From owner-freebsd-questions@FreeBSD.ORG Thu Sep 18 15:41:34 2003 Return-Path: Delivered-To: freebsd-questions@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 049E916A4B3 for ; Thu, 18 Sep 2003 15:41:34 -0700 (PDT) Received: from exgw2.lumeta.com (exgw4.lumeta.com [65.246.240.66]) by mx1.FreeBSD.org (Postfix) with ESMTP id 38636440B3 for ; Thu, 18 Sep 2003 15:41:00 -0700 (PDT) (envelope-from tal@lumeta.com) Received: from lucy.corp.lumeta.com (h65-246-245-10.lumeta.com [65.246.245.10]) by exgw2.lumeta.com (Postfix) with ESMTP id 24014373850 for ; Thu, 18 Sep 2003 18:40:59 -0400 (EDT) Received: from localhost (localhost.corp.lumeta.com [127.0.0.1]) by lucy.corp.lumeta.com (Postfix) with ESMTP id 05CE2A8934 for ; Thu, 18 Sep 2003 18:40:59 -0400 (EDT) Received: from lucy.corp.lumeta.com ([127.0.0.1]) by localhost (lucy.corp.lumeta.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 04316-09 for ; Thu, 18 Sep 2003 18:40:58 -0400 (EDT) Received: from lumeta.com (talmacmm.corp.lumeta.com [65.246.246.66]) by lucy.corp.lumeta.com (Postfix) with ESMTP id 6A02BA892C for ; Thu, 18 Sep 2003 18:40:58 -0400 (EDT) Date: Thu, 18 Sep 2003 18:40:57 -0400 Mime-Version: 1.0 (Apple Message framework v552) Content-Type: text/plain; delsp=yes; charset=US-ASCII; format=flowed From: Tom Limoncelli To: freebsd-questions@freebsd.org Content-Transfer-Encoding: 7bit Message-Id: <2BA835A4-EA29-11D7-809A-000A956888C8@lumeta.com> X-Mailer: Apple Mail (2.552) X-Virus-Scanned: by amavisd-new at lumeta.com Subject: How many nfsd's to run on a Xeon? X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 18 Sep 2003 22:41:34 -0000 Is it possible to run too many nfsd's? I remember back in 1992-1994 someone published a paper at Usenix (they were from Sun or Auspex, I don't remember) saying that performance went down if you ran more nfsd's than there were built-in process slots in your CPU. So, a Sparc that had 16 hardware context slots, running more than 15 nfsd's hurt performance because the CPU was thrashing. Is there an equiv. setting for a Xeon processor? --Tom P.S. Here is the CPU data from dmesg from the host I'm trying to tune: CPU: Intel(R) XEON(TM) CPU 2.20GHz (2199.76-MHz 686-class CPU) Origin = "GenuineIntel" Id = 0xf24 Stepping = 4 Features=0x3febfbff Hyperthreading: 2 logical CPUs