From owner-freebsd-threads@FreeBSD.ORG Tue Feb 10 17:13:13 2004 Return-Path: Delivered-To: freebsd-threads@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 4296A16A4CE for ; Tue, 10 Feb 2004 17:13:13 -0800 (PST) Received: from mail.pcnet.com (mail.pcnet.com [204.213.232.4]) by mx1.FreeBSD.org (Postfix) with ESMTP id 0220043D3F for ; Tue, 10 Feb 2004 17:13:13 -0800 (PST) (envelope-from eischen@vigrid.com) Received: from mail.pcnet.com (mail.pcnet.com [204.213.232.4]) by mail.pcnet.com (8.12.10/8.12.1) with ESMTP id i1B1D1fo005892; Tue, 10 Feb 2004 20:13:01 -0500 (EST) Date: Tue, 10 Feb 2004 20:13:01 -0500 (EST) From: Daniel Eischen X-Sender: eischen@pcnet5.pcnet.com To: Julian Elischer In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII cc: freebsd-threads@freebsd.org Subject: Re: Question about threads [beaver challenge] X-BeenThere: freebsd-threads@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Threading on FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 11 Feb 2004 01:13:13 -0000 On Tue, 10 Feb 2004, Julian Elischer wrote: > > > On Tue, 10 Feb 2004, Craig Rodrigues wrote: > > > On Tue, Feb 10, 2004 at 06:22:28PM -0500, Daniel Eischen wrote: > > > That's why I wanted to know what top -H showed. We saw the > > > same problem with python -- it was using system scope threads > > > by default. The port has since been changed to use process > > > scope threads. > > > > Is this a Linux-ism that we will need to be aware of > > for other ports that use pthreads? Probably, yes. > > process scope threads a re perfectly reasonable.. they however > use more resources and are more heavily limitted. > > > On the other hand process scope threads can lead to nasty surprises with > the limits as they can "suddenly" hit th ekernel limit after running > usccessfully for a time when they all (by some fluke) all decide to > enter the kernel at the same time. Well, since the kernel limit for scope system threads is much smaller than that for "threads blocked in kernel", you'd hit the limit much sooner if the port/application used scope system threads instead of scope process threads. > We probably should increase the limits from 150 and 50 to > 600 and 300 or something. That's fine by me. It's all a guessing game for me 'cause I don't really know what you'd expect to see with some of these applications. -- Dan Eischen