From owner-freebsd-bugs Mon Sep 28 22:11:23 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id WAA12165 for freebsd-bugs-outgoing; Mon, 28 Sep 1998 22:11:23 -0700 (PDT) (envelope-from owner-freebsd-bugs@FreeBSD.ORG) Received: from peak.mountin.net ([207.227.119.2]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id WAA12160; Mon, 28 Sep 1998 22:11:14 -0700 (PDT) (envelope-from jeff@mountin.net) Received: (from daemon@localhost) by peak.mountin.net (8.9.1/8.9.1) id AAA14414; Tue, 29 Sep 1998 00:11:18 -0500 (CDT) Received: from aridius-9.isdn.mke.execpc.com(169.207.66.136) by peak.mountin.net via smap (V1.3) id sma014412; Tue Sep 29 00:11:00 1998 Message-Id: <3.0.3.32.19980929000935.00718e6c@207.227.119.2> X-Sender: jeffm@207.227.119.2 X-Mailer: QUALCOMM Windows Eudora Pro Version 3.0.3 (32) Date: Tue, 29 Sep 1998 00:09:35 -0500 To: freebsd-isp@FreeBSD.ORG From: "Jeffrey J. Mountin" Subject: MFS, mount, and vmdaemon Cc: freebsd-bugs@FreeBSD.ORG Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Sender: owner-freebsd-bugs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Posted to both bugs and isp, much as I don't like to do this. Maybe this should be on hackers too. 8-) Don't quite know if this is a bug, but when using a MFS for /tmp and you do a 'mount -a' a second MFS for /tmp is created. This shows up with 'df' and 'ps', but one can be killed without problem it seems. Using noauto, like with a CDROM, means it will not mount initially. Doesn't seem like good behaviour. This also leads to a question. Can you have 2 MFS in fstab? Something like: /dev/wd0s1b none swap sw 0 0 /dev/wd0s1b /tmp mfs rw 0 0 /dev/wd0s1b /htdocs mfs rw 0 0 I presume that any MFS must use the swap (or a swap), but can't find anything saying you can or cannot use more than one. Or is it best to (sym)link /htdocs to /tmp instead, which works fine, but I'm toying with a few ideas. In past I thought of running a web site off of MFS and in part plan to do so still. One thing about running FBSD with plenty of memory is how fast things get once the cache and active pages increase: last pid: 8865; load averages: 0.21, 0.21, 0.19 00:01:28 40 processes: 2 running, 36 sleeping, 2 zombie CPU states: % user, % nice, % system, % interrupt, % idle Mem: 137M Active, 46M Inact, 28M Wired, 34M Cache, 8318K Buf, 5444K Free It took less than 4 seconds to copy 11MB from the htdocs directory to the MFS version of the htdocs directory, which tells me that many of the files were in memory. Subsequent copies are almost instant. Normal performance with only 96MB vs 256MB is astronmical. Must say the vmdaemon does a good job. Even the MFS's resident size will shrink and I pushed it to 80MB from the startup of 7MB, but without keeping the files present and it dropped in size as the memory load went up. Still would be nice to start with all htdocs on MFS, but also wonder if the vmdaemon duplicates this by keeping files from the MFS in memory, which seems kind of redundant and wasteful. comments? Jeff Mountin - Unix Systems TCP/IP networking jeff@mountin.net To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-bugs" in the body of the message