From owner-freebsd-net@FreeBSD.ORG Mon Apr 10 20:10:31 2006 Return-Path: X-Original-To: freebsd-net@freebsd.org Delivered-To: freebsd-net@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id F087916A400 for ; Mon, 10 Apr 2006 20:10:31 +0000 (UTC) (envelope-from cswiger@mac.com) Received: from pi.codefab.com (pi.codefab.com [199.103.21.227]) by mx1.FreeBSD.org (Postfix) with ESMTP id 8C3D243D45 for ; Mon, 10 Apr 2006 20:10:31 +0000 (GMT) (envelope-from cswiger@mac.com) Received: from localhost (localhost [127.0.0.1]) by pi.codefab.com (Postfix) with ESMTP id DE6DE5FEB; Mon, 10 Apr 2006 16:10:30 -0400 (EDT) Received: from pi.codefab.com ([127.0.0.1]) by localhost (pi.codefab.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 02896-07; Mon, 10 Apr 2006 16:10:29 -0400 (EDT) Received: from [199.103.21.238] (pan.codefab.com [199.103.21.238]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by pi.codefab.com (Postfix) with ESMTP id D64695C97; Mon, 10 Apr 2006 16:10:29 -0400 (EDT) In-Reply-To: References: Mime-Version: 1.0 (Apple Message framework v749.3) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Charles Swiger Date: Mon, 10 Apr 2006 16:10:28 -0400 To: dima <_pppp@mail.ru> X-Mailer: Apple Mail (2.749.3) X-Virus-Scanned: amavisd-new at codefab.com Cc: freebsd-net@freebsd.org Subject: Re: is NFS production-ready ? X-BeenThere: freebsd-net@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Networking and TCP/IP with FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 10 Apr 2006 20:10:32 -0000 On Apr 10, 2006, at 3:26 PM, dima wrote: > First, searching through the archives I'm about to say "No". > > My goal is to provide NFS service to many FreeBSD clients sharing > the exports. The usage pattern appears to be "many reads and not as > much writes". The deployment might look like the following: a SAN > and 2 NFS servers sharing its LUNs. The servers use hot-standby > scheme provided by CARP (or its equivalent). Many FreeBSD clients > would share their exports. I wish servers ran FreeBSD also since > it's the best known OS for the company administrators. The NFS protocol is stateless, but most clients doing writes will use a locking mechanism which is not stateless. In other words, you can easily cluster read-only NFS shares, but this is not true of read- write shares. > The majors are: > - no data corruption > - no hangs (this seems to be the largest problem with current > implementation) > - client retry on failure These two suggest you might be happier with Samba/CIFS. > - a reasonable read speed > > My questions: > 1. NFS/UDP (it's stateless!) is considered to be "evil". Why > (assuming I can grant a balanced network bandwidth)? Dunno, NFS over UDP works just fine. > 2. NFS server implementation seems to be very buggy. Any success > stories? Well, NFS servers can easily run Linux, Solaris etc. NFS works reasonably well on FreeBSD, modulo rpc.lockd. Solaris probably has the best NFS implementation available, and would be a better fileserver platform than almost anything else you've mentioned. NFS on Linux is probably more buggy than NFS on FreeBSD, from what I've seen. > 3. Is at least implementation of NFS client (either kernel-side or > user-space) stable enough for production use? Client OS replacement > is impossible (hardly suitable, really) in my project. NFS on FreeBSD is stable but perhaps not bullet-proof. > PS: The competing options are either SMB or CODA for now. Any other > suggestions? > PPS: I'd be happy to hear that FreeBSD supports at least one really > clustered FS (proprietary ones are also OK). But I think I wouldn't :( I think you can get some amount of the Veritas suite for FreeBSD... -- -Chuck