From owner-freebsd-stable@FreeBSD.ORG  Mon Mar  7 21:28:25 2005
Return-Path: <owner-freebsd-stable@FreeBSD.ORG>
Delivered-To: freebsd-stable@freebsd.org
Received: by hub.freebsd.org (Postfix, from userid 758)
	id 52D9916A4CF; Mon,  7 Mar 2005 21:28:25 +0000 (GMT)
Date: Mon, 7 Mar 2005 21:28:25 +0000
From: Kris Kennaway <kris@FreeBSD.org>
To: Brian Reichert <reichert@numachi.com>
Message-ID: <20050307212825.GB34940@hub.freebsd.org>
References: <1110213509.593.18.camel@p4-3200.local>
	<20050307171009.GA37856@e-Gitt.NET> <20050307171244.GO22873@hub.freebsd.org>
	<20050307210539.GJ90757@numachi.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20050307210539.GJ90757@numachi.com>
User-Agent: Mutt/1.4.2.1i
cc: Kris Kennaway <kris@FreeBSD.org>
cc: freebsd-stable@freebsd.org
Subject: Re: FreeBSD 5.3 freezes under heavy hdd load
X-BeenThere: freebsd-stable@freebsd.org
X-Mailman-Version: 2.1.1
Precedence: list
List-Id: Production branch of FreeBSD source code
	<freebsd-stable.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-stable>,
	<mailto:freebsd-stable-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-stable>
List-Post: <mailto:freebsd-stable@freebsd.org>
List-Help: <mailto:freebsd-stable-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-stable>,
	<mailto:freebsd-stable-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Mon, 07 Mar 2005 21:28:25 -0000

On Mon, Mar 07, 2005 at 04:05:39PM -0500, Brian Reichert wrote:
> On Mon, Mar 07, 2005 at 05:12:44PM +0000, Kris Kennaway wrote:
> > It is necessary to increase KSTACK_PAGES (e.g. to 4) on machines with
> > certain patterns of heavy disk write load to avoid double faults in
> > the softupdates code (softupdates has potentially unbounded call stack
> > and can overflow the default kernel stack size without too much
> > effort).
> 
> Can these 'certain patterns' be characterized?

Probably combinations of lots of writes and removes of directories.

> I've got a system
> that hangs on haavy rsync usage, where the file sizes can me measured
> in a half-gig or more...
> 
> > This would cause panics though, not freezes.
> 
> I can't tell remotely if all of my hangs are panics, as it's a
> remote box.  Once, on the console of a hung box, I found
> 
>   panic: worklist_remove: not on list
> 
> This box did not reboot of it's own accord, and has wedged again
> under a similar workload...

That's probably different - as I said, the problem I'm talking about
causes double faults.

Kris