From owner-freebsd-questions Thu Nov 16 18:01:05 1995 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id SAA00731 for questions-outgoing; Thu, 16 Nov 1995 18:01:05 -0800 Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.211]) by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id SAA00724 for ; Thu, 16 Nov 1995 18:01:02 -0800 Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id SAA04239; Thu, 16 Nov 1995 18:59:26 -0700 From: Terry Lambert Message-Id: <199511170159.SAA04239@phaeton.artisoft.com> Subject: Re: Poor NFS performance under FreeBSD To: mikhail@klm.com (Mikhail Kuperblum) Date: Thu, 16 Nov 1995 18:59:26 -0700 (MST) Cc: freebsd-questions@freebsd.org In-Reply-To: <9511161845.ZM1174@klm.com> from "Mikhail Kuperblum" at Nov 16, 95 06:45:34 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1331 Sender: owner-questions@freebsd.org Precedence: bulk > It there any way to tune up a NFS performance with FreeBSD? I have > FreeBSD v2.0.5 box as a server, on the client side is SCO Unix OSE 5 > connected via ethernet (SMC 8216). I've done all I could on the > client's side, but still it takes about 8 times as long to copy > a file over NFS than, for example, via ftp. Is there anything > tweakable in FreeBSD setup that can make a difference? BTW, I can't > find any trace of rpc.lockd or rpc.statd under FreeBSD. Aren't > they required? Thanks for any help you can provide and, since > I'm not a subscriber to this list, if you'll respond, please > Cc to mikhail@klm.com. It's your NFS write performance that sucks. This is compliant with the spec. If we cached, like SCO, we'd be in violation of the spec (like SCO). You can temporarily enable async when mounting (see the per fs mount man pages) but unless you want to be utterly screwed the first time the power fails, I wouldn't make this your first choice. The lockd/stad have to do with file locking andwon't affect write performance at all (unless you use an unlock as a trigger for client cache flushing, and even then your processes would have to know about it and cooperate). Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.