). > Do the numbers that top and ps show for per process memory usage also lie? What I see here on my news server, is that I run out of swap (256M), but according to top and ps a rough calculation of the total of all the processes is less than half that. Inn then typically show a usage of ~70M according to top, but as soon as I kill and restart it the swap usage go down to ~5M. I once even added a 128M vn swapfile and it filled that also without inn showing a usage of more than 70M, but killing and restarting it takes the swap usage down to ~5M. John -- John Hay -- John.Hay@mikom.csir.co.za