From owner-freebsd-fs@FreeBSD.ORG Fri Sep 3 23:41:07 2010 Return-Path: Delivered-To: freebsd-fs@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id EC61510656C1 for ; Fri, 3 Sep 2010 23:41:07 +0000 (UTC) (envelope-from rmacklem@uoguelph.ca) Received: from esa-jnhn.mail.uoguelph.ca (esa-jnhn.mail.uoguelph.ca [131.104.91.44]) by mx1.freebsd.org (Postfix) with ESMTP id AB79D8FC1A for ; Fri, 3 Sep 2010 23:41:07 +0000 (UTC) X-IronPort-Anti-Spam-Filtered: true X-IronPort-Anti-Spam-Result: ApwEAGokgUyDaFvO/2dsb2JhbACDGJ5wrWmRdIEigVaBUHMEihk X-IronPort-AV: E=Sophos;i="4.56,315,1280721600"; d="scan'208";a="92787044" Received: from erie.cs.uoguelph.ca (HELO zcs3.mail.uoguelph.ca) ([131.104.91.206]) by esa-jnhn-pri.mail.uoguelph.ca with ESMTP; 03 Sep 2010 19:41:05 -0400 Received: from zcs3.mail.uoguelph.ca (localhost.localdomain [127.0.0.1]) by zcs3.mail.uoguelph.ca (Postfix) with ESMTP id 0006EB3F20; Fri, 3 Sep 2010 19:41:05 -0400 (EDT) Date: Fri, 3 Sep 2010 19:41:05 -0400 (EDT) From: Rick Macklem To: David Brodbeck Message-ID: <697498209.486386.1283557265921.JavaMail.root@erie.cs.uoguelph.ca> In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit X-Originating-IP: [24.65.230.102] X-Mailer: Zimbra 6.0.7_GA_2476.RHEL4 (ZimbraWebClient - SAF3 (Mac)/6.0.7_GA_2473.RHEL4_64) Cc: freebsd-fs@freebsd.org Subject: Re: Weird Linux - FreeBSD/ZFS NFSv4 interoperability problem X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 03 Sep 2010 23:41:08 -0000 > (Note: I originally posted this to freebsd-questions, and it was > suggested I take it here.) > > While doing some interoperability testing between Linux and FreeBSD, I > came up with this unusual issue. I could use some help figuring out > if this is a bug, and if so, where to file it. Here's the scenario: > > - FreeBSD 8.1-RELEASE server, sharing a ZFS filesystem via NFSv4. > - Linux client (I've tested with RHEL 5.4 and Debian Lenny) mounting > said filesystem with NFSv4. > - A user on the Linux client does a Subversion checkout onto the > mounted filesystem. > > At the end of the checkout, access to the filesystem hangs. nfsd on > the FreeBSD server and rpciod on the Linux client seem to be in a > tight loop, and there's lots of network traffic between them. I can > reproduce this every time. > > The problem does not occur if the backing filesystem is UFS instead of > ZFS, if NFSv3 is used instead of NFSv4, or if the client is FreeBSD > instead of Linux. > If you email me a packet capture (probably only need a few seconds when it is the tight loop), I can look at it in wireshark. Something like "tcpdump -s 0 -w xxx host " run on the server should do it. Then email me "xxx" as an attachment. However, I don't know diddly w.r.t. ZFS, so I don't know how much I'll be able to help? rick ps: You either need the correct kernel modules or a kernel built with "options NFSD" to run the exp. server (which is the one that does NFSv4).