From owner-freebsd-stable Mon Apr 15 16:20:40 2002 Delivered-To: freebsd-stable@freebsd.org Received: from mail3.texoma.net (mail3.texoma.net [209.151.96.18]) by hub.freebsd.org (Postfix) with SMTP id 89E7437B405 for ; Mon, 15 Apr 2002 16:20:34 -0700 (PDT) Received: (qmail 2751 invoked from network); 15 Apr 2002 23:20:33 -0000 Received: from unknown (HELO love2golf.texoma.net) (209.151.96.67) by mail3.texoma.net with SMTP; 15 Apr 2002 23:20:33 -0000 Message-Id: <5.1.0.14.2.20020415180842.057ff460@mail3.texoma.net> X-Sender: ldvhomeu@mail3.texoma.net X-Mailer: QUALCOMM Windows Eudora Version 5.1 Date: Mon, 15 Apr 2002 18:20:32 -0500 To: David Malone From: Larry Vaden Subject: Re: best current practice re: number of NFS servers Cc: freebsd-stable@FreeBSD.ORG In-Reply-To: <20020415223359.GA46835@lanczos.maths.tcd.ie> References: <5.1.0.14.2.20020415140823.05130eb0@mail1.texoma.net> <000f01c1e493$2b67e880$265c27d4@evrocom.net> <000f01c1e493$2b67e880$265c27d4@evrocom.net> <5.1.0.14.2.20020415140823.05130eb0@mail1.texoma.net> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG David, Thanks *very much* for the real world input. I appreciate your time and hope you'll answer another question or two. We're running Linux clients and a FreeBSD server; the clients are mounting with mount -o "hard,intr,nfsvers=3,nodev,nolock,nosuid,rsize=32768,rw,tcp,wsize=32768" silo:/mail /mail How can I guarantee at the server level that 'noasync' is the case or do you recommend that? In other words, what do you recommend in /etc/exports for maximum reliability? Are there any parameters in the mount you would change? [No, we can't ignore performance ...] By the way, searching the area between 4 and 64 with a binary search indicates you can have at least 16 servers, but not as many as 24. Again, thanks for your input. rgds/ldv At 05:33 PM 4/15/2002, David Malone wrote: >On Mon, Apr 15, 2002 at 02:16:57PM -0500, Larry Vaden wrote: > > root 69 0.0 0.0 364 196 ?? Is 1:52PM 0:00.00 nfsd: master > > root 71 0.0 0.0 356 188 ?? S 1:52PM 0:00.22 nfsd: server > > root 72 0.0 0.0 356 188 ?? I 1:52PM 0:00.00 nfsd: server > > root 73 0.0 0.0 356 188 ?? I 1:52PM 0:00.00 nfsd: server > > root 74 0.0 0.0 356 188 ?? I 1:52PM 0:00.00 nfsd: server > >It looks like only the second nfsd has done any work (note the >0:00.22 time), so I don't think you need any more nfsds at the >moment. Keep an eye on their CPU usage and see how many of them are >actually getting used - that should give you a reasonable idea of >how many to run. > >I've included the equivelent output from a busy NFS server we have >(up for 80 days). We could possibly do with 1 more nfsd on this >machine, but we're not seeing any performance problems. > > David. > >root 199 0.3 0.0 356 56 ?? S 25Jan02 3142:36.14 nfsd: server >root 200 0.0 0.0 356 56 ?? S 25Jan02 373:45.14 nfsd: server >root 201 0.0 0.0 356 56 ?? S 25Jan02 38:55.37 nfsd: server >root 202 0.0 0.0 356 56 ?? I 25Jan02 7:48.41 nfsd: server To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message