Date: Sun, 9 Jun 1996 20:26:28 -0500 (EST) From: "John S. Dyson" <toor@dyson.iquest.net> To: taob@io.org (Brian Tao) Cc: dyson@freebsd.org, freebsd-hackers@freebsd.org Subject: Re: Memory leak or reporting problem in 2.2-960501-SNAP? Message-ID: <199606100126.UAA02692@dyson.iquest.net> In-Reply-To: <Pine.NEB.3.92.960609210401.8414I-100000@zap.io.org> from "Brian Tao" at Jun 9, 96 09:12:36 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > True, but the partitioning of memory is different, and that's what > I'm interested in. I just rebooted our Web/FTP server. Here's the > header from 'top', again with about 60 httpd's and 10 ftpd's running: > > load averages: 0.41, 0.42, 0.28 21:07:47 > 103 processes: 1 running, 102 sleeping > Cpu states: 5.3% user, 0.0% nice, 12.9% system, 8.3% interrupt, 73.5% idle > Mem: 61M Active, 1536K Inact, 19M Wired, 22M Cache, 6413K Buf, 25M Free > Swap: 262M Total, 4000K Used, 258M Free, 2% Inuse > > This intuitively looks more "correct" than the 91M active I saw > earlier, under similar load. I generally ignore the amount of cache > reported, since it grows and shrinks with the amount of otherwise free > memory. > I think I know what is happening now, the pageout daemon is necessary for the statistics that you are talking about to be meaningful. There is really no mechanism to show what is happening. The pages are placed onto the cache queue by the pageout daemon (or the vfs_bio) when memory is in short supply. Pages normally cycled through the buffer cache or when processes exit are left on the active queue. The "excessive" active queue numbers shouldn't clog up your system or anything evil like that. When you want to get a reasonable snapshot of your memory loading under light load conditions, run a program that loads memory for a very short while, and the system will reach an equilibrium with short term accurate statistics. John
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199606100126.UAA02692>