From owner-freebsd-mobile  Wed Jun  7 11:59:16 2000
Delivered-To: freebsd-mobile@freebsd.org
Received: from salmon.maths.tcd.ie (salmon.maths.tcd.ie [134.226.81.11])
	by hub.freebsd.org (Postfix) with SMTP id 0F63737BC75
	for <freebsd-mobile@freebsd.org>; Wed,  7 Jun 2000 11:59:06 -0700 (PDT)
	(envelope-from iedowse@maths.tcd.ie)
Received: from bell.maths.tcd.ie by salmon.maths.tcd.ie with SMTP
          id <aa73034@salmon>; 7 Jun 2000 19:58:55 +0100 (BST)
To: sec@42.org
Cc: freebsd-mobile@freebsd.org
Subject: Re: spinning down the drive
Date: Wed, 07 Jun 2000 19:58:54 +0100
From: Ian Dowse <iedowse@maths.tcd.ie>
Message-ID:  <200006071958.aa73034@salmon.maths.tcd.ie>
Sender: owner-freebsd-mobile@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.org


Stefan `Sec` Zehl <sec@42.org> writes:

>On 2.2.x Systems you could increase the sync interval. Which I set to 3
>hours. I then typed 'sync' whenever the hard drive spun up after a read.
>That way I got a significant longer uptime from my battery.

>Unfortunately on 3.0 and greater you can't defer writes that much.

In FreeBSD 3.x and newer I've used something along the lines of:

	echo 'set syncdelay = 600' | gdb -k -wcore /kernel /dev/mem

This doesn't have quite the same effect as the old 'kern.update' sysctl,
but if you then set up a script such as

	#!/bin/sh
	while :; do
		sleep 600
		sync
	done

and run it in the background, the overall effect is similar. The idea
is that syncing of buffers is normally spread out evenly over time,
with each buffer being delayed no more than 'syncdelay' seconds. With
the above script, all buffers get flushed together, so no further
(timeout-based) buffer flushing will occur for 'syncdelay' seconds.

Ian


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-mobile" in the body of the message